1 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
4 does not have a conditional at the top.
6 2018-01-02 Richard Biener <rguenther@suse.de>
8 * ipa-inline.c (big_speedup_p): Fix expression.
10 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
13 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
16 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
20 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
21 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
22 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
23 cond_taken_branch_cost 3->4.
25 2018-01-01 Jakub Jelinek <jakub@redhat.com>
27 PR tree-optimization/83581
28 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
29 TODO_cleanup_cfg if any changes have been made.
32 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
33 convert_modes if target mode has the right side, but different mode
37 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
38 last argument when extracting from CONCAT. If either from_real or
39 from_imag is NULL, use expansion through memory. If result is not
40 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
41 the parts directly to inner mode, if even that fails, use expansion
45 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
46 check for bswap in mode rather than HImode and use that in expand_unop
49 2017-12-31 Jakub Jelinek <jakub@redhat.com>
52 * config/i386/i386.c (ix86_attribute_table): Remove excess
53 initializer for "shared" attribute.
55 2017-12-30 Tom de Vries <tom@codesourcery.com>
58 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
60 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
61 offload_funcs entries that no longer have a corresponding cgraph_node.
62 Mark the remaining ones as DECL_PRESERVE_P.
63 (output_lto): Call prune_offload_funcs.
65 2017-12-30 Jakub Jelinek <jakub@redhat.com>
67 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
68 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
69 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
70 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
71 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
72 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
73 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
74 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
75 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
76 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
77 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
78 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
79 vaesenc_<mode>, vpclmulqdq_<mode>,
80 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
82 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
84 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
85 _Float<N> and _Float<N>X variants for rounding built-in
87 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
88 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
89 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
90 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
91 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
92 * builtins.c (mathfn_built_in_2): Likewise.
93 * internal-fn.def (CEIL): Likewise.
95 (NEARBYINT): Likewise.
99 * convert.c (convert_to_integer_1): Likewise.
100 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
101 (integer_valued_real_call_p): Likewise.
102 * fold-const-call.c (fold_const_call_ss): Likewise.
103 * gencfn-macros.c (print_case_cfn): Change CFN and operator
104 printers to take a const char * suffix instead of a bool.
105 (print_define_operator_list): Likewise.
106 (fltall_suffixes): New list of suffixes, that include the
107 traditional suffixes as well as all of the _Float<N> and
109 (main): For _Float<N> and _Float<N>X functions, emit both
110 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
111 has the _Float<N> and _Float<N>X case names or operators. The
112 <name>_ALL variant has both the traditional and the
113 _Float<N>/_Float<N>X case names or operators.
114 * match.pd (COPYSIGN optimizations): Provide optimizations for
115 _Float<N> and _Float<N>X types where possible.
116 (MIN/MAX optimizations): Likewise.
117 (sqrt optimizations): Likewise.
118 (rounding optimizations): Likewise.
120 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
122 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
123 gen_int_mode rather than GEN_INT.
124 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
126 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
128 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
129 instead of CONSTANT_P.
130 (gen_vec_series): Likewise.
131 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
133 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
135 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
138 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
140 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
141 (aarch64_ld1x2<VDC:mode>): Likewise.
142 (aarch64_simd_ld1<mode>_x2): Likewise.
143 (aarch64_simd_ld1<mode>_x2): Likewise.
144 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
145 (vld1_s8_x2): Likewise.
146 (vld1_u16_x2): Likewise.
147 (vld1_s16_x2): Likewise.
148 (vld1_u32_x2): Likewise.
149 (vld1_s32_x2): Likewise.
150 (vld1_u64_x2): Likewise.
151 (vld1_s64_x2): Likewise.
152 (vld1_f16_x2): Likewise.
153 (vld1_f32_x2): Likewise.
154 (vld1_f64_x2): Likewise.
155 (vld1_p8_x2): Likewise.
156 (vld1_p16_x2): Likewise.
157 (vld1_p64_x2): Likewise.
158 (vld1q_u8_x2): Likewise.
159 (vld1q_s8_x2): Likewise.
160 (vld1q_u16_x2): Likewise.
161 (vld1q_s16_x2): Likewise.
162 (vld1q_u32_x2): Likewise.
163 (vld1q_s32_x2): Likewise.
164 (vld1q_u64_x2): Likewise.
165 (vld1q_s64_x2): Likewise.
166 (vld1q_f16_x2): Likewise.
167 (vld1q_f32_x2): Likewise.
168 (vld1q_f64_x2): Likewise.
169 (vld1q_p8_x2): Likewise.
170 (vld1q_p16_x2): Likewise.
171 (vld1q_p64_x2): Likewise.
173 2017-12-27 Martin Liska <mliska@suse.cz>
175 PR tree-optimization/83552
176 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
177 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
179 2017-12-27 Tom de Vries <tom@codesourcery.com>
181 * config/nvptx/nvptx.c (nvptx_option_override): Disable
182 -gstatement-frontiers.
184 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
186 PR rtl-optimization/83513
187 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
188 before priority comparison.
190 2017-12-25 Jakub Jelinek <jakub@redhat.com>
193 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
194 ix86_isa_flags2 to ix86_isa_flags.
195 * config/i386/i386-c.c (ix86_target_macros_internal): Test
196 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
197 isa_flags rather than isa_flags2.
198 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
199 and -mavx512bitalg from isa2_opts to isa_opts.
200 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
201 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
202 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
204 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
205 OPTION_MASK_ISA_AVX512F ored with another option similarly to
206 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
207 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
208 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
209 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
210 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
211 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
212 builtins from bdesc_args2 to bdesc_args section.
213 (__builtin_ia32_compressstoreuqi512_mask,
214 __builtin_ia32_compressstoreuhi512_mask,
215 __builtin_ia32_compressstoreuqi256_mask,
216 __builtin_ia32_expandloadqi512_mask,
217 __builtin_ia32_expandloadqi512_maskz,
218 __builtin_ia32_expandloadhi512_mask,
219 __builtin_ia32_expandloadhi512_maskz,
220 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
221 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
222 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
223 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
224 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
225 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
226 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
227 __builtin_ia32_vpshldv_v32hi_maskz,
228 __builtin_ia32_vpopcountb_v64qi_mask,
229 __builtin_ia32_vpopcountw_v32hi_mask,
230 __builtin_ia32_vpshufbitqmb512_mask,
231 __builtin_ia32_vpshufbitqmb256_mask): Add
232 " | OPTION_MASK_ISA_AVX512BW".
233 (__builtin_ia32_expandloadqi256_mask,
234 __builtin_ia32_expandloadqi256_maskz,
235 __builtin_ia32_vpopcountb_v32qi_mask): Add
236 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
237 (__builtin_ia32_expandloadhi256_mask,
238 __builtin_ia32_expandloadhi256_maskz,
239 __builtin_ia32_expandloadqi128_mask,
240 __builtin_ia32_expandloadqi128_maskz,
241 __builtin_ia32_expandloadhi128_mask,
242 __builtin_ia32_expandloadhi128_maskz,
243 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
244 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
245 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
246 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
247 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
248 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
249 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
250 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
251 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
252 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
253 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
254 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
255 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
256 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
257 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
258 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
259 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
260 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
261 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
262 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
263 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
264 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
265 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
266 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
267 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
268 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
269 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
270 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
271 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
272 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
273 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
274 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
275 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
276 __builtin_ia32_vpopcountw_v8hi_mask): Add
277 " | OPTION_MASK_ISA_AVX512VL".
278 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
279 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
280 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
281 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
282 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
283 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
284 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
285 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
286 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
287 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
288 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
289 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
290 _mm_mask_bitshuffle_epi64_mask): Likewise.
291 * common/config/i386/i386-common.c
292 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
293 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
294 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
295 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
296 OPTION_MASK_ISA_AVX512BITALG_UNSET.
297 (OPTION_MASK_ISA2_AVX512F_UNSET,
298 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
299 (ix86_handle_option): For -mno-general-regs-only, clear from
300 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
301 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
302 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
303 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
304 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
305 for moving from ix86_isa_flags2 to ix86_isa_flags.
307 2017-12-23 Jakub Jelinek <jakub@redhat.com>
310 * fold-const.c (struct contains_label_data): New type.
311 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
312 inside of a SWITCH_BODY seen during the walk.
313 (contains_label_p): Use walk_tree instead of
314 walk_tree_without_duplicates, prepare data for contains_label_1 and
317 2017-12-22 Jakub Jelinek <jakub@redhat.com>
320 * tree-iterator.c (alloc_stmt_list): Start with cleared
321 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
325 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
326 enable avx512f explicitly in #pragma GCC target.
327 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
328 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
329 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
330 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
331 __builtin_ia32_vpdpbusds_v8si_mask,
332 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
333 __builtin_ia32_vpdpbusds_v4si_mask,
334 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
335 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
336 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
337 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
338 __builtin_ia32_vpdpwssds_v8si_mask,
339 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
340 __builtin_ia32_vpdpwssds_v4si_mask,
341 __builtin_ia32_vpdpwssds_v4si_maskz): Use
342 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
343 just OPTION_MASK_ISA_AVX512VNNI.
345 2017-12-22 Martin Jambor <mjambor@suse.cz>
348 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
351 2017-12-22 Julia Koval <julia.koval@intel.com>
352 Sebastian Peryt <sebastian.peryt@intel.com>
354 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
355 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
356 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
357 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
358 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
359 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
360 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
361 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
362 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
363 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
364 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
365 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
366 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
367 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
368 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
369 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
370 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
371 _mm256_maskz_popcnt_epi64): New intrinsics.
372 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
373 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
374 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
375 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
376 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
377 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
378 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
379 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
380 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
381 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
382 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
383 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
384 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
385 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
386 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
387 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
388 __builtin_ia32_vpshufbitqmb256_mask,
389 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
390 * config/i386/i386-c.c (__AVX512BITALG__): New.
391 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
392 (ix86_valid_target_attribute_inner_p): Ditto.
393 (ix86_expand_args_builtin): Handle new types.
394 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
395 * config/i386/i386.opt: Add -mavx512bitalg.
396 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
397 avx512bitalgintrin.h.
398 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
399 (vpopcount<mode><mask_name>): Add more types.
400 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
401 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
403 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
405 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
406 Or in OPTION_MASK_ISA_AVX512F_SET.
407 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
408 OPTION_MASK_ISA_AVX512VNNI_UNSET.
409 (ix86_handle_option): Adjust for
410 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
411 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
413 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
414 isa_flag instead of isa_flag2.
415 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
416 isa_opts2 to isa_opts.
417 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
420 2017-12-22 Mike Stump <mikestump@comcast.net>
421 Eric Botcazou <ebotcazou@adacore.com>
423 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
425 2017-12-21 Jakub Jelinek <jakub@redhat.com>
428 * config/i386/i386.c (ix86_function_arg_boundary): Return
429 PARM_BOUNDARY for TYPE_EMPTY_P types.
432 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
433 if navail is >= dir.len.
435 2017-12-21 Steve Ellcey <sellcey@cavium.com>
437 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
440 2017-12-21 Jakub Jelinek <jakub@redhat.com>
442 PR rtl-optimization/80747
443 PR rtl-optimization/83512
444 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
445 succ edge from ENTRY, copy partition from e->dest to the newly
447 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
448 ENTRY, use BB_PARTITION of its successor block as current_partition.
449 Don't copy partition when splitting succ edge from ENTRY.
451 PR tree-optimization/83523
452 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
453 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
454 (convert_mult_to_fma): Likewise.
456 PR tree-optimization/83521
457 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
458 gimple_build_assign without code on result of
459 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
462 2017-12-21 Andrew Pinski <apinski@cavium.com>
463 Steve Ellcey <sellcey@cavium.com>
465 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
466 multi-arch for ilp32.
468 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
471 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
472 constraints to operand 2.
473 (*ashl<mode>3_mask_1): Ditto.
474 (*<shift_insn><mode>3_mask): Ditto.
475 (*<shift_insn><mode>3_mask_1): Ditto.
476 (*<rotate_insn><mode>3_mask): Ditto.
477 (*<rotate_insn><mode>3_mask_1): Ditto.
479 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
481 * reorg.c (make_return_insns): Reemit each insn with its own location.
483 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
485 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
486 generation for cases where splatting a value is not useful.
487 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
488 across a vec_duplicate and a paradoxical subreg forming a vector
489 mode to a vec_concat.
491 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
493 * combine.c (simplify_set): Do not transform subregs to zero_extends
494 if the destination is not a scalar int mode.
496 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
499 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
500 shared zero if the input has overflowed.
502 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
504 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
505 support for Cortex-A55 and Cortex-A75.
507 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
509 * common/config/arm/arm-common.c (compare_opt_names): Add function
510 comment. Use strcmp instead of manual loop.
512 2017-12-21 Martin Liska <mliska@suse.cz>
514 PR gcov-profile/83509
515 * gcov-dump.c (dump_gcov_file): Do not read info about
516 support_unexecuted_blocks for gcda files.
518 2017-12-21 Jakub Jelinek <jakub@redhat.com>
520 PR rtl-optimization/82973
521 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
522 (valid_for_const_vector_p): ... this.
523 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
524 (valid_for_const_vector_p): ... this. Adjust function comment.
525 (gen_vec_duplicate): Adjust caller.
526 * optabs.c (expand_vector_broadcast): Likewise.
527 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
528 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
530 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
531 instead of GET_CODE == CONST_FIXED.
532 (simplify_subreg): Use CONST_FIXED_P macro instead of
533 GET_CODE == CONST_FIXED.
536 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
537 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
538 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
539 (ix86_option_override_internal): Adjust for
540 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
541 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
542 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
543 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
544 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
545 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
546 def_builtin2 for CET builtins.
547 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
548 up formatting in IX86_BUILTIN_RDPID code.
549 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
550 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
551 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
552 ix86_isa_flags2 to ix86_isa_flags.
553 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
555 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
556 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
557 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
559 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
560 Or in OPTION_MASK_ISA_AVX512F_SET.
561 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
562 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
563 (ix86_handle_option): Adjust for
564 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
565 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
567 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
568 Alan Hayward <alan.hayward@arm.com>
569 David Sherwood <david.sherwood@arm.com>
571 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
572 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
573 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
574 Track polynomial offsets.
576 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
577 Alan Hayward <alan.hayward@arm.com>
578 David Sherwood <david.sherwood@arm.com>
580 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
581 Treat drb->init as a poly_int. Fail if its misalignment wrt
582 vector_alignment isn't known.
584 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
585 Alan Hayward <alan.hayward@arm.com>
586 David Sherwood <david.sherwood@arm.com>
588 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
590 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
591 (vect_loop_versioning): Take the loop versioning threshold as a
593 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
594 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
595 versioning_threshold.
596 (vect_analyze_loop_2): Compute the loop versioning threshold
597 whenever loop versioning is needed, and store it in the new
598 field rather than combining it with the cost model threshold.
599 (vect_transform_loop): Update call to vect_loop_versioning.
600 Try to combine the loop versioning and cost thresholds here.
602 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
603 Alan Hayward <alan.hayward@arm.com>
604 David Sherwood <david.sherwood@arm.com>
606 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
607 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
608 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
609 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
610 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
611 (group_compare_offset): Update accordingly.
612 (split_small_address_groups_p): Likewise.
613 (record_use): Take addr_offset as a poly_uint64 rather than
614 an unsigned HOST_WIDE_INT.
615 (strip_offset): Return the offset as a poly_uint64 rather than
616 an unsigned HOST_WIDE_INT.
617 (record_group_use, split_address_groups): Track polynomial offsets.
618 (add_iv_candidate_for_use): Likewise.
619 (addr_offset_valid_p): Take the offset as a poly_int64 rather
620 than a HOST_WIDE_INT.
621 (strip_offset_1): Return the offset as a poly_int64 rather than
624 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
625 Alan Hayward <alan.hayward@arm.com>
626 David Sherwood <david.sherwood@arm.com>
628 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
629 rather than a HOST_WIDE_INT.
630 * tree.c (get_binfo_at_offset): Likewise.
632 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
633 Alan Hayward <alan.hayward@arm.com>
634 David Sherwood <david.sherwood@arm.com>
636 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
637 rather than a HOST_WIDE_INT.
638 * tree-sra.c (build_ref_for_offset): Likewise.
640 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
641 Alan Hayward <alan.hayward@arm.com>
642 David Sherwood <david.sherwood@arm.com>
644 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
646 * tree.c (mem_ref_offset): Likewise.
647 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
648 * builtins.c (get_object_alignment_2): Likewise.
649 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
650 * gimple-fold.c (get_base_constructor): Likewise.
651 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
652 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
654 * ipa-polymorphic-call.c
655 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
656 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
657 (get_ancestor_addr_info): Likewise.
658 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
659 * match.pd: Likewise.
660 * tree-data-ref.c (dr_analyze_innermost): Likewise.
661 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
662 * tree-eh.c (tree_could_trap_p): Likewise.
663 * tree-object-size.c (addr_object_size): Likewise.
664 * tree-ssa-address.c (copy_ref_info): Likewise.
665 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
666 (indirect_refs_may_alias_p): Likewise.
667 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
668 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
669 (non_rewritable_mem_ref_base): Likewise.
670 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
671 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
672 * varasm.c (decode_addr_const): Likewise.
674 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
675 Alan Hayward <alan.hayward@arm.com>
676 David Sherwood <david.sherwood@arm.com>
678 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
679 offsets for MEM_REFs.
681 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
682 Alan Hayward <alan.hayward@arm.com>
683 David Sherwood <david.sherwood@arm.com>
685 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
686 as a poly_uint64 rather than an unsigned int.
687 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
689 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
690 Alan Hayward <alan.hayward@arm.com>
691 David Sherwood <david.sherwood@arm.com>
693 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
695 (decode_addr_const): Update accordingly.
697 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
698 Alan Hayward <alan.hayward@arm.com>
699 David Sherwood <david.sherwood@arm.com>
701 * tree.h (bit_field_size, bit_field_offset): New functions.
702 * hsa-gen.c (gen_hsa_addr): Use them.
703 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
704 (simplify_vector_constructor): Likewise.
705 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
706 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
707 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
708 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
711 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
712 Alan Hayward <alan.hayward@arm.com>
713 David Sherwood <david.sherwood@arm.com>
715 * expr.h (emit_group_load, emit_group_load_into_temps)
716 (emit_group_store): Take the size as a poly_int64 rather than an int.
717 * expr.c (emit_group_load_1, emit_group_load): Likewise.
718 (emit_group_load_into_temp, emit_group_store): Likewise.
720 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
721 Alan Hayward <alan.hayward@arm.com>
722 David Sherwood <david.sherwood@arm.com>
724 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
725 an unsigned int to a poly_uint64.
726 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
727 sizes as poly_uint64s rather than unsigned ints.
728 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
731 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
732 Alan Hayward <alan.hayward@arm.com>
733 David Sherwood <david.sherwood@arm.com>
735 * lra-constraints.c (emit_inc): Change inc_amount from an int
738 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
739 Alan Hayward <alan.hayward@arm.com>
740 David Sherwood <david.sherwood@arm.com>
742 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
744 (add_stack_var, stack_var_cmp, partition_stack_vars)
745 (dump_stack_var_partition): Update accordingly.
746 (alloc_stack_frame_space): Take the size as a poly_int64 rather
747 than a HOST_WIDE_INT.
748 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
749 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
750 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
751 rather than a HOST_WIDE_INT.
753 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
754 Alan Hayward <alan.hayward@arm.com>
755 David Sherwood <david.sherwood@arm.com>
757 * target.def (return_pops_args): Treat both the input and output
758 sizes as poly_int64s rather than HOST_WIDE_INTS.
759 * targhooks.h (default_return_pops_args): Update accordingly.
760 * targhooks.c (default_return_pops_args): Likewise.
761 * doc/tm.texi: Regenerate.
762 * emit-rtl.h (incoming_args): Change pops_args, size and
763 outgoing_args_size from int to poly_int64_pod.
764 * function.h (expr_status): Change x_pending_stack_adjust and
765 x_stack_pointer_delta from int to poly_int64.
766 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
767 (ARGS_SIZE_RTX): Update accordingly.
768 * calls.c (highest_outgoing_arg_in_use): Change from int to
770 (stack_usage_watermark, stored_args_watermark): New variables.
771 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
772 (emit_call_1): Change the stack_size and rounded_stack_size
773 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
775 (save_fixed_argument_area): Check stack_usage_watermark.
776 (initialize_argument_information): Change old_pending_adj from
777 a HOST_WIDE_INT * to a poly_int64_pod *.
778 (compute_argument_block_size): Return the size as a poly_int64
780 (finalize_must_preallocate): Track polynomial argument sizes.
781 (compute_argument_addresses): Likewise.
782 (internal_arg_pointer_based_exp): Track polynomial offsets.
783 (mem_overlaps_already_clobbered_arg_p): Rename to...
784 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
785 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
786 Check stored_args_used_watermark.
787 (load_register_parameters): Update accordingly.
788 (check_sibcall_argument_overlap_1): Likewise.
789 (combine_pending_stack_adjustment_and_call): Take the unadjusted
790 args size as a poly_int64 rather than an int. Return a bool
791 indicating whether the optimization was possible and return
792 the new adjustment by reference.
793 (check_sibcall_argument_overlap): Track polynomail argument sizes.
794 Update stored_args_watermark.
795 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
796 (expand_call): Likewise. Maintain stack_usage_watermark and
797 stored_args_watermark. Update calls to
798 combine_pending_stack_adjustment_and_call.
799 (emit_library_call_value_1): Handle polynomial argument sizes.
800 Call stack_region_maybe_used_p and mark_stack_region_used.
801 Maintain stack_usage_watermark.
802 (store_one_arg): Likewise. Update call to
803 mem_overlaps_already_clobbered_arg_p.
804 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
806 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
807 * config/microblaze/microblaze.c (microblaze_function_prologue):
809 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
810 TARGET_RETURN_POPS_ARGS interface.
811 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
813 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
814 * config/i386/i386.c (ix86_return_pops_args): Update for new
815 TARGET_RETURN_POPS_ARGS interface.
816 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
817 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
818 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
819 TARGET_RETURN_POPS_ARGS interface.
820 * config/vax/vax.c (vax_return_pops_args): Likewise.
821 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
822 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
823 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
824 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
825 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
826 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
827 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
828 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
829 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
830 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
831 and x_stack_pointer_delta from int to poly_int64.
832 * dojump.c (do_pending_stack_adjust): Update accordingly.
833 * explow.c (allocate_dynamic_stack_space): Handle polynomial
834 stack_pointer_deltas.
835 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
836 (pad_to_arg_alignment): Track polynomial offsets.
837 (assign_parm_find_stack_rtl): Likewise.
838 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
839 * toplev.c (output_stack_usage): Update reference to
840 current_function_pushed_stack_size.
842 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
843 Alan Hayward <alan.hayward@arm.com>
844 David Sherwood <david.sherwood@arm.com>
846 * function.c (in_arg_offset, var_offset, dynamic_offset)
847 (out_arg_offset, cfa_offset): Change from int to poly_int64.
848 (instantiate_new_reg): Return the new offset as a poly_int64_pod
849 rather than a HOST_WIDE_INT.
850 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
851 (instantiate_virtual_regs_in_insn): Likewise.
853 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
854 Alan Hayward <alan.hayward@arm.com>
855 David Sherwood <david.sherwood@arm.com>
857 * rtl.h (get_args_size, add_args_size_note): New functions.
858 (find_args_size_adjust): Return a poly_int64 rather than a
860 (fixup_args_size_notes): Likewise. Make the same change to the
861 end_args_size parameter.
862 * rtlanal.c (get_args_size, add_args_size_note): New functions.
863 * builtins.c (expand_builtin_trap): Use add_args_size_note.
864 * calls.c (emit_call_1): Likewise.
865 * explow.c (adjust_stack_1): Likewise.
866 * cfgcleanup.c (old_insns_match_p): Update use of
867 find_args_size_adjust.
868 * combine.c (distribute_notes): Track polynomial arg sizes.
869 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
870 end_true_args_size, beg_delay_args_size and end_delay_args_size
871 from HOST_WIDE_INT to poly_int64.
872 (add_cfi_args_size): Take the args_size as a poly_int64 rather
873 than a HOST_WIDE_INT.
874 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
875 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
876 polynomial arg sizes.
877 * emit-rtl.c (try_split): Use get_args_size.
878 * recog.c (peep2_attempt): Likewise.
879 * reload1.c (reload_as_needed): Likewise.
880 * expr.c (find_args_size_adjust): Return the adjustment as a
881 poly_int64 rather than a HOST_WIDE_INT.
882 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
883 to a poly_int64 and change the return type in the same way.
884 (emit_single_push_insn): Track polynomial arg sizes.
886 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
887 Alan Hayward <alan.hayward@arm.com>
888 David Sherwood <david.sherwood@arm.com>
890 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
891 from HOST_WIDE_INT to poly_int64.
892 * expr.c (push_block, emit_push_insn): Likewise.
894 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
895 Alan Hayward <alan.hayward@arm.com>
896 David Sherwood <david.sherwood@arm.com>
898 * function.h (frame_space): Change start and length from HOST_WIDE_INT
900 (get_frame_size): Return the size as a poly_int64 rather than a
902 (frame_offset_overflow): Take the offset as a poly_int64 rather
903 than a HOST_WIDE_INT.
904 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
905 (assign_stack_temp): Likewise for the size.
906 * function.c (get_frame_size): Return a poly_int64 rather than
908 (frame_offset_overflow): Take the offset as a poly_int64 rather
909 than a HOST_WIDE_INT.
910 (try_fit_stack_local): Take the start, length and size as poly_int64s
911 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
912 rather than a HOST_WIDE_INT.
913 (add_frame_space): Take the start and end as poly_int64s rather than
915 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
916 (assign_stack_temp): Likewise for the size.
917 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
919 (find_temp_slot_from_address): Handle polynomial offsets.
920 (combine_temp_slots): Likewise.
921 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
923 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
924 poly_int64 rather than a HOST_WIDE_INT.
925 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
926 than a HOST_WIDE_INT.
927 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
928 polynomial frame offsets.
929 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
930 as a poly_int64 rather than an int.
931 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
932 * config/v850/v850-protos.h (compute_frame_size): Likewise.
933 * config/v850/v850.c (compute_frame_size): Likewise.
934 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
935 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
936 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
937 * config/pa/pa.c (pa_compute_frame_size): Likewise.
938 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
939 rather than a HOST_WIDE_INT.
940 * explow.c (get_dynamic_stack_base): Likewise.
941 * final.c (final_start_function): Use the constant lower bound
942 of the frame size for -Wframe-larger-than.
943 * ira.c (do_reload): Adjust for new get_frame_size return type.
944 * lra.c (lra): Likewise.
945 * reload1.c (reload): Likewise.
946 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
947 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
948 * rtlanal.c (get_initial_register_offset): Return the offset as
949 a poly_int64 rather than a HOST_WIDE_INT.
951 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
952 Alan Hayward <alan.hayward@arm.com>
953 David Sherwood <david.sherwood@arm.com>
955 * reload1.c (elim_table): Change initial_offset, offset and
956 previous_offset from HOST_WIDE_INT to poly_int64_pod.
957 (offsets_at): Change the target array's element type from
958 HOST_WIDE_INT to poly_int64_pod.
959 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
960 (elimination_costs_in_insn, update_eliminable_offsets)
961 (verify_initial_elim_offsets, set_offsets_for_label)
962 (init_eliminable_invariants): Update after above changes.
964 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
965 Alan Hayward <alan.hayward@arm.com>
966 David Sherwood <david.sherwood@arm.com>
968 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
969 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
970 (decomposition): Change start and end from HOST_WIDE_INT
972 (decompose, immune_p): Update accordingly.
973 (find_inc_amount): Return a poly_int64 rather than an int.
974 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
977 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
978 Alan Hayward <alan.hayward@arm.com>
979 David Sherwood <david.sherwood@arm.com>
981 * tree.h (get_inner_reference): Return the bitsize and bitpos
982 as poly_int64_pods rather than HOST_WIDE_INT.
983 * fold-const.h (ptr_difference_const): Return the pointer difference
984 as a poly_int64_pod rather than a HOST_WIDE_INT.
985 * expr.c (get_inner_reference): Return the bitsize and bitpos
986 as poly_int64_pods rather than HOST_WIDE_INT.
987 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
989 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
990 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
991 (optimize_bit_field_compare): Update call to get_inner_reference.
992 (decode_field_reference): Likewise.
993 (fold_unary_loc): Track polynomial offsets and sizes.
994 (split_address_to_core_and_offset): Return the bitpos as a
995 poly_int64_pod rather than a HOST_WIDE_INT.
996 (ptr_difference_const): Likewise for the pointer difference.
997 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
998 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
999 * dbxout.c (dbxout_expand_expr): Likewise.
1000 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1001 (loc_list_from_tree_1, fortran_common): Likewise.
1002 * gimple-laddress.c (pass_laddress::execute): Likewise.
1003 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1004 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1005 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1006 * tree-affine.c (tree_to_aff_combination): Likewise.
1007 (get_inner_reference_aff): Likewise.
1008 * tree-data-ref.c (split_constant_offset_1): Likewise.
1009 (dr_analyze_innermost): Likewise.
1010 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1011 * tree-sra.c (ipa_sra_check_caller): Likewise.
1012 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1013 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1014 (instrument_bool_enum_load, instrument_object_size): Likewise.
1015 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1016 to get_inner_reference.
1017 * hsa-gen.c (gen_hsa_addr): Likewise.
1018 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1019 * tsan.c (instrument_expr): Likewise.
1020 * match.pd: Update call to ptr_difference_const.
1022 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1023 Alan Hayward <alan.hayward@arm.com>
1024 David Sherwood <david.sherwood@arm.com>
1026 * fold-const.c (fold_comparison): Track sizes and offsets as
1027 poly_int64s rather than HOST_WIDE_INTs when folding address
1030 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1031 Alan Hayward <alan.hayward@arm.com>
1032 David Sherwood <david.sherwood@arm.com>
1034 * expr.h (get_bit_range): Return the bitstart and bitend as
1035 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1036 as a poly_int64 rather than a HOST_WIDE_INT.
1037 * expr.c (get_bit_range): Likewise.
1038 (expand_assignment): Update call accordingly.
1039 * fold-const.c (optimize_bit_field_compare): Likewise.
1041 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1042 Alan Hayward <alan.hayward@arm.com>
1043 David Sherwood <david.sherwood@arm.com>
1045 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1046 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1047 as an optional final parameter.
1048 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1049 (aarch64_legitimate_address_p): Likewise.
1050 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1051 instead of an rtx code.
1052 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1053 (aarch64_legitimate_address_hook_p): Likewise.
1054 (aarch64_print_ldpstp_address): Likewise.
1055 (aarch64_print_operand_address): Likewise.
1056 (aarch64_address_cost): Likewise.
1057 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1058 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1059 (aarch64_mem_pair_lanes_operand): Likewise.
1061 2017-12-20 Richard Biener <rguenther@suse.de>
1063 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1064 update_stmt_if_modified.
1066 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1068 PR tree-optimization/83491
1069 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1070 before walking uses. Improve coding style and comments.
1072 2017-12-20 Tom de Vries <tom@codesourcery.com>
1074 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1076 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1079 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1081 (pass_ipa_free_fn_summary): Add small_p private data member,
1082 initialize to false in the ctor.
1083 (pass_ipa_free_fn_summary::clone,
1084 pass_ipa_free_fn_summary::set_pass_param,
1085 pass_ipa_free_fn_summary::gate): New methods.
1086 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1087 | TODO_dump_symtab if small_p.
1088 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1089 entry and add another instance of the pass with false parm after
1091 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1092 ipa_free_fn_summary here.
1094 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1096 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1098 2017-12-20 Martin Sebor <msebor@redhat.com>
1101 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1102 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1103 (expand_builtin_strncmp): Same.
1105 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1108 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1109 allowed debug stmts before labels.
1110 (expand_gimple_basic_block): Likewise.
1111 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1112 * gimple-iterator.h (gsi_after_labels): Likewise.
1113 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1114 rename reused variable, and simplify using gsi_move_before.
1115 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1116 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1117 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1118 (gimple_verify_flow_info, gimple_block_label): Likewise.
1119 (make_blocks): Move debug markers after adjacent labels.
1120 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1121 allowed debug insns outside blocks.
1122 * df-scan.c (df_insn_delete): Likewise.
1123 * lra-constraints.c (update_ebb_live_info): Likewise.
1124 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1125 (vt_initialize, delete_vta_debug_insns): Likewise.
1126 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1128 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1129 Alan Hayward <alan.hayward@arm.com>
1130 David Sherwood <david.sherwood@arm.com>
1132 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1133 (round_up_to_byte_boundary): Likewise.
1134 * expr.h (get_bit_range): Add temporary shim.
1135 * gimple-ssa-store-merging.c (store_operand_info): Change the
1136 bitsize, bitpos, bitregion_start and bitregion_end fields from
1137 unsigned HOST_WIDE_INT to poly_uint64.
1138 (merged_store_group): Likewise load_align_base.
1139 (compatible_load_p, compatible_load_p): Update accordingly.
1140 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1141 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1142 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1143 bitregion_start and bitregion_end as poly_uint64s rather than
1144 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1145 (handled_load): Take the bitsize, bitpos,
1146 bitregion_start and bitregion_end as poly_uint64s rather than
1147 unsigned HOST_WIDE_INTs.
1148 (pass_store_merging::process_store): Update call to
1149 mem_valid_for_store_merging.
1151 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1152 Alan Hayward <alan.hayward@arm.com>
1153 David Sherwood <david.sherwood@arm.com>
1155 * builtins.c (get_object_alignment_2): Track polynomial offsets
1156 and sizes. Update the alignment handling.
1158 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1159 Alan Hayward <alan.hayward@arm.com>
1160 David Sherwood <david.sherwood@arm.com>
1162 * tree.h (get_inner_reference): Add a version that returns the
1163 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1164 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1165 the case in which bitpos is not associated with the first byte.
1167 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1168 Alan Hayward <alan.hayward@arm.com>
1169 David Sherwood <david.sherwood@arm.com>
1171 * tree-affine.h (get_inner_reference_aff): Return the size as a
1173 * tree-affine.c (get_inner_reference_aff): Likewise.
1174 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1175 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1177 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1178 Alan Hayward <alan.hayward@arm.com>
1179 David Sherwood <david.sherwood@arm.com>
1181 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1182 HOST_WIDE_INT rather than a poly_int64.
1184 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1185 Alan Hayward <alan.hayward@arm.com>
1186 David Sherwood <david.sherwood@arm.com>
1188 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1189 HOST_WIDE_INT to poly_int64_pod.
1190 (perform_symbolic_merge): Update accordingly.
1191 (bswap_replace): Likewise.
1193 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1194 Alan Hayward <alan.hayward@arm.com>
1195 David Sherwood <david.sherwood@arm.com>
1197 * tree-affine.h (aff_tree::offset): Change from widest_int
1199 (wide_int_ext_for_comb): Delete.
1200 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1201 constants as poly_widest_int rather than widest_int.
1202 (aff_combination_constant_multiple_p): Return the multiplier
1203 as a poly_widest_int.
1204 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1206 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1207 version static and add an overload for poly_widest_int.
1208 (aff_combination_const, aff_combination_add_cst)
1209 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1210 the constants as poly_widest_int rather than widest_int.
1211 (tree_to_aff_combination): Generalize INTEGER_CST case to
1213 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1214 (aff_combination_add_product, aff_combination_mult): Handle
1216 (aff_combination_constant_multiple_p): Return the multiplier
1217 as a poly_widest_int.
1218 * tree-predcom.c (determine_offset): Return the offset as a
1220 (split_data_refs_to_components, suitable_component_p): Update
1222 (valid_initializer_p): Update call to
1223 aff_combination_constant_multiple_p.
1224 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1225 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1226 as a poly_int64 rather than a HOST_WIDE_INT.
1227 (get_address_cost): Handle polynomial offsets.
1228 (iv_elimination_compare_lt): Likewise.
1229 (rewrite_use_nonlinear_expr): Likewise.
1231 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1232 Alan Hayward <alan.hayward@arm.com>
1233 David Sherwood <david.sherwood@arm.com>
1235 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1236 as a poly_int64_pod rather than a HOST_WIDE_INT.
1237 (get_addr_base_and_unit_offset): Likewise.
1238 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1239 (get_addr_base_and_unit_offset): Likewise.
1240 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1241 to poly_int64 in example.
1242 * fold-const.c (fold_binary_loc): Update call to
1243 get_addr_base_and_unit_offset.
1244 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1245 (maybe_canonicalize_mem_ref_addr): Likewise.
1246 (gimple_fold_stmt_to_constant_1): Likewise.
1247 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1249 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1250 * match.pd: Likewise.
1251 * omp-low.c (lower_omp_target): Likewise.
1252 * tree-sra.c (build_ref_for_offset): Likewise.
1253 (build_debug_ref_for_model): Likewise.
1254 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1255 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1256 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1257 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1258 (constant_pointer_difference): Likewise.
1259 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1260 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1261 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1262 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1263 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1264 (set_ssa_val_to): Likewise.
1265 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1266 (maybe_diag_stxncpy_trunc): Likewise.
1267 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1268 * tree.c (build_simple_mem_ref_loc): Likewise.
1269 (array_at_struct_end_p): Likewise.
1271 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1272 Alan Hayward <alan.hayward@arm.com>
1273 David Sherwood <david.sherwood@arm.com>
1275 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1276 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1277 (get_ref_base_and_extent_hwi): Declare.
1278 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1279 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1280 (get_ref_base_and_extent_hwi): New function.
1281 * cfgexpand.c (expand_debug_expr): Update call to
1282 get_ref_base_and_extent.
1283 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1284 * gimple-fold.c (get_base_constructor): Return the offset as a
1285 poly_int64_pod rather than a HOST_WIDE_INT.
1286 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1287 * ipa-polymorphic-call.c
1288 (ipa_polymorphic_call_context::set_by_invariant)
1289 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1290 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1291 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1292 rather than get_ref_base_and_extent.
1293 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1294 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1295 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1297 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1298 call to get_ref_base_and_extent.
1299 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1300 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1301 (stmt_kills_ref_p): Likewise.
1302 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1303 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1305 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1306 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1307 when calling native_encode_expr.
1308 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1309 call to get_ref_base_and_extent.
1310 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1311 get_ref_base_and_extent.
1312 * var-tracking.c (track_expr_p): Likewise.
1314 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1315 Alan Hayward <alan.hayward@arm.com>
1316 David Sherwood <david.sherwood@arm.com>
1318 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1319 HOST_WIDE_INT to poly_int64_pod.
1320 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1321 polynomail parameter offsets.
1323 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1324 Alan Hayward <alan.hayward@arm.com>
1325 David Sherwood <david.sherwood@arm.com>
1327 * gengtype.c (main): Handle poly_int64_pod.
1328 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1329 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1330 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1331 from HOST_WIDE_INT to poly_int64_pod.
1332 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1333 (copy_cfa): New function.
1334 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1335 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1336 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1337 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1338 (initial_return_save): Treat offsets as poly_ints.
1339 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1340 offset is nonconstant.
1341 (reg_save): Take the offset as a poly_int64. Fall back to
1342 DW_CFA_expression for nonconstant offsets.
1343 (queue_reg_save): Take the offset as a poly_int64.
1344 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1346 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1347 Alan Hayward <alan.hayward@arm.com>
1348 David Sherwood <david.sherwood@arm.com>
1350 * rtl.h (operand_subword, operand_subword_force): Take the offset
1351 as a poly_uint64 an unsigned int.
1352 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1354 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1355 Alan Hayward <alan.hayward@arm.com>
1356 David Sherwood <david.sherwood@arm.com>
1358 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1359 'p' format code. Use INT_LIST rather than SUBREG as the example of
1360 a code with an XINT and an XEXP. Remove the implication that
1361 accessing an rtx field using XINT is expected to work.
1362 * rtl.def (SUBREG): Change format from "ei" to "ep".
1363 * rtl.h (rtunion::rt_subreg): New field.
1364 (XCSUBREG): New macro.
1365 (SUBREG_BYTE): Use it.
1366 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1367 Update constructor accordingly.
1368 (subreg_shape::operator ==): Update accordingly.
1369 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1370 than an unsigned int.
1371 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1372 a poly_uint64 rather than an unsigned int.
1373 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1374 than an unsigned int.
1375 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1376 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1377 an unsigned int. Take the sizes as poly_uint64s.
1378 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1379 an unsigned int. Take the shift as a poly_uint64 rather than
1381 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1382 as a poly_uint64 rather than an unsigned int.
1383 (simplify_subreg_regno): Likewise.
1384 (byte_lowpart_offset): Return the memory offset as a poly_int64
1386 (subreg_memory_offset): Likewise. Take the subreg offset as a
1387 poly_uint64 rather than an unsigned int.
1388 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1389 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1390 poly_uint64 rather than an unsigned int.
1391 * rtl.c (rtx_format): Describe 'p' in comment.
1392 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1393 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1394 offset as a poly_uint64 rather than an unsigned int.
1395 (byte_lowpart_offset): Return the memory offset as a poly_int64
1397 (subreg_memory_offset): Likewise. Take the subreg offset as a
1398 poly_uint64 rather than an unsigned int.
1399 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1400 mode sizes as poly_uint64s rather than unsigned ints. Return a
1401 poly_uint64 rather than an unsigned int.
1402 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1403 (copy_insn_1): Handle 'p'.
1404 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1405 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1406 an unsigned int. Return the shift in the same way.
1407 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1409 (subreg_size_offset_from_lsb): Take the sizes and shift as
1410 poly_uint64s rather than unsigned ints. Return the offset as
1412 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1413 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1415 * rtlhash.c (add_rtx): Handle 'p'.
1416 * genemit.c (gen_exp): Likewise.
1417 * gengenrtl.c (type_from_format, gendef): Likewise.
1418 * gensupport.c (subst_pattern_match, get_alternatives_number)
1419 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1420 (subst_dup): Likewise.
1421 * gengtype.c (adjust_field_rtx_def): Likewise.
1422 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1423 (match_pattern_2): Likewise.
1424 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1425 (rtx_test::subreg_field): New function.
1426 (operator ==, safe_to_hoist_p, transition_parameter_type)
1427 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1428 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1429 * genpeep.c (match_rtx): Likewise.
1430 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1431 (rtx_writer::print_rtx_operand): Handle 'p'.
1432 (print_value): Handle SUBREG.
1433 * read-rtl.c (apply_int_iterator): Likewise.
1434 (rtx_reader::read_rtx_operand): Handle 'p'.
1435 * alias.c (rtx_equal_for_memref_p): Likewise.
1436 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1437 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1439 * calls.c (expand_call): Likewise.
1440 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1441 (make_extraction, gen_lowpart_for_combine): Likewise.
1442 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1444 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1445 rather than an unsigned int. Treat subreg offsets as poly_ints.
1446 (exp_equiv_p): Handle 'p'.
1447 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1448 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1449 * dse.c (find_shift_sequence): Likewise.
1450 * dwarf2out.c (rtl_for_decl_location): Likewise.
1451 * expmed.c (extract_low_bits): Likewise.
1452 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1453 (expand_expr_real_2): Likewise.
1454 * final.c (alter_subreg): Likewise.
1455 (leaf_renumber_regs_insn): Handle 'p'.
1456 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1457 Treat subreg offsets as poly_ints.
1458 * fwprop.c (forward_propagate_and_simplify): Likewise.
1459 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1460 * ira.c (get_subreg_tracking_sizes): Likewise.
1461 * ira-conflicts.c (go_through_subreg): Likewise.
1462 * ira-lives.c (process_single_reg_class_operands): Likewise.
1463 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1464 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1465 as a poly_uint64 rather than an unsigned int.
1466 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1467 subreg offsets as poly_ints.
1468 * lra-constraints.c (operands_match_p): Handle 'p'.
1469 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1470 * lra-spills.c (assign_mem_slot): Likewise.
1471 * postreload.c (move2add_valid_value_p): Likewise.
1472 * recog.c (general_operand, indirect_operand): Likewise.
1473 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1474 (copyprop_hardreg_forward_1): Likewise.
1475 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1476 (record_subregs_of_mode): Likewise.
1477 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1478 * reload.c (operands_match_p): Handle 'p'.
1479 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1480 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1481 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1482 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1483 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1484 (simplify_const_poly_int_tests<N>::run): Likewise.
1485 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1486 a poly_uint64 rather than an unsigned int.
1487 * valtrack.c (debug_lowpart_subreg): Likewise.
1488 * var-tracking.c (var_lowpart): Likewise.
1489 (loc_cmp): Handle 'p'.
1491 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1492 Alan Hayward <alan.hayward@arm.com>
1493 David Sherwood <david.sherwood@arm.com>
1495 * ira.c (get_subreg_tracking_sizes): New function.
1496 (init_live_subregs): Take an integer size rather than a register.
1497 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1498 to init_live_subregs.
1500 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1501 Alan Hayward <alan.hayward@arm.com>
1502 David Sherwood <david.sherwood@arm.com>
1504 * expr.c (store_constructor_field): Change bitsize from a
1505 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1506 HOST_WIDE_INT to a poly_int64.
1507 (store_constructor): Change size from a HOST_WIDE_INT to
1509 (store_field): Likewise bitsize and bitpos.
1511 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1512 Alan Hayward <alan.hayward@arm.com>
1513 David Sherwood <david.sherwood@arm.com>
1515 * expmed.h (store_bit_field): Change bitregion_start and
1516 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1517 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1518 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1519 (store_fixed_bit_field, store_split_bit_field): Likewise.
1520 * expr.c (store_constructor_field, store_field): Likewise.
1521 (optimize_bitfield_assignment_op): Likewise. Make the same change
1522 to bitsize and bitpos.
1523 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1524 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1525 same change in the constructor arguments.
1526 (get_best_mode): Change bitregion_start and bitregion_end from
1527 unsigned HOST_WIDE_INT to poly_uint64.
1528 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1529 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1531 (bit_field_mode_iterator::next_mode): Update for new types
1532 of m_bitregion_start and m_bitregion_end.
1533 (get_best_mode): Change bitregion_start and bitregion_end from
1534 unsigned HOST_WIDE_INT to poly_uint64.
1536 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1537 Alan Hayward <alan.hayward@arm.com>
1538 David Sherwood <david.sherwood@arm.com>
1540 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1541 accepts poly_uint64 offsets.
1542 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1543 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1544 * expmed.c (lowpart_bit_field_p): Likewise.
1545 (extract_bit_field_as_subreg): New function, split out from...
1546 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1547 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1548 extractions, check that BITSIZE matches the size of the extracted
1549 value and that BITNUM is an exact multiple of that size.
1550 If all else fails, try forcing the value into memory if
1551 BITNUM is variable, and adjusting the address so that the
1552 offset is constant. Split the part that can only handle constant
1553 bitsize and bitnum out into...
1554 (extract_integral_bit_field): ...this new function.
1555 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1556 rather than unsigned HOST_WIDE_INTs.
1558 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1559 Alan Hayward <alan.hayward@arm.com>
1560 David Sherwood <david.sherwood@arm.com>
1562 * expmed.h (store_bit_field): Take bitsize and bitnum as
1563 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1564 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1565 that returns the byte size.
1566 (store_bit_field_1): Take bitsize and bitnum as
1567 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1568 to simple_mem_bitfield_p. Split the part that can only handle
1569 constant bitsize and bitnum out into...
1570 (store_integral_bit_field): ...this new function.
1571 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1572 than unsigned HOST_WIDE_INTs.
1573 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1575 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1576 Alan Hayward <alan.hayward@arm.com>
1577 David Sherwood <david.sherwood@arm.com>
1579 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1580 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1582 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1583 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1584 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1585 offset as a poly_int64 rather than an int.
1586 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1587 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1588 * lra-constraints.c (equiv_address_substitution): Track offsets
1590 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1591 (curr_insn_transform): Handle the new form of sp_offset.
1592 * lra-eliminations.c (lra_elim_table): Change previous_offset
1593 and offset from HOST_WIDE_INT to poly_int64.
1594 (print_elim_table, update_reg_eliminate): Update accordingly.
1595 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1596 (get_elimination): Update accordingly.
1597 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1598 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1599 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1600 poly_int64 offsets generally.
1601 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1602 (mark_not_eliminable, init_elimination): Update accordingly.
1603 (remove_reg_equal_offset_note): Return a bool and pass the new
1604 offset back by pointer as a poly_int64.
1605 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1606 rather than a HOST_WIDE_INT.
1607 (do_remat): Track offsets poly_int64s.
1608 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1610 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1611 Alan Hayward <alan.hayward@arm.com>
1612 David Sherwood <david.sherwood@arm.com>
1614 * rtl.h (mem_attrs): Add a default constructor. Change size and
1615 offset from HOST_WIDE_INT to poly_int64.
1616 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1617 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1618 (widen_memory_access): Take the sizes and offsets as poly_int64s
1619 rather than HOST_WIDE_INTs.
1620 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1621 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1623 (adjust_offset_for_component_ref): Change the offset from a
1624 HOST_WIDE_INT to a poly_int64.
1625 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1626 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1627 * dce.c (find_call_stack_args): Likewise.
1628 * dse.c (record_store): Likewise.
1629 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1630 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1631 * read-rtl-function.c (test_loading_mem): Likewise.
1632 * rtlanal.c (may_trap_p_1): Likewise.
1633 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1634 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1635 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1636 (mem_attrs::mem_attrs): New function.
1637 (set_mem_attributes_minus_bitpos): Change bitpos from a
1638 HOST_WIDE_INT to poly_int64.
1639 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1640 (clear_mem_offset, clear_mem_size, change_address)
1641 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1642 initialize mem_attrs.
1643 (set_mem_offset, set_mem_size, adjust_address_1)
1644 (adjust_automodify_address_1, offset_address, widen_memory_access):
1645 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1647 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1648 Alan Hayward <alan.hayward@arm.com>
1649 David Sherwood <david.sherwood@arm.com>
1651 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1652 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1653 rather than 0 to represent an unknown size. Assert that the size
1654 is known when the mode isn't BLKmode.
1655 (may_trap_p_1): Use -1 for unknown sizes.
1656 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1658 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1659 Alan Hayward <alan.hayward@arm.com>
1660 David Sherwood <david.sherwood@arm.com>
1662 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1663 to poly_int64. Update commentary for positions_needed.large.
1664 (read_info_type): Change offset and width from HOST_WIDE_INT
1666 (set_usage_bits): Likewise.
1667 (canon_address): Return the offset as a poly_int64 rather than
1668 a HOST_WIDE_INT. Use strip_offset_and_add.
1669 (set_all_positions_unneeded, any_positions_needed_p): Use
1670 positions_needed.large to track stores with non-constant widths.
1671 (all_positions_needed_p): Likewise. Take the offset and width
1672 as poly_int64s rather than ints. Assert that rhs is nonnull.
1673 (record_store): Cope with non-constant offsets and widths.
1674 Nullify the rhs of an earlier store if we can't tell which bytes
1676 (find_shift_sequence): Take the access_size and shift as poly_int64s
1678 (get_stored_val): Take the read_offset and read_width as poly_int64s
1679 rather than HOST_WIDE_INTs.
1680 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1681 non-constant offsets and widths.
1683 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1684 Alan Hayward <alan.hayward@arm.com>
1685 David Sherwood <david.sherwood@arm.com>
1687 * inchash.h (inchash::hash::add_poly_int): New function.
1688 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1689 Use poly_int64 rather than HOST_WIDE_INT.
1690 (ao_ref::max_size_known_p): New function.
1691 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1692 rather than HOST_WIDE_INT.
1693 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1694 to temporaries until its interface is adjusted to match.
1695 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1696 (aliasing_component_refs_p, decl_refs_may_alias_p)
1697 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1698 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1699 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1701 * alias.c (ao_ref_from_mem): Likewise.
1702 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1703 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1704 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1705 (maybe_trim_complex_store, maybe_trim_constructor_store)
1706 (live_bytes_read, dse_classify_store): Likewise.
1707 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1708 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1709 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1710 (vn_reference_lookup_3): Likewise.
1711 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1713 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1714 Alan Hayward <alan.hayward@arm.com>
1715 David Sherwood <david.sherwood@arm.com>
1717 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1718 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1719 instead of ranges_overlap_p.
1721 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1722 Alan Hayward <alan.hayward@arm.com>
1723 David Sherwood <david.sherwood@arm.com>
1725 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1726 sizes as poly_int64s rather than HOST_WIDE_INTs.
1728 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1729 Alan Hayward <alan.hayward@arm.com>
1730 David Sherwood <david.sherwood@arm.com>
1732 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1733 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1734 * gimple-fold.c (fold_ctor_reference): Likewise.
1736 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1737 Alan Hayward <alan.hayward@arm.com>
1738 David Sherwood <david.sherwood@arm.com>
1740 * target.def (dwarf_poly_indeterminate_value): New hook.
1741 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1742 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1743 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1744 * doc/tm.texi: Regenerate.
1745 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1746 offset as a poly_int64.
1747 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1748 offset as a poly_int64.
1749 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1750 Take the offset as a poly_int64.
1751 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1752 (frame_pointer_fb_offset): Change to a poly_int64.
1753 (int_loc_descriptor): Take the offset as a poly_int64. Use
1754 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1755 (based_loc_descr): Take the offset as a poly_int64.
1756 Use strip_offset_and_add to handle (plus X (const)).
1757 Use new_reg_loc_descr instead of an open-coded version of the
1758 previous implementation.
1759 (mem_loc_descriptor): Handle CONST_POLY_INT.
1760 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1761 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1763 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1764 Alan Hayward <alan.hayward@arm.com>
1765 David Sherwood <david.sherwood@arm.com>
1767 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1768 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1769 * inchash.h (inchash::hash::add_poly_hwi): New function.
1770 * gengtype.c (main): Register poly_int64.
1771 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1772 offset as a poly_int.
1773 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1774 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1775 offset as a poly_int64.
1776 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1777 * print-rtl.c (print_poly_int): New function.
1778 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1780 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1782 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1783 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1784 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1785 Enforce track_offset_p for parts of a PARALLEL.
1786 (vt_add_function_parameter): Use const_offset for the final
1787 offset to track. Use get_tracked_reg_offset for the parts
1790 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1791 Alan Hayward <alan.hayward@arm.com>
1792 David Sherwood <david.sherwood@arm.com>
1794 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1795 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1796 * doc/tm.texi: Regenerate.
1797 * hooks.h (hook_bool_uint_uint_true): Delete.
1798 (hook_bool_puint64_puint64_true): Declare.
1799 * hooks.c (hook_bool_uint_uint_true): Delete.
1800 (hook_bool_puint64_puint64_true): New function.
1801 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1802 instead of unsigned ints.
1803 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1804 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1806 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1807 Alan Hayward <alan.hayward@arm.com>
1808 David Sherwood <david.sherwood@arm.com>
1810 * optabs.h (expand_operand): Add an int_value field.
1811 (create_expand_operand): Add an int_value parameter and use it
1812 to initialize the new expand_operand field.
1813 (create_integer_operand): Replace with a declaration of a function
1814 that accepts poly_int64s. Move the implementation to...
1815 * optabs.c (create_integer_operand): ...here.
1816 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1817 the mode preserves the value of int_value, instead of calling
1818 const_int_operand on the rtx. Use gen_int_mode to generate
1821 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1822 Alan Hayward <alan.hayward@arm.com>
1823 David Sherwood <david.sherwood@arm.com>
1825 * dumpfile.h (dump_dec): Declare.
1826 * dumpfile.c (dump_dec): New function.
1827 * pretty-print.h (pp_wide_integer): Turn into a function and
1828 declare a poly_int version.
1829 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1831 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1832 Alan Hayward <alan.hayward@arm.com>
1833 David Sherwood <david.sherwood@arm.com>
1835 * doc/generic.texi (POLY_INT_CST): Document.
1836 * tree.def (POLY_INT_CST): New tree code.
1837 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1838 * tree-core.h (tree_poly_int_cst): New struct.
1839 (tree_node): Add a poly_int_cst field.
1840 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1841 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1842 instead of a wide_int_ref.
1843 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1845 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1846 instead of an unsigned HOST_WIDE_INT.
1847 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1848 (ptrdiff_tree_p): Declare.
1849 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1850 extern inline implementations if the target doesn't use POLY_INT_CST.
1851 (poly_int_tree_p): New function.
1852 (wi::unextended_tree): New class.
1853 (wi::int_traits <unextended_tree>): New override.
1854 (wi::extended_tree): Add a default constructor.
1855 (wi::extended_tree::get_tree): New function.
1856 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1857 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1858 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1859 (wi::tree_to_poly_wide_ref): New typedefs.
1860 (wi::ints_for): Provide overloads for extended_tree and
1862 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1863 (wi::to_wide): New functions.
1864 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1865 * tree.c (poly_int_cst_hasher): New struct.
1866 (poly_int_cst_hash_table): New variable.
1867 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1868 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1870 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1871 (init_ttree): Initialize poly_int_cst_hash_table.
1872 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1873 a poly_int64 instead of a HOST_WIDE_INT.
1874 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1875 instead of an unsigned HOST_WIDE_INT.
1876 (wide_int_to_tree): Rename to...
1877 (wide_int_to_tree_1): ...this.
1878 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1879 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1880 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1881 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1882 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1883 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1885 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1886 (streamer_read_tree_body): Likewise.
1887 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1888 (streamer_write_tree_body): Likewise.
1889 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1890 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1891 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1892 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1893 * gimple-expr.h (is_gimple_constant): Likewise.
1894 * gimplify.c (maybe_with_size_expr): Likewise.
1895 * print-tree.c (print_node): Likewise.
1896 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1897 * tree-pretty-print.c (dump_generic_node): Likewise.
1898 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1899 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1900 * tree-vrp.c (compare_values_warnv): Likewise.
1901 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1902 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1903 (force_expr_to_var_cost): Likewise.
1904 * tree-ssa-loop.c (for_each_index): Likewise.
1905 * fold-const.h (build_invariant_address, size_int_kind): Take a
1906 poly_int64 instead of a HOST_WIDE_INT.
1907 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1908 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1910 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1911 (int_const_binop_2): New function, split out from...
1912 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1913 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1914 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1915 * gimple-ssa-strength-reduction.c (slsr_process_add)
1916 (slsr_process_mul): Check for INTEGER_CSTs before using them
1918 * stor-layout.c (bits_from_bytes): New function.
1919 (bit_from_pos): Use it.
1920 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1921 by BITS_PER_UNIT to get the TYPE_SIZE.
1922 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1923 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1925 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1926 Alan Hayward <alan.hayward@arm.com>
1927 David Sherwood <david.sherwood@arm.com>
1929 * doc/rtl.texi (const_poly_int): Document. Also document the
1930 rtl sharing behavior.
1931 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1932 * rtl.h (const_poly_int_def): New struct.
1933 (rtx_def::u): Add a cpi field.
1934 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1935 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1936 (wi::rtx_to_poly_wide_ref): New typedef
1937 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1938 (poly_int_rtx_p): New functions.
1939 (trunc_int_for_mode): Declare a poly_int64 version.
1940 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1941 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1943 (strip_offset): Declare.
1944 (strip_offset_and_add): New function.
1945 * rtl.def (CONST_POLY_INT): New rtx code.
1946 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1947 (shared_const_p): Use poly_int_rtx_p.
1948 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1950 (gen_int_shift_amount): Likewise.
1951 * emit-rtl.c (const_poly_int_hasher): New class.
1952 (const_poly_int_htab): New variable.
1953 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1954 (const_poly_int_hasher::hash): New function.
1955 (const_poly_int_hasher::equal): Likewise.
1956 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1957 (immed_wide_int_const): Rename to...
1958 (immed_wide_int_const_1): ...this and make static.
1959 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1960 instead of a wide_int_ref.
1961 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1962 (gen_lowpart_common): Handle CONST_POLY_INT.
1963 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1964 * cselib.c (cselib_hash_rtx): Likewise.
1965 * dwarf2out.c (const_ok_for_output_1): Likewise.
1966 * expr.c (convert_modes): Likewise.
1967 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1968 * rtlhash.c (add_rtx): Likewise.
1969 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1970 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1971 Handle existing CONST_POLY_INT rtxes.
1972 * expmed.h (expand_shift): Take a poly_int64 instead of a
1974 * expmed.c (expand_shift): Likewise.
1975 * rtlanal.c (strip_offset): New function.
1976 (commutative_operand_precedence): Give CONST_POLY_INT the same
1977 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1979 * rtl-tests.c (const_poly_int_tests): New struct.
1980 (rtl_tests_c_tests): Use it.
1981 * simplify-rtx.c (simplify_const_unary_operation): Handle
1983 (simplify_const_binary_operation): Likewise.
1984 (simplify_binary_operation_1): Fold additions of symbolic constants
1985 and CONST_POLY_INTs.
1986 (simplify_subreg): Handle extensions and truncations of
1988 (simplify_const_poly_int_tests): New struct.
1989 (simplify_rtx_c_tests): Use it.
1990 * wide-int.h (storage_ref): Add default constructor.
1991 (wide_int_ref_storage): Likewise.
1992 (trailing_wide_ints): Use GTY((user)).
1993 (trailing_wide_ints::operator[]): Add a const version.
1994 (trailing_wide_ints::get_precision): New function.
1995 (trailing_wide_ints::extra_size): Likewise.
1997 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1998 Alan Hayward <alan.hayward@arm.com>
1999 David Sherwood <david.sherwood@arm.com>
2001 * emit-rtl.h (gen_int_shift_amount): Declare.
2002 * emit-rtl.c (gen_int_shift_amount): New function.
2003 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2005 * calls.c (shift_return_value): Likewise.
2006 * cse.c (fold_rtx): Likewise.
2007 * dse.c (find_shift_sequence): Likewise.
2008 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2009 (expand_shift, expand_smod_pow2): Likewise.
2010 * lower-subreg.c (shift_cost): Likewise.
2011 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2012 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2013 (expand_vec_perm_var): Likewise.
2014 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2015 (simplify_binary_operation_1): Likewise.
2016 * combine.c (try_combine, find_split_point, force_int_to_mode)
2017 (simplify_shift_const_1, simplify_shift_const): Likewise.
2018 (change_zero_ext): Likewise. Use simplify_gen_binary.
2020 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2022 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2024 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2026 * doc/invoke.texi (ARM Options): Document accepted extension options
2027 for -march=armv8.3-a.
2029 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2032 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2033 or --with-float=softfp, set the default CPU to arm10e.
2035 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2037 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2038 * config/visium/predicates.md (const_shift_operand): Likewise.
2039 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2040 (visium_legitimize_reload_address): Likewise.
2042 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2045 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2047 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2050 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2052 2017-12-20 Julia Koval <julia.koval@intel.com>
2054 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2055 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2056 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2057 * config.gcc: Include vpclmulqdqintrin.h.
2058 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2059 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2061 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2062 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2063 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2064 * config/i386/i386.c (isa2_opts): Add -mcx16.
2065 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2066 (ix86_option_override_internal): Move mcx16 to flags2.
2067 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2068 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2069 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2070 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2071 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2072 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2073 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2074 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2075 * doc/invoke.texi: Add -mvpclmulqdq.
2077 2017-12-20 Tom de Vries <tom@codesourcery.com>
2080 * config/i386/i386.c (ix86_static_chain): Move
2081 DECL_STATIC_CHAIN test ...
2082 * calls.c (rtx_for_static_chain): ... here. New function.
2083 * calls.h (rtx_for_static_chain): Declare.
2084 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2085 instead of targetm.calls.static_chain.
2086 * df-scan.c (df_get_entry_block_def_set): Same.
2088 2017-12-19 Tom de Vries <tom@codesourcery.com>
2090 PR tree-optimization/83493
2091 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2094 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2096 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2097 inform with hardcoded english plural handling.
2099 2017-12-18 Jeff Law <law@redhat.com>
2101 PR tree-optimization/83477
2102 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2103 a non-virtual PHI, always push a new range.
2105 2017-12-19 Martin Sebor <msebor@redhat.com>
2108 * builtins.c (compute_objsize): Handle non-constant offsets.
2110 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2112 PR tree-optimization/83444
2113 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2114 character load case, if get_stridx on MEM_REF's operand doesn't
2115 look usable, retry with get_addr_stridx.
2117 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2120 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2121 (variable_tracking_main_1): Keep markers even when VTA fails.
2124 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2125 even if there are markers before it.
2126 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2128 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2130 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2131 typical order conditions.
2132 * sel-sched.c (extract_new_fences_from): Likewise.
2133 * config/visium/constraints.md (J, K, L): Likewise.
2134 * config/visium/predicates.md (const_shift_operand): Likewise.
2135 * config/visium/visium.c (visium_legitimize_address,
2136 visium_legitimize_reload_address): Likewise.
2137 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2138 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2139 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2140 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2141 SET_DUMP_DETAIL): Likewise.
2142 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2143 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2144 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2145 avr_set_core_architecture, avr_set_current_function,
2146 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2147 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2148 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2149 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2150 avr_map_decompose, avr_fold_builtin): Likewise.
2151 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2152 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2153 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2154 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2155 * config/m32c/m32c.c (m32c_conditional_register_usage,
2156 m32c_address_cost): Likewise.
2157 * config/m32c/predicates.md (shiftcount_operand,
2158 longshiftcount_operand): Likewise.
2159 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2160 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2161 can_use_cdx_ldstw): Likewise.
2162 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2163 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2165 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2166 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2167 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2168 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2169 * config/vax/vax.c (adjacent_operands_p): Likewise.
2170 * config/ft32/constraints.md (L, b, KA): Likewise.
2171 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2173 * cfgexpand.c (expand_stack_alignment): Likewise.
2174 * gcse.c (insert_expr_in_table): Likewise.
2175 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2176 * cgraphunit.c (cgraph_node::expand): Likewise.
2177 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2178 * emit-rtl.c (add_insn): Likewise.
2179 * input.c (dump_location_info): Likewise.
2180 * passes.c (NEXT_PASS): Likewise.
2181 * read-rtl-function.c (parse_note_insn_name,
2182 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2184 * sched-rgn.c (sched_rgn_init): Likewise.
2185 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2186 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2187 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2189 * explow.c (eliminate_constant_term): Likewise.
2190 * final.c (leaf_renumber_regs_insn): Likewise.
2191 * cfgrtl.c (print_rtl_with_bb): Likewise.
2192 * genhooks.c (emit_init_macros): Likewise.
2193 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2194 * tree-data-ref.c (conflict_fn): Likewise.
2195 * selftest.c (assert_streq): Likewise.
2196 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2197 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2198 fold_binary_loc, multiple_of_p): Likewise.
2199 * reload.c (push_reload, find_equiv_reg): Likewise.
2200 * et-forest.c (et_nca, et_below): Likewise.
2201 * dbxout.c (dbxout_symbol_location): Likewise.
2202 * reorg.c (relax_delay_slots): Likewise.
2203 * dojump.c (do_compare_rtx_and_jump): Likewise.
2204 * gengtype-parse.c (type): Likewise.
2205 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2206 simplify_const_relational_operation): Likewise.
2207 * reload1.c (do_output_reload): Likewise.
2208 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2209 * gengtype.c (type_for_name): Likewise.
2210 * gimple-ssa-sprintf.c (format_directive): Likewise.
2212 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2215 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2216 accessing it. Adjust comment.
2218 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2221 * predict.c (zero_one_minusone): New function.
2222 (apply_return_prediction): Avoid return prediction for functions
2223 returning only -1, 0 and 1 values, unless they only return -1 and 0
2226 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2228 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2229 fall-through warning.
2230 (arc_compute_frame_size): Remove unused variables.
2231 (arc_print_operand): Fix fprintif format.
2232 (arc_can_follow_jump): Clean fall-through warning.
2234 2017-12-19 Marek Polacek <polacek@redhat.com>
2237 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2240 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2242 * config/arc/arc.c (overriderregs): New variable.
2243 (arc_override_options): Track fixed/call saved/call options.
2244 (arc_conditional_register_usage): Check against overrideregs
2245 variable whenever we change register properties.
2247 2017-12-19 Nathan Sidwell <nathan@acm.org>
2249 * opts.c (finish_options): Don't prefix dump_base_name if it
2250 already contains directories.
2252 2017-12-19 Martin Liska <mliska@suse.cz>
2254 PR rtl-optimization/82675
2255 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2256 more element in sbitmap.
2258 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2260 * gimplify.c (gimplify_expr): Use error_operand_p.
2262 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2265 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2268 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2270 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2271 the jump is to a label.
2273 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2275 PR tree-optimization/83444
2276 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2281 * ipa-inline.c (flatten_remove_node_hook): New function.
2282 (ipa_inline): Keep only nodes with flatten attribute at the end of
2283 the array in the order from ipa_reverse_postorder, only walk that
2284 portion of array for flattening, if there is more than one such
2285 node, temporarily register a removal hook and ignore removed nodes.
2287 PR tree-optimization/80631
2288 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2289 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2290 IFN_REDUC_MAX or IFN_REDUC_MIN.
2292 2017-12-18 Jeff Law <law@redhat.com>
2294 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2295 record symbolic equivalences from backedges in the CFG.
2298 2017-11-19 Jeff Law <law@redhat.com>
2300 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2301 of degenerates resulting from ignoring an edge.
2303 2017-12-18 Martin Sebor <msebor@redhat.com>
2306 PR tree-optimization/78450
2307 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2308 (handle_builtin_strlen): Call it.
2310 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2312 PR rtl-optimization/83424
2313 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2315 2017-12-18 Marek Polacek <polacek@redhat.com>
2318 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2319 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2322 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2324 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2325 (umaddsidi4, umaddsidi_split): Likewise.
2327 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2329 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2332 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2335 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2337 2017-12-18 Richard Biener <rguenther@suse.de>
2339 PR tree-optimization/81877
2340 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2341 (outermost_indep_loop): Adjust.
2342 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2343 (can_sm_ref_p): Adjust.
2345 2017-12-18 Richard Biener <rguenther@suse.de>
2348 * tree.c (array_at_struct_end_p): Return true if the underlying
2349 object has space for at least one element in excess of what
2350 the array domain specifies.
2352 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2354 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2355 example to avoid overfull hbox.
2356 * doc/invoke.texi (Option Summary): Add missing @gol.
2357 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2360 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2361 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2363 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2364 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2365 (AArch64 Options): Document -mverbose-cost-dump.
2366 (ARM Options): Likewise, plus -mflip-thumb.
2368 2017-12-17 Martin Sebor <msebor@redhat.com>
2371 * gimple-ssa-warn-restrict.c
2372 (builtin_memref::offset_out_of_bounds): Correct the handling of
2375 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2377 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2379 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2381 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2383 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2385 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2388 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2390 * doc/invoke.texi: Fix some typos.
2392 2017-12-16 Martin Sebor <msebor@redhat.com>
2394 PR tree-optimization/78918
2395 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2396 * builtins.c (check_sizes): Rename...
2397 (check_access): ...to this. Rename function arguments for clarity.
2398 (check_memop_sizes): Adjust names.
2399 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2400 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2401 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2402 (check_strncat_sizes, expand_builtin_strncat): Same.
2403 (expand_builtin_strncpy, expand_builtin_memset): Same.
2404 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2405 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2406 (maybe_emit_sprintf_chk_warning): Same.
2407 (expand_builtin_strcpy): Adjust.
2408 (expand_builtin_stpcpy): Same.
2409 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2410 in pointer-checking forms of memcpy, memmove, and mempcpy.
2411 (gcall_to_tree_minimal, max_object_size): Define new functions.
2412 * builtins.h (max_object_size): Declare.
2413 * calls.c (alloc_max_size): Call max_object_size instead of
2414 hardcoding ssizetype limit.
2415 (get_size_range): Handle new argument.
2416 * calls.h (get_size_range): Add a new argument.
2417 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2418 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2419 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2421 (gimple_fold_builtin_memory_chk): Same.
2422 (gimple_fold_builtin_stxcpy_chk): New function.
2423 * gimple-ssa-warn-restrict.c: New source.
2424 * gimple-ssa-warn-restrict.h: New header.
2425 * gimple.c (gimple_build_call_from_tree): Propagate location.
2426 * passes.def (pass_warn_restrict): Add new pass.
2427 * tree-pass.h (make_pass_warn_restrict): Declare.
2428 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2430 (handle_builtin_strcat): Same.
2431 (strlen_optimize_stmt): Rename...
2432 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2433 stpncpy, strncpy, and their checking forms.
2435 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2437 PR rtl-optimization/82849
2438 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2439 and get_max_loop_iterations_int.
2441 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2442 Alan Hayward <alan.hayward@arm.com>
2443 David Sherwood <david.sherwood@arm.com>
2445 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2446 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2447 size as a poly_uint64.
2448 (mode_for_vector, mode_for_int_vector): Take the number of vector
2449 elements as a poly_uint64.
2450 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2451 size as a poly_uint64.
2452 (mode_for_vector, mode_for_int_vector): Take the number of vector
2453 elements as a poly_uint64.
2455 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2456 Alan Hayward <alan.hayward@arm.com>
2457 David Sherwood <david.sherwood@arm.com>
2459 * machmode.h (MACRO_MODE): New macro.
2460 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2461 * targhooks.c (default_libcall_value, default_secondary_reload)
2462 (default_memory_move_cost, default_register_move_cost)
2463 (default_class_max_nregs): Likewise.
2465 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2466 Alan Hayward <alan.hayward@arm.com>
2467 David Sherwood <david.sherwood@arm.com>
2469 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2470 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2472 * genautomata.c (main): Likewise.
2473 * genconditions.c (write_header): Likewise.
2474 * genemit.c (main): Likewise.
2475 * genextract.c (print_header): Likewise.
2476 * genopinit.c (main): Likewise.
2477 * genoutput.c (output_prologue): Likewise.
2478 * genpeep.c (main): Likewise.
2479 * genpreds.c (write_insn_preds_c): Likewise.
2480 * genrecog.c (writer_header): Likewise.
2481 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2482 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2483 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2484 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2485 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2486 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2487 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2488 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2489 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2490 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2491 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2492 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2493 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2494 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2495 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2496 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2497 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2498 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2499 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2500 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2501 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2502 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2503 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2504 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2505 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2506 * config/darwin.c (IN_TARGET_CODE): Likewise.
2507 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2508 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2509 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2510 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2511 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2512 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2513 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2514 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2515 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2516 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2517 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2518 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2519 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2520 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2521 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2522 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2523 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2524 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2525 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2526 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2527 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2528 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2529 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2530 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2531 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2532 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2533 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2534 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2535 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2536 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2537 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2538 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2539 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2540 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2541 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2542 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2543 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2544 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2545 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2546 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2547 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2548 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2549 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2550 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2551 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2552 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2553 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2554 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2555 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2556 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2557 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2558 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2559 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2560 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2561 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2562 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2563 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2564 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2565 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2566 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2567 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2568 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2569 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2570 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2571 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2572 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2573 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2574 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2575 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2576 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2577 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2578 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2579 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2580 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2581 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2582 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2583 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2584 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2585 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2586 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2587 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2588 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2589 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2590 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2591 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2592 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2593 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2594 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2595 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2596 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2597 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2598 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2599 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2600 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2601 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2602 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2603 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2604 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2605 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2606 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2607 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2608 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2609 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2610 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2611 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2612 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2614 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2615 Alan Hayward <alan.hayward@arm.com>
2616 David Sherwood <david.sherwood@arm.com>
2618 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2621 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2622 Alan Hayward <alan.hayward@arm.com>
2623 David Sherwood <david.sherwood@arm.com>
2625 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2626 * doc/md.texi (vec_series@var{m}): Document.
2627 * tree.def (VEC_SERIES_EXPR): New tree code.
2628 * tree.h (build_vec_series): Declare.
2629 * tree.c (build_vec_series): New function.
2630 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2631 * tree-pretty-print.c (dump_generic_node): Likewise.
2632 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2633 * tree-inline.c (estimate_operator_cost): Likewise.
2634 * expr.c (expand_expr_real_2): Likewise.
2635 * optabs-tree.c (optab_for_tree_code): Likewise.
2636 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2637 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2638 * expmed.c (make_tree): Handle VEC_SERIES.
2639 * optabs.def (vec_series_optab): New optab.
2640 * optabs.h (expand_vec_series_expr): Declare.
2641 * optabs.c (expand_vec_series_expr): New function.
2642 * tree-vect-generic.c (expand_vector_operations_1): Check that
2643 the operands also have vector type.
2645 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2646 Alan Hayward <alan.hawyard@arm.com>
2647 David Sherwood <david.sherwood@arm.com>
2649 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2650 (VEC_COND_EXPR): Add missing @tindex.
2651 * doc/md.texi (vec_duplicate@var{m}): Document.
2652 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2653 * tree.c (build_vector_from_val): Add stubbed-out handling of
2654 variable-length vectors, using VEC_DUPLICATE_EXPR.
2655 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2656 * cfgexpand.c (expand_debug_expr): Likewise.
2657 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2658 * tree-inline.c (estimate_operator_cost): Likewise.
2659 * tree-pretty-print.c (dump_generic_node): Likewise.
2660 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2661 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2662 (test_vec_duplicate_folding): New function.
2663 (fold_const_c_tests): Call it.
2664 * optabs.def (vec_duplicate_optab): New optab.
2665 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2666 * optabs.h (expand_vector_broadcast): Declare.
2667 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2668 vec_duplicate_optab.
2669 * expr.c (store_constructor): Try using vec_duplicate_optab for
2671 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2673 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2676 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2677 div/mod latencies a bit.
2679 2017-12-15 Jeff Law <law@redhat.com>
2681 PR tree-optimization/36550
2682 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2683 (mark_threaded_blocks): Rewrite code to avoid block copying when
2684 optimizing for size. Don't pessimize blocks which will be
2685 copied, but all the statements will be dead.
2687 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2689 PR tree-optimization/81165
2690 * tree-ssa-threadupdate.c (uses_in_bb): New.
2691 (estimate_threading_killed_stmts): New.
2692 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2693 * tree-ssa-threadedge.c
2694 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2697 2017-12-15 Jeff Law <law@redhat.com>
2699 PR tree-optimization/83410
2700 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2701 threads when parallelizing loops.
2703 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2705 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2707 * config/alpha/alpha.c (vms_attribute_table): Swap
2708 affects_type_identity and handler fields, adjust comments.
2709 * config/mips/mips.c (mips_attribute_table): Likewise.
2710 * config/visium/visium.c (visium_attribute_table): Likewise.
2711 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2712 * config/microblaze/microblaze.c (microblaze_attribute_table):
2714 * config/spu/spu.c (spu_attribute_table): Likewise.
2715 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2716 * config/arc/arc.c (arc_attribute_table): Likewise.
2717 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2718 * config/v850/v850.c (v850_handle_interrupt_attribute,
2719 v850_handle_data_area_attribute): Formatting fixes.
2720 (v850_attribute_table): Swap affects_type_identity and handler
2721 fields, adjust comments.
2722 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2723 * config/arm/arm.c (arm_attribute_table): Likewise.
2724 * config/avr/avr.c (avr_attribute_table): Likewise.
2725 * config/s390/s390.c (s390_attribute_table): Likewise.
2726 * config/sh/sh.c (sh_attribute_table): Likewise.
2727 * config/i386/i386.c (ix86_handle_cconv_attribute,
2728 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2729 (ix86_attribute_table): Swap affects_type_identity and handler
2730 fields, adjust comments.
2731 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2732 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2733 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2734 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2735 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2736 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2737 * config/rx/rx.c (rx_attribute_table): Likewise.
2738 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2739 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2740 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2741 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2742 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2743 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2744 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2745 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2746 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2747 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2748 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2750 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2752 (register_scoped_attributes, decl_attributes): Formatting fixes.
2754 PR tree-optimization/83269
2755 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2756 subtraction in arg0's type if type is signed and arg0 is unsigned.
2760 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2761 convert? on @0 instead of convert. Check type of @1, not @0.
2762 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2763 convert? on @0 instead of convert. Check type of @1, not @0.
2764 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2765 only optimize if either both @1 and @2 types are narrower
2766 precision, or both are wider or equal precision, and in the former
2767 case only if both have undefined overflow.
2769 2017-12-15 Richard Biener <rguenther@suse.de>
2772 * internal-fn.def (IFN_NOP): Add.
2773 * internal-fn.c (expand_NOP): Do nothing.
2774 * lto-streamer-in.c (input_function): Instead of removing
2775 sanitizer calls replace them with IFN_NOP calls.
2777 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2778 Alan Hayward <alan.hayward@arm.com>
2779 David Sherwood <david.sherwood@arm.com>
2781 * dse.c (store_info, read_info_type): Replace begin and end with
2783 (print_range): New function.
2784 (set_all_positions_unneeded, any_positions_needed_p)
2785 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2787 (record_store): Likewise. Optimize the case in which all positions
2789 (get_stored_val): Replace read_begin and read_end with read_offset
2791 (replace_read): Update call accordingly.
2793 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2795 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2796 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2797 (loop_cand::loop_cand): Initialize above members.
2798 (loop_cand::supported_operations): Delete.
2799 (loop_cand::can_interchange_p): Inline above function.
2800 (loop_cand::classify_simple_reduction): Record number of constant
2801 initialized simple reductions.
2802 (should_interchange_loops): New parameters. Check stmt cost of loops
2804 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2805 Update call to should_interchange_loops.
2806 (should_interchange_loop_nest): Update call to
2807 should_interchange_loops.
2809 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2812 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2813 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2814 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2816 2017-12-15 Julia Koval <julia.koval@intel.com>
2818 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2819 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2821 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2822 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2823 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2825 2017-12-15 Julia Koval <julia.koval@intel.com>
2827 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2828 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2829 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2830 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2831 _mm_aesenc_epi128): New intrinsics.
2833 2017-12-15 Julia Koval <julia.koval@intel.com>
2835 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2836 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2838 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2839 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2840 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2842 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2844 * gimple-ssa-strength-reduction.c (analyze_increments):
2845 Distinguish replacement costs for constant strides from those for
2848 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2850 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2852 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2854 * doc/invoke.texi: Document -Wcast-function-type.
2855 * recog.h (stored_funcptr): Change signature.
2856 * tree-dump.c (dump_node): Avoid warning.
2857 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2859 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2862 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2865 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2867 PR tree-optimization/83312
2868 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2869 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2870 "val" param, and to cope with arbitrary basic blocks.
2871 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2872 handle NULL_TREE for "val", dropping "bb" param.
2873 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2874 drop "bb" param. Handle NULL_TREE for "val".
2875 (find_case_label_for_value): Make "switch_stmt" param const.
2876 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2878 (vrp_prop::check_all_array_refs): Reimplement as...
2879 (check_array_bounds_dom_walker::before_dom_children): ...this new
2880 vfunc. Replace linear search through BB block list, excluding
2881 those with non-executable in-edges via dominator walk.
2883 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2885 * config/arm/arm.opt (mverbose-cost-dump): New option.
2886 * config/arm/arm.c (arm_rtx_costs): Use it.
2888 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2891 * reload1.c (emit_input_reload_insns): Skip debug markers.
2893 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2895 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2899 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2900 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2901 * config/sh/sh.c: Skip debug insns besides notes.
2902 * config/sh/sh.md: Likewise.
2903 * config/sh/sh_treg_combine.cc: Likewise.
2904 * config/sh/sync.md: Likewise.
2906 2017-12-14 Tom de Vries <tom@codesourcery.com>
2908 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2909 Add item for weak_undefined.
2911 2017-12-14 Richard Biener <rguenther@suse.de>
2913 PR tree-optimization/67842
2914 * sese.h (bb_in_region): Remove #if 0'ed code.
2916 2017-12-14 Richard Biener <rguenther@suse.de>
2918 PR tree-optimization/83326
2919 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2920 may_be_zero parameter and handle it by not marking the first
2921 peeled copy as not exiting the loop.
2922 (try_peel_loop): Likewise.
2923 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2924 to handle the case of constant or zero iterations and perform
2925 loop header copying on-the-fly.
2927 2017-12-14 Richard Biener <rguenther@suse.de>
2929 PR tree-optimization/83418
2930 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2931 Instead of asserting we don't get unfolded comparisons deal with them.
2933 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2936 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2937 basic blocks. Assert debug bind insns don't appear outside of bbs,
2938 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2941 PR tree-optimization/83198
2942 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2943 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2944 value if arg is a REAL_CST with incompatible type.
2946 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2947 Bin Cheng <bin.cheng@arm.com>
2950 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2952 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2955 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2957 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2958 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2960 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2961 Alan Hayward <alan.hayward@arm.com>
2962 David Sherwood <david.sherwood@arm.com>
2964 * poly-int.h: New file.
2965 * poly-int-types.h: Likewise.
2966 * coretypes.h: Include them.
2967 (POLY_INT_CONVERSION): Define.
2968 * target.def (estimated_poly_value): New hook.
2969 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2970 * doc/tm.texi: Regenerate.
2971 * doc/poly-int.texi: New file.
2972 * doc/gccint.texi: Include it.
2973 * doc/rtl.texi: Describe restrictions on subreg modes.
2974 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2975 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2976 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2977 * targhooks.h (default_estimated_poly_value): Declare.
2978 * targhooks.c (default_estimated_poly_value): New function.
2979 * target.h (estimated_poly_value): Likewise.
2980 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2981 (wi::unary_traits): Delete.
2982 (wi::binary_traits::signed_shift_result_type): Define for
2983 offset_int << HOST_WIDE_INT, etc.
2984 (generic_wide_int::operator <<=): Define for all types and use
2985 wi::lshift instead of <<.
2986 (wi::hwi_with_prec): Add a default constructor.
2987 (wi::ints_for): New class.
2988 (operator <<): Define for all wide-int types.
2989 (operator /): New function.
2990 (operator %): Likewise.
2991 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2992 (ASSERT_MAYBE_NE_AT): New macros.
2994 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2995 Dominik Vogt <vogt@linux.vnet.ibm.com>
2998 * emit-rtl.c (init_emit): Remove ??? comment.
2999 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3000 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3001 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3002 alignment of 3 virtual registers to BITS_PER_WORD.
3004 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3006 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3008 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3009 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3011 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3014 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3016 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3018 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3019 with the exception of debug begin stmt markers appear before
3023 * final.c (rest_of_handle_final): Call variable_tracking_main only
3024 if !flag_var_tracking.
3026 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3027 Jakub Jelinek <jakub@redhat.com>
3031 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3032 labels debug stmts that can only appear after labels.
3034 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3036 PR rtl-optimization/82398
3037 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3038 EXPR_USEFULNESS in priority comparison.
3040 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3042 PR rtl-optimization/83393
3043 * combine.c (move_deaths): If reg_stat points to a too new insn in
3044 last_death, do not use it: find the proper insn instead.
3046 2017-12-12 Jeff Law <law@redhat.com>
3048 PR tree-optimization/83298
3049 PR tree-optimization/83362
3050 PR tree-optimization/83383
3051 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3052 push_value_range a public interface. Add new argument to
3053 record_ranges_from_stmt.
3054 * gimple-ssa-evrp-analyze.c
3055 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3056 Update comments. Handle recording temporary equivalences.
3057 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3058 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3059 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3060 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3061 gimple-ssa-evrp-analyze.h.
3062 (record_temporary_equivalences_from_phis): Add new argument. When
3063 the PHI arg is an SSA_NAME, set the result's range to the range
3065 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3066 from statements too.
3067 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3068 Pass it down to children as needed.
3069 (thread_outgoing_edges): Likewise.
3070 (thread_across_edge): Likewise. Push/pop range state as needed.
3071 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3073 2017-12-12 Julia Koval <julia.koval@intel.com>
3075 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3076 (PTA_CANNONLAKE): Remove PTA_CLWB.
3078 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3081 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3082 (vcondv2div2df): Likewise.
3083 (vconduv2dfv2di): Likewise.
3085 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3087 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3088 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3089 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3090 ATTR_MATHFN_FPROUNDING_ERRNO.
3092 2017-12-12 Richard Biener <rguenther@suse.de>
3094 PR tree-optimization/83385
3095 * graphite-scop-detection.c (get_order, order): Remove.
3096 (bb_to_rpo): New global.
3098 (build_scops): Sort pbbs in RPO order.
3100 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3102 * combine.c (simplify_set): Do not transform subregs to zero_extends
3103 if the destination mode is a vector mode.
3105 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3107 PR tree-optimization/83359
3108 * tree-cfg.h (fold_loop_internal_call): Declare.
3109 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3110 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3111 (find_loop_dist_alias): New function.
3112 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3113 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3116 PR tree-optimization/80631
3117 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3118 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3119 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3120 hardcoding zero as the value if COND_EXPR is never true. For
3121 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3122 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3123 hardcoding MAX_EXPR as the reduction operation.
3124 (is_nonwrapping_integer_induction): Allow negative step.
3125 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3126 vect_create_epilog_for_reduction, if no value is suitable, don't
3127 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3129 2017-12-12 Richard Biener <rguenther@suse.de>
3131 PR tree-optimization/81889
3132 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3133 range info from the non-wrapping IV instead of just the range
3136 2017-12-12 Julia Koval <julia.koval@intel.com>
3138 * config.gcc: Add vaesintrin.h.
3139 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3141 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3142 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3144 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3145 * config/i386/immintrin.h: Include vaesintrin.h.
3146 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3147 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3148 _mm_aesdec_epi128): New intrinsics.
3150 2017-12-12 Julia Koval <julia.koval@intel.com>
3152 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3153 OPTION_MASK_ISA_VAES_UNSET): New.
3154 (ix86_handle_option): Handle -mvaes.
3155 * config/i386/cpuid.h: Define bit_VAES.
3156 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3157 * config/i386/i386-c.c (__VAES__): New.
3158 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3159 (ix86_valid_target_attribute_inner_p): Ditto.
3160 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3161 * config/i386/i386.opt: Add -mvaes.
3162 * doc/invoke.texi: Ditto.
3164 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3166 * debug.h (gcc_debug_hooks): Add inline_entry.
3167 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3168 * debug.c (do_nothing_debug_hooks): Likewise.
3169 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3170 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3171 (dwarf2_lineno_debug_hooks): Likewise.
3173 * common.opt (gstatement-frontiers): New, setting
3174 debug_nonbind_markers_p.
3175 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3176 * toplev.c (process_options): Autodetect value for debug statement
3178 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3179 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3181 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3182 markers. Integrate source bind into debug stmt expand loop.
3183 (pass_expand::execute): Check debug marker limit. Avoid deep
3184 TER and expand debug locations for debug bind insns only.
3185 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3187 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3188 * final.c (reemit_insn_block_notes): Take current block from
3189 nonbind markers. Declare note where it's first set.
3190 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3191 begin stmt markers if enabled.
3192 (notice_source_line): Handle nonbind markers. Fail if their
3193 location is unknown or that of builtins.
3194 (rest_of_handle_final): Convert begin stmt markers to notes if
3195 var-tracking didn't run.
3196 (rest_of_clean_state): Skip begin stmt markers.
3197 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3199 * function.c (allocate_struct_function): Set begin_stmt_markers.
3200 * function.h (struct function): Add debug_marker_count counter
3201 and debug_nonbind_markers flag.
3202 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3203 * gimple-low.c (lower_function_body): Adjust
3204 debug_nonbind_markers.
3205 (lower_stmt): Drop or skip gimple debug stmts.
3206 (lower_try_catch): Skip debug stmts.
3207 * gimple.c (gimple_build_debug_begin_stmt): New.
3208 (gimple_copy): Increment debug_marker_count if copying one.
3209 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3210 * gimplify.c (rexpr_location): New.
3211 (rexpr_has_location): New.
3212 (warn_switch_unreachable_r): Handle gimple debug stmts.
3213 (shortcut_cond_r): Call expr_location.
3215 (find_goto_label): New.
3216 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3218 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3220 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3221 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3222 (LANG_HOOKS_INITIALIZER): ... this.
3223 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3224 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3226 (update_ebb_live_info): Skip debug insn markers.
3227 * lra.c (debug_insn_static_data): Rename to...
3228 (debug_bind_static_data): ... this.
3229 (debug_marker_static_data): New.
3230 (lra_set_insn_recog_data): Select one of the above depending
3232 (lra_update_isn_regno_info): Don't assume debug insns have
3234 (push_insns): Skip debug insns.
3235 * lto-streamer-in.c (input_function): Drop debug stmts
3236 depending on active options. Adjust debug_nonbind_markers.
3237 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3238 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3239 begin stmt marker notes.
3240 (print_insn): Likewise.
3241 * recog.c (extract_insn): Recognize rtl for debug markers.
3242 * rtl.def (DEBUG_MARKER): New.
3243 * tree-inline.c: Include params.h.
3244 (remap_gimple_stmt): Handle nonbind markers.
3245 (maybe_move_debug_stmts_to_successors): Likewise.
3246 (copy_debug_stmt): Likewise.
3247 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3248 markers regardless of no side effects.
3249 (tsi_link_before): Don't update container's side effects when adding
3250 a begin stmt marker.
3251 (tsi_link_after): Likewise.
3252 (expr_first): Skip begin stmt markers.
3253 (expr_last): Likewise.
3254 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3255 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3256 Disregard nonbind markers.
3257 * tree.c (make_node_stat): Don't set side effects for begin stmt
3259 (build1_stat): Likewise.
3260 * tree.def (DEBUG_BEGIN_STMT): New.
3261 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3262 * var-tracking.c (delete_debug_insns): Renamed to...
3263 (delete_vta_debug_insns): ... this.
3264 (reemit_marker_as_note): New.
3265 (vt_initialize): Reemit markers.
3266 (delete_vta_debug_insns): Likewise.
3267 (vt_debug_insns_local): Reemit or delete markers.
3268 (variable_tracking_main_1): Likewise.
3269 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3270 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3271 (gimple_debug_nonbind_marker_p): New.
3272 (gimple_build_debug_bind): Adjust.
3273 (gimple_build_debug_begin_stmt): New.
3274 * doc/invoke.texi (max-debug-marker-count): New param.
3275 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3276 (debug_parameter_ref, debug_marker): New.
3277 (NOTE_INSN_BEGIN_STMT): New.
3278 (DEBUG_INSN): Describe begin stmt markers.
3280 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3281 without debug insns, we wouldn't, but clean up debug insns
3282 after a control flow insn nevertheless.
3284 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3285 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3286 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3287 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3288 (rtl_tidy_fallthru_edge): Likewise.
3289 (rtl_verify_fallthru): Likewise.
3290 (rtl_verify_bb_layout): Likewise.
3291 (skip_insns_after_block): Likewise.
3292 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3293 * dwarf2out.c: Include print-rtl.h.
3294 (dwarf2out_next_real_insn): New.
3295 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3296 Dump debug binds in asm comments.
3297 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3298 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3299 callers to use gsi_start_nondebug_bb instead.
3300 (gsi_after_labels): Skip gimple debug stmts.
3301 (gsi_start_nondebug): New.
3302 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3303 (proper_loop_form_for_interchange): Adjust.
3304 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3305 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3306 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3307 (collect_fallthrough_labels): Likewise.
3308 (should_warn_for_implicit_fallthrough): Likewise.
3309 (warn_implicit_fallthrough_r): Likewise.
3310 (expand_FALLTHROUGH_r): Likewise.
3311 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3312 (graphite_copy_stmts_from_block): Skip nonbind markers.
3313 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3314 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3315 * jump.c (clean_barriers): Skip debug insns.
3316 * omp-expand.c (expand_parallel_call): Skip debug insns.
3317 (expand_task_call): Likewise.
3318 (remove_exit_barrier): Likewise.
3319 (expand_omp_taskreg): Likewise.
3320 (expand_omp_for_init_counts): Likewise.
3321 (expand_omp_for_generic): Likewise.
3322 (expand_omp_for_static_nochunk): Likewise.
3323 (expand_omp_for_static_chunk): Likewise.
3324 (expand_omp_simd): Likewise.
3325 (expand_omp_taskloop_for_outer): Likewise.
3326 (expand_omp_taskloop_for_inner): Likewise.
3327 (expand_oacc_for): Likewise.
3328 (expand_omp_sections): Likewise.
3329 (expand_omp_single): Likewise.
3330 (expand_omp_synch): Likewise.
3331 (expand_omp_atomic_load): Likewise.
3332 (expand_omp_atomic_store): Likewise.
3333 (expand_omp_atomic_fetch_op): Likewise.
3334 (expand_omp_atomic_pipeline): Likewise.
3335 (expand_omp_atomic_mutex): Likewise.
3336 (expand_omp_target): Likewise.
3337 (grid_expand_omp_for_loop): Likewise.
3338 (grid_expand_target_grid_body): Likewise.
3339 (build_omp_regions_1): Likewise.
3340 * omp-low.c (check_combined_parallel): Skip debug stmts.
3341 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3342 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3343 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3345 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3346 (make_edges): Likewise.
3347 (cleanup_dead_labels): Likewise.
3348 (gimple_can_merge_blocks_p): Likewise.
3349 (stmt_starts_bb_p): Likewise.
3350 (gimple_block_label): Likewise.
3351 (gimple_redirect_edge_and_branch): Likewise.
3352 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3354 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3356 * tree-pretty-print (print_declaration): Omit initializer in slim
3358 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3360 (eliminate_unnecessary_stmts): Stabilize block removal order.
3361 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3362 * var-tracking.c (get_first_insn): New.
3363 (vt_emit_notes): Call it.
3364 (vt_initialize): Walk any insns before the first BB.
3365 (delete_debug_insns): Likewise.
3367 * gimple.h (enum gimple_debug_subcode): Add
3368 GIMPLE_DEBUG_BEGIN_STMT.
3369 (gimple_debug_begin_stmt_p): New.
3370 (gimple_debug_nonbind_marker_p): New.
3371 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3372 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3373 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3374 * insn-notes.def (BEGIN_STMT): New.
3375 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3376 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3377 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3378 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3379 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3380 (INSN_DEBUG_MARKER_KIND): New.
3381 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3382 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3383 (INSN_VAR_LOCATION_PTR): New.
3384 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3386 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3387 deps only if debug bind insns are enabled.
3388 (pass_expand::execute): Avoid deep TER and expand
3389 debug locations for debug bind insns only.
3390 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3391 debug stmts special handling down to debug bind stmts.
3392 * combine.c (try_combine): Narrow debug insns special handling
3393 down to debug bind insns.
3394 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3395 Narrow debug insns preexisting special handling down to debug
3397 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3398 handling down to debug bind insns.
3399 * function.c (instantiate_virtual_regs): Skip debug markers,
3400 adjust handling of debug binds.
3401 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3402 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3403 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3404 insns to debug bind insns.
3405 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3406 special handling of debug stmts to debug bind stmts.
3407 * ipa-split.c (split_function): Likewise.
3408 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3409 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3411 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3412 * regrename.c (build_def_use): Likewise.
3413 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3414 (pass_cprop_hardreg): Narrow special casing of debug insns to
3416 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3417 * reload1.c (reload): Likewise.
3418 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3419 casing of debug stmts to debug bind stmts.
3420 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3421 * ssa-iterators.h (num_imm_uses): Likewise.
3422 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3423 debug stmts to debug bind stmts.
3424 * tree-inline.c (tree_function_versioning): Narrow special casing
3425 of debug stmts to debug bind stmts.
3426 * tree-loop-distribution.c (generate_loops_for_partition):
3427 Narrow special casing of debug stmts to debug bind stmts.
3428 * tree-sra.c (analyze_access_subtree): Narrow special casing
3429 of debug stmts to debug bind stmts.
3430 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3431 stmts to debug bind stmts.
3432 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3433 casing of debug stmts to debug bind stmts.
3434 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3435 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3436 casing of debug stmts to debug bind stmts.
3437 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3439 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3440 debug stmts to debug bind stmts.
3441 (gimple_replace_ssa_lhs): Likewise.
3442 (insert_debug_temp_for_var_def): Likewise.
3443 (insert_debug_temps_for_defs): Likewise.
3444 (reset_debug_uses): Likewise.
3445 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3446 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3447 (adjust_debug_stmts): Likewise.
3448 (adjust_phi_and_debug_stmts): Likewise.
3449 (vect_do_peeling): Likewise.
3450 * tree-vect-loop.c (vect_transform_loop): Likewise.
3451 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3452 * var-tracking.c (adjust_mems): Narrow special casing of debug
3453 insns to debug bind insns.
3454 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3455 (compute_bb_dataflow, vt_find_locations): Likewise.
3456 (vt_expand_loc, emit_notes_for_changes): Likewise.
3457 (vt_init_cfa_base): Likewise.
3458 (vt_emit_notes): Likewise.
3459 (vt_initialize): Likewise.
3460 (vt_finalize): Likewise.
3462 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3463 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3464 (next_nonnote_nondebug_insn_bb): New.
3465 (prev_nonnote_nondebug_insn_bb): New.
3466 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3467 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3468 (prev_nonnote_nondebug_insn_bb): Declare.
3469 (next_nonnote_nondebug_insn_bb): Declare.
3470 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3471 * cfgrtl.c (get_last_bb_insn): Likewise.
3472 * lra.c (push_insns): Likewise.
3474 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3477 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3479 * selftest.c (selftest::test_runner::test_runner): New ctor.
3480 (selftest::test_runner::~test_runner): New dtor.
3481 * selftest.h (class selftest::test_runner): New class.
3483 2017-12-11 Carl Love <cel@us.ibm.com>
3485 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3486 vec_extract_fp32_from_shortl]): Add #defines.
3487 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3488 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3489 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3490 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3491 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3492 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3493 * doc/extend.texi: Add documentation for the added builtins.
3495 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3497 PR rtl-optimization/80693
3498 PR rtl-optimization/81019
3499 PR rtl-optimization/81020
3500 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3501 are not mentioned in i3. Place the REG_UNUSED note on i2,
3502 possibly modified to REG_DEAD, if it did not originate in i3.
3504 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3506 * recog.c (store_data_bypass_p_1): New function.
3507 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3508 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3510 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3512 PR rtl-optimization/83361
3513 * ifcvt.c (if_convert): Call fixup_partitions.
3515 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3517 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3518 early folding of splat_u{8,16,32}.
3520 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3522 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3523 output_operand_lossage first argument with capital letter.
3524 (aarch64_override_options): Don't start error and sorry first argument
3525 with capital letter.
3527 2017-12-11 Andi Kleen <ak@linux.intel.com>
3529 PR gcov-profile/83355
3530 * auto-profile.c (string_table::get_index_by_decl): Don't
3531 recurse when abstract origin points to itself.
3533 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3535 PR tree-optimization/83320
3536 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3537 (prune_datarefs_not_in_loop): Ditto.
3539 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3541 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3543 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3545 PR tree-optimization/83337
3546 * gimple-loop-interchange.cc (compute_access_stride): Handle
3547 bitfield DRs properly.
3549 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3551 PR tree-optimization/83338
3552 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3553 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3554 vector of unsigned integers to vector of signed integers.
3556 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3558 PR rtl-optimization/83317
3559 * lra-constraints.c (process_address_1): Add insn code check.
3561 2017-12-08 Michael Matz <matz@suse.de>
3563 Fix PR tree-optimization/83323
3564 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3565 head-controlled loops and loop BBs.
3566 * common.opt (funroll-and-jam): Remove, instead ...
3567 (floop-unroll-and-jam): ... reuse this option.
3568 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3569 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3570 (-floop-unroll-and-jam): ... this option.
3572 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3574 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3575 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3576 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3578 2017-12-08 Richard Biener <rguenther@suse.de>
3581 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3582 handle accesses outside of zero-sized vars.
3584 2017-12-08 Martin Jambor <mjambor@suse.cz>
3586 PR tree-optimization/83141
3587 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3588 test for MEM_REFs implicitely changing types with padding. Remove
3590 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3592 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3594 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3596 * config/avr/avr.c (avr_attribute_table): Likewise.
3597 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3598 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3599 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3600 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3601 of struct attribute_spec.
3602 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3604 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3607 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3609 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3612 * profile-count.c (profile_count::from_gcov_type): Move from
3613 profile-count.h; handle overflow.
3614 * profile-count.h (profile_count::from_gcov_type): Move offline.
3616 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3618 PR rtl-optimization/83304
3619 * combine.c (move_deaths): If we do not know where a register died,
3622 2017-12-08 Richard Biener <rguenther@suse.de>
3624 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3625 Provide -fopt-info-loop feedback when we interchange in a nest.
3627 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3629 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3630 for armv6 ARM CPU IDs.
3632 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3634 * common/config/arm/arm-common.c: Include <algorithm>.
3635 (INCLUDE_VECTOR): Define.
3636 (compare_opt_names): New function.
3637 (arm_rewrite_selected_arch): Only strip out extensions that can be
3638 expressed through -mfpu. Sort the remaining extensions
3641 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3643 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3644 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3645 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3646 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3648 2017-06-08 Tristan Gingold <gindold@adacore.com>
3651 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3652 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3654 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3656 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3657 of struct attribute_spec.
3659 2017-12-08 Julia Koval <julia.koval@intel.com>
3661 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3662 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3663 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3664 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3665 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3666 _mm_maskz_dpwssds_epi32): Ditto.
3668 2017-12-08 Richard Biener <rguenther@suse.de>
3670 PR tree-optimization/81303
3671 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3672 conditions try to create a comparison vector type matching
3673 the data vector type.
3674 (vectorizable_condition): Adjust.
3675 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3676 Leave invariant conditions alone in case we can vectorize those.
3678 2017-12-08 Julia Koval <julia.koval@intel.com>
3680 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3681 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3682 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3683 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3684 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3686 2017-12-08 Julia Koval <julia.koval@intel.com>
3688 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3689 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3690 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3691 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3692 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3693 _mm_maskz_dpbusds_epi32): New intrinsics.
3695 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3697 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3699 (ld<bh>io_signed): Likewise.
3700 (st<bhw>io): Likewise.
3701 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3702 SMALL_INT12 constant integer operand.
3704 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3706 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3708 (nios2_large_constant_allowed): ...this. Adjust uses.
3709 (nios2_plus_symbolic_constant_p): Rename to...
3710 (nios2_plus_large_constant_p): ...this. Adjust uses.
3711 (nios2_legitimate_address_p): Correct CONST_INT handling.
3712 (nios2_symbolic_memory_operand_p): Rename to...
3713 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3714 (nios2_large_constant_p): Check for large constant integers too.
3715 (nios2_split_large_constant): Handle constant integers.
3716 (nios2_split_symbolic_memory_operand): Rename to...
3717 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3718 (nios2_legitimize_constant_address): Handle constant integers.
3719 (r0rel_constant_p): Handle small constant integers.
3720 (nios2_print_operand_address): Handle r0-relative integer addresses.
3721 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3722 * config/nios2/nios2.md: Adjust for renamed functions.
3724 2017-12-07 Andrew Waterman <andrew@sifive.com>
3726 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3727 (TARGET_HAVE_SRODATA_SECTION): New define.
3728 (riscv_select_section): New function.
3730 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3731 Alexander Monakov <amonakov@ispras.ru>
3732 Jakub Jelinek <jakub@redhat.com>
3735 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3737 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3739 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3740 a bool success value. Don't call output_operand_lossage here.
3741 (aarch64_print_ldpstp_address): Return a bool success value.
3742 (aarch64_print_operand_address): Call output_addr_const if
3743 aarch64_print_address_internal fails.
3744 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3745 'y'; call output_operand_lossage instead. Call output_operand_lossage
3746 if aarch64_print_ldpstp_address fails.
3748 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3750 * tree-vector-builder.h
3751 (tree_vector_builder::binary_encoded_nelts): Declare.
3752 * tree-vector-builder.c
3753 (tree_vector_builder::binary_encoded_nelts): New function.
3754 * fold-const.c (negate_expr_p): Likewise.
3755 (operand_equal_p, fold_checksum_tree): Likewise.
3756 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3757 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3758 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3759 (uniform_vector_p): Likewise.
3760 * varasm.c (const_hash_1, compare_constant): Likewise.
3761 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3762 (valid_lattice_transition): Operate directly on the VECTOR_CST
3764 * ipa-icf.c: Include tree-vector-builder.h.
3765 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3766 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3768 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3770 * tree.c (build_vector): Delete.
3771 * tree.h (build_vector): Make static and move into the self-testing
3774 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3776 * vector-builder.h (vector_builder::derived): New const overload.
3777 (vector_builder::elt): New function.
3778 * tree-vector-builder.h (tree_vector_builder::type): New function.
3779 (tree_vector_builder::apply_step): Declare.
3780 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3782 * gimple-fold.h (tree_vector_builder): Declare.
3783 (gimple_build_vector): Take a tree_vector_builder instead of a
3784 type and vector of elements.
3785 * gimple-fold.c (gimple_build_vector): Likewise.
3786 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3788 (get_initial_defs_for_reduction): Likewise.
3789 (vectorizable_induction): Likewise.
3791 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3793 * tree-vector-builder.h
3794 (tree_vector_builder::new_binary_operation): Declare.
3795 * tree-vector-builder.c
3796 (tree_vector_builder::new_binary_operation): New function.
3797 * fold-const.c (fold_relational_const): Use it.
3798 (const_binop): Likewise. Check that both input vectors have
3799 the same number of elements, thus excluding things like WIDEN_SUM.
3800 Check whether it is possible to operate directly on the encodings
3803 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3805 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3806 new_unary_operation, operating only on the encoded elements.
3807 (const_unop): Likewise.
3808 (exact_inverse): Likewise.
3809 (distributes_over_addition_p): New function.
3810 (const_binop): Use tree_vector_builder and new_unary_operation
3811 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3812 on the encoded elements unless the encoding is strided and the
3813 operation does not distribute over addition.
3814 (fold_convert_const): Use tree_vector_builder and
3815 new_unary_operation. Operate only on the encoded elements
3816 for truncating integer conversions, or for non-stepped encodings.
3818 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3820 * config/sparc/sparc.c: Include tree-vector-builder.h.
3821 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3822 * expmed.c: Include tree-vector-builder.h.
3823 (make_tree): Use tree_vector_builder instead of build_vector.
3824 * fold-const.c: Include tree-vector-builder.h.
3825 (const_binop): Use tree_vector_builder instead of build_vector.
3826 (const_unop): Likewise.
3827 (native_interpret_vector): Likewise.
3828 (fold_vec_perm): Likewise.
3829 (fold_ternary_loc): Likewise.
3830 * gimple-fold.c: Include tree-vector-builder.h.
3831 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3833 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3834 (simplify_vector_constructor): Use tree_vector_builder instead
3836 * tree-vect-generic.c: Include tree-vector-builder.h.
3837 (add_rshift): Use tree_vector_builder instead of build_vector.
3838 (expand_vector_divmod): Likewise.
3839 (optimize_vector_constructor): Likewise.
3840 * tree-vect-loop.c: Include tree-vector-builder.h.
3841 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3842 of build_vector. Explicitly use a stepped encoding for
3844 * tree-vect-slp.c: Include tree-vector-builder.h.
3845 (vect_get_constant_vectors): Use tree_vector_builder instead
3847 (vect_transform_slp_perm_load): Likewise.
3848 (vect_schedule_slp_instance): Likewise.
3849 * tree-vect-stmts.c: Include tree-vector-builder.h.
3850 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3851 (vect_gen_perm_mask_any): Likewise.
3852 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3853 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3855 (build_vector_from_val): Likewise. Explicitly use a duplicate
3858 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3860 * doc/generic.texi (VECTOR_CST): Describe new representation of
3862 * vector-builder.h: New file.
3863 * tree-vector-builder.h: Likewise.
3864 * tree-vector-builder.c: Likewise.
3865 * Makefile.in (OBJS): Add tree-vector-builder.o.
3866 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3867 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3868 (tree_vector): Change the number of elements to
3869 vector_cst_encoded_nelts.
3870 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3871 (VECTOR_CST_ELTS): Delete.
3872 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3873 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3874 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3875 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3876 (VECTOR_CST_ENCODED_ELT): Likewise.
3877 (vector_cst_encoded_nelts): New function.
3878 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3879 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3880 (vector_cst_int_elt, vector_cst_elt): Declare.
3881 * tree.c: Include tree-vector-builder.h.
3882 (tree_code_size): Abort if passed VECTOR_CST.
3883 (tree_size): Update for new VECTOR_CST layout.
3884 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3885 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3886 (build_vector): Use tree_vector_builder.
3887 (vector_cst_int_elt, vector_cst_elt): New functions.
3888 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3889 encoded elements and then create the vector in the canonical form.
3890 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3891 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3892 (tree_c_tests): Call test_vector_cst_patterns.
3893 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3895 (hash_tree): Likewise.
3896 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3897 (streamer_write_tree_header): Likewise.
3898 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3899 (streamer_alloc_tree): Likewise. Update call to make_vector.
3900 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3902 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3904 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3905 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3906 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3908 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3909 Richard Biener <rguenther@suse.de>
3911 PR tree-optimization/81303
3912 * Makefile.in (gimple-loop-interchange.o): New object file.
3913 * common.opt (floop-interchange): Reuse the option from graphite.
3914 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3915 -floop-interchange and mention it for -O3.
3916 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3917 * gimple-loop-interchange.cc: New file.
3918 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3919 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3920 * passes.def (pass_linterchange): New pass.
3921 * timevar.def (TV_LINTERCHANGE): New time var.
3922 * tree-pass.h (make_pass_linterchange): New declaration.
3923 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3924 interchange. Record IV before/after increment in new parameters.
3925 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3926 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3928 (check_reduction_path): ...New function here.
3929 * tree-vectorizer.h (check_reduction_path): New declaration.
3931 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3934 PR rtl-optimization/80818
3935 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3936 always early clobbered.
3937 * lra-lives.c (process_bb_lives): Check input hard regs for early
3938 clobbered non-operand hard reg.
3940 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3943 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3944 types_compatible_p, just that TYPE_MODE is the same.
3946 2017-12-07 Martin Sebor <msebor@redhat.com>
3949 * attribs.c (empty_attribute_table): Initialize new member of
3950 struct attribute_spec.
3951 (decl_attributes): Add argument. Handle mutually exclusive
3952 combinations of attributes.
3953 (selftests::test_attribute_exclusions): New function.
3954 (selftests::attribute_c_tests): Ditto.
3955 * attribs.h (decl_attributes): Add default argument.
3956 * selftest.h (attribute_c_tests): Declare.
3957 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3958 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3960 * doc/extend.texi (Common Function Attributes): Update const and pure.
3961 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3962 of struct attribute_spec.
3963 * config/arc/arc.c (arc_attribute_table): Same.
3964 * config/arm/arm.c (arm_attribute_table): Same.
3965 * config/avr/avr.c ( avr_attribute_table): Same.
3966 * config/bfin/bfin.c (bfin_attribute_table): Same.
3967 * config/cr16/cr16.c (cr16_attribute_table): Same.
3968 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3969 * config/h8300/h8300.c (h8300_attribute_table): Same.
3970 * config/i386/i386.c (ix86_attribute_table): Same.
3971 * config/ia64/ia64.c (ia64_attribute_table): Same.
3972 * config/m32c/m32c.c (m32c_attribute_table): Same.
3973 * config/m32r/m32r.c (m32r_attribute_table): Same.
3974 * config/m68k/m68k.c (m68k_attribute_table): Same.
3975 * config/mcore/mcore.c (mcore_attribute_table): Same.
3976 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3977 * config/mips/mips.c (mips_attribute_table): Same.
3978 * config/msp430/msp430.c (msp430_attribute_table): Same.
3979 * config/nds32/nds32.c (nds32_attribute_table): Same.
3980 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3981 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3982 * config/rl78/rl78.c (rl78__attribute_table): Same.
3983 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3984 * onfig/rx/rx.c (rx_attribute_table): Same.
3985 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3986 * config/sh/sh.c (sh_attribute_table): Same.
3987 * config/sparc/sparc.c (sparc_attribute_table): Same.
3988 * config/spu/spu.c (spu_attribute_table): Same.
3989 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3990 * config/v850/v850.c (v850_attribute_table): Same.
3991 * config/visium/visium.c (visium_attribute_table): Same.
3993 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3996 * config/arm/arm.c (INCLUDE_STRING): Define.
3997 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3998 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4001 2017-12-07 Michael Matz <matz@suse.de>
4003 Add unroll and jam pass
4005 * gimple-loop-jam.c: New file.
4006 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4007 * common.opt (funroll-and-jam): New option.
4008 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4009 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4010 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4011 * passes.def: Add pass_loop_jam.
4012 * timevar.def (TV_LOOP_JAM): Add.
4013 * tree-pass.h (make_pass_loop_jam): Declare.
4014 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4015 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4016 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4017 to flow_loop_tree_node_add.
4018 (duplicate_subloops, copy_loops_to): Append to sibling list.
4019 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4020 * doc/invoke.texi (-funroll-and-jam): Document new option.
4021 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4023 2017-12-07 Richard Biener <rguenther@suse.de>
4025 PR tree-optimization/83296
4026 PR tree-optimization/67769
4027 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4028 flow sensitive info in an unrelated BB.
4029 (value_replacement): Use reset_flow_sensitive_info.
4030 (minmax_replacement): Reset flow sensitive info on the def
4031 we move. Do not reset flow sensitive info in the whole BB
4032 we move the stmt to.
4033 (abs_replacement): Likewise.
4035 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4038 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4039 rs6000_cpu to the given -mcpu=, or to the default processor.
4041 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4043 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4044 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4045 instead of rs6000_cpu_attr.
4046 (rs6000_adjust_cost): Ditto.
4047 (is_microcoded_insn): Ditto.
4048 (rs6000_adjust_priority): Ditto.
4049 (rs6000_issue_rate): Ditto.
4050 (rs6000_use_sched_lookahead): Ditto.
4051 (rs6000_use_sched_lookahead_guard): Ditto.
4052 (rs6000_sched_reorder): Ditto.
4053 (force_new_group): Ditto.
4054 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4055 (group_ending_nop): Ditto.
4057 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4059 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4060 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4061 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4062 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4064 (rs6000_reassociation_width): Ditto.
4065 (rs6000_emit_epilogue): Ditto.
4066 (rs6000_adjust_cost): Ditto.
4067 (is_microcoded_insn): Ditto.
4068 (is_cracked_insn): Ditto.
4069 (rs6000_adjust_priority): Ditto.
4070 (rs6000_sched_reorder): Ditto.
4071 (rs6000_sched_reorder2): Ditto.
4072 (insn_must_be_first_in_group): Ditto.
4073 (insn_must_be_last_in_group): Ditto.
4074 (rs6000_register_move_cost): Ditto.
4075 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4078 2017-12-07 Julia Koval <julia.koval@intel.com>
4080 * config.gcc: Add vaesintrin.h.
4081 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4083 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4084 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4086 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4087 * config/i386/immintrin.h: Include vaesintrin.h.
4088 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4089 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4090 _mm_aesdec_epi128): New intrinsics.
4092 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4094 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4095 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4096 c-family/c-spellcheck.cc.
4097 (best_macro_match::best_macro_match): Likewise.
4099 (struct edit_distance_traits<cpp_hashnode *>): Move to
4100 c-family/c-spellcheck.h.
4101 (class best_macro_match): Likewise.
4103 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4105 PR tree-optimization/83293
4106 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4107 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4108 might insert into empty bb.
4111 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4112 simplify for plus with :c added, and pointer_plus without that.
4113 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4114 with undefined overflow and the conversion is not widening,
4115 perform negation in utype and only convert to type afterwards.
4116 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4117 simplify for plus with :c added, and pointer_plus without that.
4118 If type is integral with undefined overflow and the conversion is
4119 not widening, perform minus in utype and only convert to type
4120 afterwards. Move the last pointer_diff_expr simplify into the
4123 2017-12-06 Martin Sebor <msebor@redhat.com>
4125 PR tree-optimization/82646
4126 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4127 strncpy, not maxlen.
4129 2017-12-06 Martin Sebor <msebor@redhat.com>
4131 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4134 PR tree-optimization/83075
4135 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4136 strncat/strncpy don't change length of source string.
4138 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4141 2017-11-29 Martin Aberg <maberg@gaisler.com>
4143 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4144 to prevent b2bst errata sequence.
4145 (sqrtdf2_fix): Likewise.
4147 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4149 PR tree-optimization/81945
4150 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4151 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4152 to dest_cfun has orig_loop_num set, either remap it to the new
4153 loop number if the loop got moved too, or clear it.
4155 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4157 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4159 (thunderx2t99_nothing): New insn reservation.
4160 (thunderx2t99_mrs): New insn reservation.
4161 (thunderx2t99_multiple): New insn reservation.
4162 (thunderx2t99_alu_basi): Add bfx to reservation.
4163 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4165 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4168 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4170 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4172 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4173 tree-ssa-pre.c::remove_dead_inserted_code.
4174 * tree-ssa-dce.h: New file.
4175 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4176 (remove_dead_inserted_code): Move and rename to function
4177 tree-ssa-dce.c::simple_dce_from_worklist.
4178 (pass_pre::execute): Update use.
4180 2017-12-05 Richard Biener <rguenther@suse.de>
4182 PR tree-optimization/83277
4183 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4184 to code-gen liveout vars.
4186 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4188 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4189 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4190 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4193 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4195 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4197 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4198 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4201 2017-12-05 Richard Biener <rguenther@suse.de>
4203 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4204 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4205 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4206 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4208 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4210 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4212 2017-12-05 Martin Liska <mliska@suse.cz>
4213 Jakub Jelinek <jakub@redhat.com>
4215 * doc/invoke.texi: Document the options.
4216 * flag-types.h (enum sanitize_code): Add
4217 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4218 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4219 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4220 * opts.c: Define new sanitizer options.
4221 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4222 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4224 2017-12-05 Julia Koval <julia.koval@intel.com>
4226 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4227 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4228 (ix86_handle_option): Handle -mavx512vnni.
4229 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4230 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4231 * config/i386/i386-c (__AVX512VNNI__): New.
4232 * config/i386/i386.c (ix86_target_string): Handle new option.
4233 (ix86_valid_target_attribute_inner_p): Handle new option.
4234 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4235 * config/i386/i386.opt (mavx512vnni): New option.
4237 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4240 * athlon.md: Disable for generic.
4241 * haswell.md: Enable for generic.
4242 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4243 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4245 (ix86_adjust_cost): Move generic to haswell path.
4247 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4249 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4250 instead of MEM_P in a couple more places. Fix formatting issues.
4252 2017-12-04 Jim Wilson <jimw@sifive.com>
4254 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4255 instead of GP_REG_LAST-1.
4256 (riscv_adjust_libcall_cfi_prologue): Likewise.
4257 (riscv_adjust_libcall_cri_epilogue): Likewise.
4258 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4261 2017-12-04 Luis Machado <luis.machado@linaro.org>
4263 * ipa-pure-const.c (check_decl): Add missing newline.
4264 (state_from_flags): Likewise.
4266 2017-12-04 Jeff Law <law@redhat.com>
4268 PR tree-optimizatin/78496
4269 * gimple-ssa-evrp-analyze.h
4270 (evrp_range_analyzer::get_vr_values): Simplify.
4271 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4272 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4273 and gimple-ssa-evrp-analyze.h.
4274 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4275 (simplify_stmt_for_jump_threading): Copy a blob of code from
4276 tree-vrp.c to use ranges to simplify statements.
4277 (dom_opt_dom_walker::before_dom_children): Call
4278 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4279 (dom_opt_dom_walker::after_dom_children): Similarly for
4280 evrp_range_analyzer::leave.
4281 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4284 * gimple-ssa-evrp-analyze.c
4285 (evrp_range_analyzer::extract_range_from_stmt): Always use
4286 vr_values::update_value_range so preexisting range info is
4287 medged with new range info, even if the new range is VR_VARYING.
4289 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4291 * combine.c: Adjust comment.
4292 (use_crosses_set_p): Delete.
4293 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4294 (try_combine): Ditto.
4296 2017-12-04 Richard Biener <rguenther@suse.de>
4298 PR tree-optimization/83255
4299 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4300 Re-add zero-iteration check.
4302 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4304 PR rtl-optimization/83245
4305 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4306 hard registers as earlyclobber, also if not in an asm.
4308 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4312 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4315 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4316 rs6000_cpu based on cpu_index, not tune_index.
4318 2017-12-04 Richard Biener <rguenther@suse.de>
4320 PR tree-optimization/83238
4321 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4322 code match comment, rejecting invalid SESE regions.
4324 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4326 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4327 require base operand is a REG_POINTER prior to reload on targets
4328 with non-equivalent space registers.
4330 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4332 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4333 (update_specialized_profile): Fix updating of counts.
4334 (perhaps_add_new_callers): Likewise.
4336 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4339 * x86-tune.def: Remove obsolette FIXMEs.
4340 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4341 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4342 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4344 (X86_TUNE_PAD_RETURNS): Disable for generic.
4346 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4348 PR tree-optimization/83170
4349 PR tree-optimization/83241
4350 * gimple-ssa-store-merging.c
4351 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4352 gimple_vuse (ins_stmt) in case it has changed.
4353 (imm_store_chain_info::output_merged_store): Likewise.
4355 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4359 * tree-cfg.c (pass_warn_function_return::execute): Handle
4360 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4361 with BUILTINS_LOCATION.
4365 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4366 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4367 instead of DECL_MODE.
4369 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4370 last argument to rtx pointer.
4371 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4372 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4373 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4374 depending on the chosen ISAs.
4375 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4376 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4377 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4379 * config/i386/sse.md (mov<mode>_internal): Likewise.
4380 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4382 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4384 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4386 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4388 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4389 parameters from prototype.
4390 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4391 parameters. Don't print a comment.
4392 (emit_fusion_gpr_load): Adjust.
4393 (emit_fusion_load_store): Adjust.
4394 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4395 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4396 comment on the second line.
4398 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4401 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4402 rs6000_cpu based on cpu_index, not tune_index.
4404 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4406 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4407 which_alternative instead of which_alternative + 1.
4408 (output_asm_insn): Print an extra tab if the template is short.
4410 2017-12-01 Jim Wilson <jimw@sifive.com>
4412 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4414 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4415 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4417 * doc/tm.texi: Regenerate.
4419 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4422 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4423 whether we can allocate pseudos before trying to fix an address.
4424 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4425 memory address is indexed or indirect.
4426 (floatuns_<mode>si2_hw2): Likewise.
4428 2017-12-01 Jason Merrill <jason@redhat.com>
4430 * Makefile.in (TAGS): Add c-family/*.cc.
4432 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4434 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4435 (call_value_insn): Likewise.
4436 (sibcall_insn): Likewise.
4437 (sibcall_value_insn): Likewise.
4438 (movsi_aarch64): Likewise.
4439 (movdi_aarch64): Likewise.
4440 (add_losym_): Likewise.
4441 (ldr_got_small_): Likewise.
4442 (ldr_got_small_sidi): Likewise.
4443 (ldr_got_small_28k_): Likewise.
4444 (ldr_got_small_28k_sidi): Likewise.
4445 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4446 Move output_addr_const to symbolic case. Add error check.
4448 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4450 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4451 * config/s390/s390.md: Disable MVC merging peephole if it would
4452 disable operand forwarding.
4453 (new peephole2): Split MVCs if it would turn them into up to 2
4456 2017-12-01 Richard Biener <rguenther@suse.de>
4458 PR tree-optimization/83232
4459 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4460 detection of same access. Instead of breaking the group here
4461 do not consider the duplicate. Add comment explaining real fix.
4463 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4465 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4467 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4469 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4471 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4473 * function.h (struct function): Remove cilk_frame_decl,
4474 is_cilk_function and calls_cilk_spawn fields.
4475 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4477 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4478 cilk_elemental field.
4479 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4480 * target.def: Adjust comment.
4481 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4482 Don't test cilk_elemental.
4484 PR tree-optimization/83233
4485 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4486 bswap_stat name for the struct.
4489 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4490 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4491 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4492 SWITCH_BREAK_LABEL_P set on the label.
4493 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4494 added for default case if it was missing and not all cases covered.
4495 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4496 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4498 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4499 to avoid -Wimplicit-fallthrough warning.
4500 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4501 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4503 PR tree-optimization/83221
4504 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4506 (init_reassoc): Formatting fix.
4509 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4510 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4514 * tree-cfg.c: Include asan.h.
4515 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4517 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4519 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4521 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4523 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4524 vpcmpeqd instruction.
4526 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4528 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4529 registers type for 128bit mode.
4531 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4533 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4535 * gengtype-state.c (read_state_pair): Likewise.
4536 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4537 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4538 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4539 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4540 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4541 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4542 * builtins.c (fold_builtin_3): Likewise.
4543 * graphite-scop-detection.c
4544 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4545 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4548 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4551 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4552 parameter. Return true if ignore_decl_rtl_set_p is true and other
4553 conditions are satisfied.
4554 * asan.h (asan_protect_global): Add new parameter.
4555 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4556 to asan_protect_global calls.
4558 2017-11-30 Jim Wilson <jimw@sifive.com>
4560 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4561 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4562 -mstrict-align, add info on default value. Delete redundant lines for
4563 -mabi. Add missing -mexplicit-relocs docs.
4565 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4567 * config/arc/arc.md (trap): New pattern.
4569 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4571 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4572 instruction to end into a delay slot.
4573 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4574 can be placed into a delay slot against reg_note.
4576 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4578 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4579 labels number of usages.
4581 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4583 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4585 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4587 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4590 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4591 multiplication by power of 2 constant into two shifts + comparison.
4593 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4596 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4598 2017-11-30 Richard Biener <rguenther@suse.de>
4600 PR tree-optimization/83202
4601 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4602 allow_peel argument and guard peeling.
4603 (canonicalize_loop_induction_variables): Likewise.
4604 (canonicalize_induction_variables): Pass false.
4605 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4606 peeling from cunrolli.
4608 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4610 * combine.c (try_combine): Print a message to dump file whenever
4611 I0, I1, or I2 cannot be combined into I3.
4613 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4615 PR rtl-optimization/83156
4616 PR rtl-optimization/82621
4617 * combine.c (try_combine): Don't split an I2 if one of the dests is
4618 set again before I3. Allow unused dests.
4620 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4622 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4624 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4626 PR rtl-optimization/80818
4627 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4628 recursively. Use insn code for clobber.
4629 (lra_set_insn_recog_data): Pass the new arg to
4630 collect_non_operand_hard_regs.
4631 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4633 (lra_update_insn_regno_info): Pass insn to
4634 add_regs_to_insn_regno_info.
4636 2017-11-29 Jim Wilson <jimw@sifive.com>
4637 Andrew Waterman <andrew@sifive.com>
4639 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4640 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4641 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4642 use SINGLE_SHIFT_COST.
4643 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4644 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4646 2017-11-29 Julia Koval <julia.koval@intel.com>
4648 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4649 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4650 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4651 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4652 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4653 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4654 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4655 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4656 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4657 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4658 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4659 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4660 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4661 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4662 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4663 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4664 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4665 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4666 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4667 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4668 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4669 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4670 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4671 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4672 __builtin_ia32_vpshldv_v2di_mask,
4673 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4674 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4675 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4677 2017-11-29 Julia Koval <julia.koval@intel.com>
4679 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4680 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4681 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4682 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4683 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4684 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4685 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4686 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4687 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4688 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4689 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4690 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4691 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4692 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4693 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4694 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4695 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4696 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4697 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4698 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4699 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4701 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4703 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4704 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4707 2017-11-29 Martin Aberg <maberg@gaisler.com>
4709 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4710 to prevent b2bst errata sequence.
4711 (sqrtdf2_fix): Likewise.
4713 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4715 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4716 (div_sqrt_insn_p): New function.
4717 (sparc_do_work_around_errata): Insert NOP instructions to
4718 prevent sequences that could trigger the TN-0013 errata for
4719 certain LEON3 processors.
4720 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4721 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4722 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4723 (in_branch_delay): Prevent div and sqrt in delay slot if
4725 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4727 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4729 * config/sparc/sparc.c (atomic_insn_p): New function.
4730 (sparc_do_work_around_errata): Insert NOP instructions to
4731 prevent sequences that could trigger the TN-0010 errata for
4733 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4734 instruction referable in atomic_insns_p.
4736 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4738 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4739 (atomic_compare_and_swap_leon3_1): Likewise.
4742 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4744 * config/sparc/sparc.c (fpop_insn_p): New function.
4745 (sparc_do_work_around_errata): Insert NOP instructions to
4746 prevent sequences that could trigger the TN-0012 errata for
4748 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4749 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4750 (in_branch_annul_delay): Prevent floating-point instructions
4751 in delay slot of annulled integer branch.
4753 2017-11-29 Richard Biener <rguenther@suse.de>
4755 PR tree-optimization/83202
4756 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4758 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4759 nodes vectorized to the same stmts multiple times.
4760 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4761 (vect_analyze_slp_instance): Adjust.
4762 (scalar_stmts_to_slp_tree_map_t): New typedef.
4763 (vect_schedule_slp_instance): Add a map recording the SLP node
4764 representing the vectorized stmts for a set of scalar stmts.
4765 Avoid code-generating redundancies.
4766 (vect_schedule_slp): Allocate map and pass it down.
4768 2017-11-29 Nathan Sidwell <nathan@acm.org>
4771 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4774 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4776 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4777 cases for printing LDP/STP memory addresses.
4778 (aarch64_print_address_internal): Renamed from
4779 aarch64_print_operand_address, added parameter, add Pmode check.
4780 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4781 (aarch64_print_operand_address): Indirect to
4782 aarch64_print_address_internal.
4783 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4784 'y' operand output specifier.
4786 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4789 * tree.c (build_simple_mem_ref_loc): Handle
4790 get_addr_base_and_unit_offset returning a MEM_REF.
4793 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4796 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4799 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4801 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4803 (layout::move_to_column): Likewise.
4805 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4808 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4809 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4811 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4812 Martin Jambor <mjambor@suse.cz>
4815 * tree.h (expr_type_first_operand_type_p): Declare
4816 * tree.c (expr_type_first_operand_type_p): New function.
4817 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4818 (ipa_value_from_jfunc): Adjust declaration.
4819 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4820 Use it as result type for arithmetics, unless it is NULL in which case
4821 be more conservative.
4822 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4823 ipa_get_jf_pass_through_result.
4824 (propagate_vals_across_pass_through): Likewise.
4825 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4826 is to propagate_vals_across_pass_through.
4827 (propagate_constants_across_call): Pass PARM_TYPE to
4828 propagate_scalar_across_jump_function.
4829 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4830 ipa_value_from_jfunc.
4831 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4832 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4833 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4834 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4835 target_type, pass it to ipa_value_from_jfunc.
4836 (update_indirect_edges_after_inlining): Pass parameter type to
4837 try_make_edge_direct_simple_call.
4839 2017-11-28 Jeff Law <law@redhat.com>
4841 * gimple-ssa-evrp-analyze.c
4842 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4843 refine ranges if scev_initialized_p returns true.
4844 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4846 2017-11-28 Julia Koval <julia.koval@intel.com>
4848 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4849 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4850 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4851 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4852 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4853 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4854 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4855 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4856 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4857 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4858 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4859 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4860 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4861 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4862 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4863 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4864 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4865 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4866 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4867 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4868 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4869 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4871 2017-11-28 Julia Koval <julia.koval@intel.com>
4873 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4874 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4875 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4876 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4877 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4878 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4879 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4880 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4881 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4882 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4883 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4884 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4885 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4886 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4887 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4888 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4889 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4890 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4891 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4892 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4893 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4894 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4895 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4896 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4897 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4898 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4899 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4900 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4901 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4902 __builtin_ia32_vpshld_v2di_mask): New builtins.
4903 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4904 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4906 2017-11-28 Richard Biener <rguenther@suse.de>
4908 PR tree-optimization/80776
4909 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4911 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4913 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4914 If the incoming edge is an effective fallthru because the other
4915 edge only reaches a __builtin_unreachable () then record ranges
4916 derived from the controlling condition in SSA info.
4917 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4918 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4920 2017-11-28 Olivier Hainque <hainque@adacore.com>
4922 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4923 on mingw build hosts.
4925 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4928 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4930 2017-11-28 Richard Biener <rguenther@suse.de>
4933 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4934 copies generated from memcpy use a character array as reference
4937 2017-11-28 Julia Koval <julia.koval@intel.com>
4938 Sebastian Peryt <sebastian.peryt@intel.com>
4940 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4941 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4942 cilk-common.o, cilk.h, cilk-common.c): Remove.
4944 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4945 * builtins.c (is_builtin_name): Remove cilkplus condition.
4946 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4947 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4948 cilk-builtins.def, cilkplus.def): Remove.
4949 * cif-code.def (CILK_SPAWN): Remove.
4950 * cilk-builtins.def: Delete.
4951 * cilk-common.c: Ditto.
4953 * cilkplus.def: Ditto.
4954 * config/darwin.h (fcilkplus): Delete.
4955 * cppbuiltin.c: Ditto.
4956 * doc/extend.texi: Remove cilkplus doc.
4957 * doc/generic.texi: Ditto.
4958 * doc/invoke.texi: Ditto.
4959 * doc/passes.texi: Ditto.
4960 * gcc.c (fcilkplus): Remove.
4961 * gengtype.c (cilk.h): Remove.
4962 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4964 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4966 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4968 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4969 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4970 cilkplus conditions.
4971 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4972 inline_read_section): Ditto.
4973 * ipa-inline-analysis.c (cilk.h): Remove.
4974 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4975 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4976 append_linker_options): Remove condition for fcilkplus.
4977 * lto/lto-lang.c (cilk.h): Remove.
4978 (lto_init): Remove condition for fcilkplus.
4979 * omp-expand.c (expand_cilk_for_call): Delete.
4980 (expand_omp_taskreg, expand_omp_for_static_chunk,
4981 expand_omp_for): Remove cilkplus
4983 (expand_cilk_for): Delete.
4984 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4985 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4986 execute_lower_omp, diagnose_sb_0): Ditto.
4987 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4988 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4989 * tree-nested.c: Ditto.
4990 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4991 (dump_generic_node): Ditto.
4992 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4993 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4995 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4997 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4999 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5000 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5001 for vector mode and !TARGET_SIMD.
5003 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5005 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5007 * tree.h (SWITCH_LABELS): Remove.
5008 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5009 assert SWITCH_BODY is non-NULL.
5010 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5012 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5014 PR tree-optimization/80788
5015 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5016 has TREE_OVERFLOW set, call drop_tree_overflow.
5018 2017-11-28 Richard Biener <rguenther@suse.de>
5020 PR tree-optimization/83158
5021 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5023 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5026 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5027 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5029 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5031 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5032 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5034 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5037 * varasm.c (compare_constant): Take the mode of the constants into
5038 account when comparing floating point constants.
5040 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5042 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5043 from explicit instantiation of debug_helper.
5044 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5046 2017-11-27 Richard Biener <rguenther@suse.de>
5048 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5051 2017-11-27 Richard Biener <rguenther@suse.de>
5053 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5054 (type_hash_canon): Also clear int_cst_hash_table entry for
5056 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5058 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5060 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5061 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5062 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5064 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5066 * hash-map.h (gt_cleare_cache): Avoid UB.
5068 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5070 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5071 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5072 (decide_unroll_runtime_iterations): Likewise.
5073 (decide_unroll_stupid): Likewise.
5075 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5078 * config/i386/i386.md: Add a loop with incssp.
5080 2017-11-27 Martin Jambor <mjambor@suse.cz>
5082 PR tree-optimization/81248
5083 * tree-sra.c (splice_param_accesses): Remove size check.
5084 (decide_one_param_reduction): Fix size check.
5085 * gimple-pretty-print.c (dump_profile): Silence warning.
5086 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5088 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5091 * dbxout.c (lastlineno): New variable.
5092 (dbx_debug_hooks): Use dbxout_switch_text_section as
5093 switch_text_section debug hook.
5094 (dbxout_function_end): Switch to current_function_section
5095 rather than function_section. If crtl->has_bb_partition,
5096 output just one N_FUN, depending on in_cold_section_p.
5097 (dbxout_source_line): Remember last lineno in lastlineno.
5098 (dbxout_switch_text_section): New function.
5099 (dbxout_function_decl): Adjust dbxout_block caller.
5100 (dbx_block_with_cold_children): New function.
5101 (dbxout_block): Return true if any LBRAC/RBRAC have been
5102 emitted. Use dbx_block_with_cold_children at depth == 0
5103 in second partition. Add PARENT_BLOCKNUM argument, pass
5104 it optionally adjusted to children. Output LBRAC/RBRAC
5105 around recursive call only if the block is in the current
5106 partition, if not and anything was output, emit empty
5108 * final.c (final_scan_insn): Compute cold_function_name
5109 before calling switch_text_section debug hook. Call
5110 that hook even if dwarf2out_do_frame if not emitting
5114 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5115 TREE_READONLY decls.
5117 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5119 PR rtl-optimization/82488
5120 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5122 2017-11-26 Julia Koval <julia.koval@intel.com>
5124 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5126 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5129 2017-11-26 Julia Koval <julia.koval@intel.com>
5131 * config/i386/driver-i386.c (host_detect_local_cpu):
5132 Detect skylake-avx512.
5134 2017-11-26 Julia Koval <julia.koval@intel.com>
5136 * config.gcc: Add -march=cannonlake.
5137 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5138 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5139 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5140 (PTA_CANNONLAKE): New.
5141 (processor_target_table): Add cannonlake.
5142 (ix86_option_override_internal): Ditto.
5143 (fold_builtin_cpu): Ditto.
5144 (get_builtin_code_for_version): Handle cannonlake.
5145 (M_INTEL_COREI7_CANNONLAKE): New.
5146 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5147 * doc/invoke.texi: Add -march=cannonlake.
5149 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5151 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5152 (try_init_one_plugin): Alternative implementation for MinGW.
5153 * Makefile.in (plugin_implib): New.
5154 (gengtype-lex.c): Fix broken AIX workaround.
5155 * configure: Regenerate.
5156 * doc/plugins.texi: Document support for MinGW.
5158 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5160 PR rtl-optimization/81553
5161 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5162 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5163 is a shift where C1 has different mode than the whole shift, use C1's
5164 mode for MULT rather than the shift's mode.
5167 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5168 builtins not enabled in the currently selected ISA.
5170 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5172 PR tree-optimization/71026
5173 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5174 (insert_reciprocals): Change to insert reciprocals before a division
5175 by a square and to insert the square of a reciprocal.
5176 (execute_cse_reciprocals_1): Change to consider division by a square.
5177 (register_division_in): Add importance parameter.
5179 2017-11-24 Richard Biener <rguenther@suse.de>
5181 PR tree-optimization/82402
5182 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5183 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5185 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5187 * match.pd (0-ptr): New transformation.
5189 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5192 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5193 when skipping checking.
5195 2017-11-24 Richard Biener <rguenther@suse.de>
5197 PR tree-optimization/83128
5198 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5199 (vn_reference_lookup_3): Likewise.
5201 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5204 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5205 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5208 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5209 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5210 check for the latter with gimple_call_builtin_p. Do not handle
5211 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5213 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5215 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5218 2017-11-23 Julia Koval <julia.koval@intel.com>
5220 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5221 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5222 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5223 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5224 _mm512_maskz_expandloadu_epi16): New intrinsics.
5225 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5226 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5227 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5228 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5229 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5230 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5231 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5232 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5233 _mm256_maskz_expandloadu_epi8): New intrinsics.
5234 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5235 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5236 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5237 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5238 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5239 * config/i386/sse.md (VI248_VLBW): New iterator.
5240 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5242 2017-11-23 Julia Koval <julia.koval@intel.com>
5244 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5245 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5246 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5247 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5248 _mm512_mask_compressstoreu_epi16): New.
5249 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5250 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5251 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5252 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5253 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5254 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5255 _mm256_mask_compressstoreu_epi8): New.
5256 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5257 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5258 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5259 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5260 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5261 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5262 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5263 __builtin_ia32_compresshi128_mask,
5264 __builtin_ia32_compressstoreuqi512_mask,
5265 __builtin_ia32_compressstoreuhi512_mask,
5266 __builtin_ia32_compressstoreuqi256_mask,
5267 __builtin_ia32_compressstoreuqi128_mask,
5268 __builtin_ia32_compressstoreuhi256_mask,
5269 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5270 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5272 (ix86_expand_special_args_builtin): Handle new types.
5273 (s4fma_expand): Handle new builtin array.
5274 * config/i386/immintrin.h: Include new headers.
5275 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5276 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5278 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5280 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5281 fixes. Declare temp and g variables at the top in order to avoid
5282 {} in most of the cases.
5284 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5286 * match.pd (ptr-0): New transformation.
5288 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5290 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5291 (struct addr_mode_cost_table): New.
5292 (struct tune_params): Add field addr_mode_costs.
5293 * config/arm/arm.c (generic_addr_mode_costs): New.
5294 (arm_slowmul_tune): Initialise addr_mode_costs field.
5295 (arm_fastmul_tune): Likewise.
5296 (arm_strongarm_tune): Likewise.
5297 (arm_xscale_tune): Likewise.
5298 (arm_9e_tune): Likewise.
5299 (arm_marvell_pj4_tune): Likewise.
5300 (arm_v6t2_tune): Likewise.
5301 (arm_cortex_tune): Likewise.
5302 (arm_cortex_a8_tune): Likewise.
5303 (arm_cortex_a7_tune): Likewise.
5304 (arm_cortex_a15_tune): Likewise.
5305 (arm_cortex_a35_tune): Likewise.
5306 (arm_cortex_a53_tune): Likewise.
5307 (arm_cortex_a57_tune): Likewise.
5308 (arm_exynosm1_tune): Likewise.
5309 (arm_xgene1_tune): Likewise.
5310 (arm_cortex_a5_tune): Likewise.
5311 (arm_cortex_a9_tune): Likewise.
5312 (arm_cortex_a12_tune): Likewise.
5313 (arm_cortex_a73_tune): Likewise.
5314 (arm_v7m_tune): Likewise.
5315 (arm_cortex_m7_tune): Likewise.
5316 (arm_v6m_tune): Likewise.
5317 (arm_fa726te_tune): Likewise.
5318 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5320 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5322 * config/arm/arm.c (arm_mem_costs): New function.
5323 (arm_rtx_costs_internal): Use arm_mem_costs.
5325 2017-11-23 Mark Wielaard <mark@klomp.org>
5327 * dwarf2out.c (init_sections_and_labels): Use generation to create
5328 unique ranges_section_label and ranges_base_label. Return generation.
5329 (output_rnglists): Add generation argument. Use generation to create
5330 unique ranges labels.
5331 (dwarf2out_finish): Get generation from init_sections_and_labels
5332 and pass generation to output_rnglists.
5334 2017-11-23 Mike Stump <mikestump@comcast.net>
5335 Eric Botcazou <ebotcazou@adacore.com>
5337 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5338 * cfgloop.h (struct loop): Add unroll field.
5339 * function.h (struct function): Add has_unroll bitfield.
5340 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5341 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5342 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5343 (pass_rtl_unroll_loops::gate): Likewise.
5344 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5345 for which loop->unroll==1.
5346 (decide_unroll_constant_iterations): Use note for consistency and
5347 take loop->unroll into account. Return early if loop->unroll is set.
5348 Fix thinko in existing test.
5349 (decide_unroll_runtime_iterations): Use note for consistency and
5350 take loop->unroll into account.
5351 (decide_unroll_stupid): Likewise.
5352 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5353 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5354 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5356 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5357 (print_loop): Print loop->unroll if set.
5358 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5359 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5360 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5362 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5363 loop->unroll is set and smaller than the trip count. Otherwise bypass
5364 entirely the heuristics if loop->unroll is set. Remove dead note.
5365 Fix off-by-one bug in other note.
5366 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5367 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5369 (tree_unroll_loops_completely): Make static.
5370 (pass_complete_unroll::execute): Use correct type for variable.
5371 (pass_complete_unrolli::execute): Fix formatting.
5372 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5374 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5376 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5377 enable when TARGET_PREFER_AVX128 is set.
5379 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5381 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5383 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5384 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5386 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5388 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5389 frequencies. Fix estimation of aggregate parameters.
5391 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5393 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5396 2017-11-23 Tom de Vries <tom@codesourcery.com>
5398 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5400 2017-11-23 Tom de Vries <tom@codesourcery.com>
5402 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5405 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5408 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5409 sibcall_value_pcrel_fdpic): Use local variable instead of
5411 (calli_tbr_rel): Add missing operand 2.
5412 (call_valuei_tbr_rel): Add missing operand 3.
5414 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5417 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5418 bitpos/bitsize covering the whole destination, use store_expr only if
5419 the complex mode is the same. Otherwise, use expand_normal and if
5420 it returns CONCAT, subreg each part separately instead of trying to
5421 subreg the whole result.
5423 2017-11-23 Richard Biener <rguenther@suse.de>
5425 PR tree-optimization/23094
5426 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5427 come here from walking over backedges in the first iteration.
5428 (vn_reference_lookup_3): Skip clobbers that store the same value.
5430 2017-11-23 Richard Biener <rguenther@suse.de>
5432 PR tree-optimization/81403
5433 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5434 a block we need a leader relative to.
5435 (phi_translate_1): For nary processing require a leader from
5436 get_representative_for given we run expression simplification
5437 using match-and-simplify. Remove previous fix.
5439 2017-11-22 Jeff Law <law@redhat.com>
5441 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5442 Use new method allocate_value_range rather than accessing the
5443 vrp_value_range_pool data member directly.
5444 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5445 to use extract_range_from_stmt method to avoid need for
5446 extract_range_from_assignment method.
5447 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5448 method rather than setting values_propgated data member directly.
5449 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5450 and values propagated data members and extract_range_from_assignment
5451 method. Reorder private data members to conform to standards.
5452 Add new methods set_lattice_propagation_complete and
5453 allocate_value_range.
5455 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5457 PR rtl-optimization/83030
5458 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5459 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5460 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5461 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5462 CROSSING_JUMP_P flag is already set before setting it.
5463 * cfgrtl.c (fixup_partition_crossing): Likewise.
5464 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5467 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5469 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5470 Handle the case where both arguments are using gen_const_vec_series.
5472 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5475 * pretty-print.c (pp_format): Move quoting implementation to
5476 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5477 to pass address of "quote" local.
5478 (pp_begin_quote): New function.
5479 (pp_end_quote): New function.
5480 * pretty-print.h (printer_fn): Convert penultimate param from bool
5482 (pp_begin_quote): New decl.
5483 (pp_end_quote): New decl.
5484 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5485 param from bool to bool *.
5486 * tree-diagnostic.h (default_tree_printer): Likewise.
5488 2017-11-22 Jeff Law <law@redhat.com>
5490 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5491 Initialize vr_values.
5492 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5493 vr_values via vr_values class instance rather than delegators.
5494 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5495 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5496 (evrp_range_analyzer::push_value_range): Likewise.
5497 (evrp_range_analyzer::pop_value_range): Likewise.
5498 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5499 most delegators. Those remaining are exposed as public interfaces.
5500 Make vr_values a pointer and private.
5501 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5503 (evrp_range_analyzer::get_vr_value): New method.
5504 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5505 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5506 (class evrp_dom_walker): Attach evrp_folder class, initialize
5507 it in the ctor. Remove temporary delegators.
5508 (evrp_dom_walker::before_dom_children): Call methods in attached
5509 evrp_range_analyzer class via class instance pointer. Use
5510 free value_range_constant_singleton to remove need for
5511 op_with_constant_singleton_value delegator method. Do not
5512 create a vrp_prop class instance for every call! Narrow
5513 scope of a couple variables.
5514 (evrp_dom_walker::cleanup): Call methods in attached
5515 evrp_range_analyzer class via class instance pointer.
5516 * vr-values.h (class vr_values): Privatize many methods and
5519 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5521 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5522 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5523 * expr.c (expand_expr_real_2): Likewise.
5524 * fold-const.c (const_unop): Likewise.
5525 * optabs-tree.c (optab_for_tree_code): Likewise.
5526 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5527 * tree-inline.c (estimate_operator_cost): Likewise.
5528 * tree-pretty-print.c (dump_generic_node): Likewise.
5529 (op_code_prio): Likewise.
5530 (op_symbol_code): Likewise.
5531 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5532 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5533 * internal-fn.c (direct_internal_fn_optab): New function.
5534 (direct_internal_fn_array, direct_internal_fn_supported_p
5535 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5536 * fold-const-call.c (fold_const_reduction): New function.
5537 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5539 * tree-vect-loop.c: Include internal-fn.h.
5540 (reduction_code_for_scalar_code): Rename to...
5541 (reduction_fn_for_scalar_code): ...this and return an internal
5543 (vect_model_reduction_cost): Take an internal_fn rather than
5545 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5547 (vectorizable_reduction): Use internal functions rather than tree
5548 codes for the reduction operation. Update calls to the functions
5550 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5551 Use calls to internal functions rather than REDUC tree codes.
5552 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5554 2017-11-22 Olivier Hainque <hainque@adacore.com>
5556 * config/vxworks.c (vxworks_override_options): Pick default
5557 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5558 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5559 DWARF_GNAT_ENCODINGS_DEFAULT.
5560 * config/vxworksae.h: Likewise.
5562 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5564 PR tree-optimization/83104
5565 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5568 2017-11-22 Marek Polacek <polacek@redhat.com>
5569 H.J. Lu <hongjiu.lu@intel.com>
5570 Jason Merrill <jason@redhat.com>
5575 * calls.c (initialize_argument_information): Call
5576 warn_parameter_passing_abi target hook.
5577 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5578 argument onto stack.
5579 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5580 * common.opt: Update -fabi-version description.
5581 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5582 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5584 (ix86_is_empty_record): New function.
5585 (ix86_warn_parameter_passing_abi): New function.
5586 (TARGET_EMPTY_RECORD_P): Redefine.
5587 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5588 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5589 * doc/tm.texi: Regenerated.
5590 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5591 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5592 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5593 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5594 instead of int_size_in_bytes.
5595 * expr.c (copy_blkmode_to_reg): Likewise.
5596 * function.c (aggregate_value_p): Return 0 for empty types.
5597 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5599 (locate_and_pad_parm): Call arg size_in_bytes instead
5601 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5602 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5603 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5605 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5606 (std_gimplify_va_arg_expr): Skip empty records. Call
5607 arg_size_in_bytes instead size_in_bytes.
5608 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5609 * tree-core.h (tree_type_common): Add empty_flag.
5610 (tree_decl_common): Update comments.
5611 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5613 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5614 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5616 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5617 * tree.c (default_is_empty_type): New function.
5618 (default_is_empty_record): New function.
5619 (arg_int_size_in_bytes): New function.
5620 (arg_size_in_bytes): New function.
5621 (get_ultimate_context): New function.
5622 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5623 TRANSLATION_UNIT_WARN_EMPTY_P.
5624 (default_is_empty_record, arg_int_size_in_bytes,
5625 arg_size_in_bytes, get_ultimate_context): Declare.
5627 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5629 * config/arm/arm.c (cmse_clear_registers): New function.
5630 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5631 code by call to cmse_clear_registers.
5632 (cmse_nonsecure_entry_clear_before_return): Likewise.
5634 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5636 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5637 (vdot_s32, vdotq_s32): New.
5638 (vdot_lane_u32, vdotq_lane_u32): New.
5639 (vdot_lane_s32, vdotq_lane_s32): New.
5642 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5645 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5646 for unsigned values with fewer HWIs than the precision.
5647 (test_overflow): New function.
5648 (wide_int_cc_tests): Call it.
5650 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5652 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5653 has the same mode class as Pmode.
5655 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5657 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5658 padding_bits_to_clear_ptr.
5659 (cmse_nonsecure_entry_clear_before_return): Likewise.
5661 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5663 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5664 auto_sbitap instead of integer bitfield to control register needing
5667 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5669 PR tree-optimization/83044
5670 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5671 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5672 that need the upper bound. Subtract offset from
5673 get_addr_base_and_unit_offset only if positive and subtract it
5674 before division by eltsize rather than after it.
5677 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5678 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5679 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5683 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5684 expand_binop with *mul_widen_optab, make sure at least one of the
5685 operands doesn't have VOIDmode.
5688 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5690 PR rtl-optimization/82044
5691 PR tree-optimization/82042
5692 * dse.c (record_store): Check for overflow.
5693 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5694 add_wild_read instead of clear_rhs_from_active_local_stores on
5695 overflow and log it into dump_file.
5697 2017-11-22 Richard Biener <rguenther@suse.de>
5699 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5700 fake edges to exit when looking for a place to insert.
5701 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5703 (insert_into_preds_of_block): Commit edge insertion immediately,
5704 assert that doesn't require new BBs.
5705 (fini_pre): Release expressions.
5706 (pass_pre::execute): Shuffle things around a bit, if the fn
5707 is too large do not compute AVAIL either as this is really the
5710 2017-11-22 Richard Biener <rguenther@suse.de>
5712 PR tree-optimization/83089
5713 * tree-if-conv.c (pass_if_conversion::execute): If anything
5714 changed reset SCEV and free the number of iteration estimates.
5716 2017-11-21 Martin Sebor <msebor@redhat.com>
5718 PR tree-optimization/82945
5719 * calls.h (warn_nonstring_bound): Remove unused function.
5721 2017-11-21 Martin Sebor <msebor@redhat.com>
5723 PR tree-optimization/82945
5724 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5725 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5726 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5728 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5729 * calls.h (get_attr_nonstring_decl): Declare new function.
5730 * doc/extend.texi (attribute nonstring): Update.
5731 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5732 get_attr_nonstring_decl and handle it.
5733 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5734 detection of nul-termination.
5735 (strlen_to_stridx): Change to a pointer.
5736 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5737 (pass_strlen::execute): Same.
5739 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5741 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5742 for the new option -mprefer-vector-width=[none|128|256|512].
5743 * config/i386/i386.c (ix86_target_string): remove old style options
5744 -mprefer-avx256 and make -mprefer-avx128 as alias.
5745 (ix86_option_override_internal): Apply defaults for the
5746 -mprefer-vector-width=[128|256] option.
5747 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5748 Implement macros to work with -mprefer-vector-width=.
5749 * config/i386/i386.opt: Implemented option
5750 -mprefer-vector-width=[none|128|256|512].
5751 * doc/invoke.texi: Documentation for
5752 -mprefer-vector-width=[none|128|256|512].
5754 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5756 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5758 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5760 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5762 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5764 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5766 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5767 * expr.c (expand_expr_real_2): Likewise.
5768 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5769 fold_binary_loc): Likewise.
5770 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5771 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5772 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5773 MINUS_EXPR transformations.
5774 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5775 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5776 * tree-inline.c (estimate_operator_cost): Likewise.
5777 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5778 op_symbol_code): Likewise.
5779 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5780 * vr-values.c (extract_range_from_binary_expr): Likewise.
5781 * varasm.c (initializer_constant_valid_p_1): Likewise.
5782 * tree.def: New tree code POINTER_DIFF_EXPR.
5784 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5786 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5787 integer suffix to movbe mnemonic.
5788 (*bswaphi2_movbe): Ditto.
5789 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5791 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5794 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5795 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5796 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5798 2017-11-21 Martin Liska <mliska@suse.cz>
5800 * tree-inline.c (expand_call_inline): Remove not needed
5803 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5804 Jakub Jelinek <jakub@redhat.com>
5807 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5808 Remove static keyword from f variable.
5810 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5812 PR tree-optimization/83086
5813 * gimple-ssa-store-merging.c
5814 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5815 rather than n.base_addr.
5817 2017-11-21 Martin Liska <mliska@suse.cz>
5819 PR rtl-optimization/82044
5820 PR tree-optimization/82042
5821 * dse.c (check_mem_read_rtx): Check for overflow.
5823 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5825 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5828 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5830 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5831 and its corresponding call arg location note.
5833 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5834 Andrew Burgess <andrew.burgess@embecosm.com>
5836 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5838 (arc_return_slot_offset): Likewise.
5839 (arc_eh_return_address_location): New declaration.
5840 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5841 (MUST_SAVE_REGISTER): Add exception handler case.
5842 (MUST_SAVE_RETURN_ADDR): Likewise.
5843 (arc_frame_pointer_required): Likewise.
5844 (arc_frame_pointer_needed): New function.
5845 (arc_compute_frame_size): Changed.
5846 (arc_expand_prologue): Likewise.
5847 (arc_expand_epilogue): Likewise.
5848 (arc_initial_elimination_offset): Likewise.
5849 (arc_return_slot_offset): Delete.
5850 (arc_eh_return_address_location): New function.
5851 (arc_builtin_setjmp_frame_value): Likewise.
5852 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5853 (EH_RETURN_STACKADJ_RTX): Define.
5854 (EH_RETURN_HANDLER_RTX): Likewise.
5855 * config/arc/arc.md (eh_return): Delete.
5857 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5859 * print-tree.h (debug_vec_tree): Remove prototype.
5860 * gdbinit.in (pvt): Remove macro.
5862 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5864 PR tree-optimization/83047
5865 * gimple-ssa-store-merging.c
5866 (imm_store_chain_info::output_merged_store): If the loads with the
5867 same vuse are in different basic blocks, for load_gsi pick a load
5868 location that is dominated by the other loads.
5871 * config/i386/i386.c (ix86_memmodel_check): Start
5872 -Winvalid-memory-model diagnostics with lowercase letter.
5875 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5876 set in_cold_section_p to the partition containing loc_list->first.
5877 When seeing loc_list->last_before_switch node, update secname and
5878 perform range_across_switch second partition handling only after that.
5881 * run-rtl-passes.c: Include debug.h.
5882 (run_rtl_passes): Call debug_hooks->assembly_start.
5883 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5887 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5888 OPTAB_DIRECT in calls to expand_simple_binop.
5890 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5893 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5895 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5898 * spellcheck.h (best_match::blithely_get_best_candidate): New
5901 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5903 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5904 with lower case letter.
5906 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5908 * config/i386/i386.md (bswaphi2): New expander.
5909 (*bswaphi2_movbe): New insn pattern.
5910 (bswaphi -> rorhi pepehole2): New peephole pattern.
5912 2017-11-20 Jeff Law <law@redhat.com>
5914 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5915 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5916 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5917 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5918 gimple-ssa-evrp-analyze.h.
5920 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5921 set BB_VISITED here.
5922 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5924 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5926 * tree-predcom.c: Add general comment on Store-Store chains.
5927 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5929 (get_chain_last_ref_at): Rename into...
5930 (get_chain_last_write_at): ...this.
5931 (get_chain_last_write_before_load): New function.
5932 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5933 CT_STORE_STORE when write reference is added.
5934 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5935 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5936 (initialize_root_vars_store_elim_1): Ditto.
5937 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5938 definition is created.
5939 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5940 chain by replacing it with dominant stored value.
5942 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5944 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5946 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5948 * vr-values.c (extract_range_from_binary_expr): Use a full range
5951 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5953 * config/arm/arm.md (R4_REGNUM): Define constant.
5954 (nonsecure_call_internal): Remove r4 clobber.
5955 (nonsecure_call_value_internal): Likewise.
5956 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5957 clobber and resequence match_operands.
5958 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5959 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5960 (nonsecure_call_value_reg_thumb2): Likewise.
5962 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5964 PR tree-optimization/78821
5965 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5966 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5967 to the address of the base rather than the base itself.
5968 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5969 (find_bswap_or_nop_finalize): New function.
5970 (find_bswap_or_nop): Use it.
5971 (bswap_replace): Return a tree rather than bool, change first
5972 argument from gimple * to gimple_stmt_iterator, allow inserting
5973 into an empty sequence, allow ins_stmt to be NULL - then emit
5974 all stmts into gsi. Fix up MEM_REF address gimplification.
5975 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5976 (struct store_immediate_info): Add N and INS_STMT non-static
5978 (store_immediate_info::store_immediate_info): Initialize them
5979 from newly added ctor args.
5980 (merged_store_group::apply_stores): Formatting fixes. Sort by
5982 (stmts_may_clobber_ref_p): For stores call also
5983 refs_anti_dependent_p.
5984 (gather_bswap_load_refs): New function.
5985 (imm_store_chain_info::try_coalesce_bswap): New method.
5986 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5987 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5988 (imm_store_chain_info::output_merged_store): Fail if number of
5989 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5990 and NOP_EXPR rhs_code.
5991 (pass_store_merging::process_store): Compute n and ins_stmt, if
5992 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5993 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5995 (pass_store_merging::execute): Calculate dominators.
5997 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5998 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5999 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6000 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6001 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6002 class pass_optimize_bswap, bswap_replace,
6003 pass_optimize_bswap::execute): Moved to ...
6004 * gimple-ssa-store-merging.c: ... this file.
6005 Include optabs-tree.h.
6006 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6007 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6008 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6009 anonymous namespace, remove static keywords.
6010 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6011 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6013 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6016 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6018 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6020 * vec.h (debug_helper): New function.
6021 (DEFINE_DEBUG_VEC): New macro.
6022 * hash-set.h (debug_helper): New function.
6023 (DEFINE_DEBUG_HASH_SET): New macro.
6024 * cfg.c (debug_slim (edge)): New function.
6025 Call DEFINE_DEBUG_VEC for edges.
6026 Call DEFINE_DEBUG_HASH_SET for edges.
6027 * cfghooks.c (debug_slim (basic_block)): New function.
6028 Call DEFINE_DEBUG_VEC for basic blocks.
6029 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6030 * print-tree.c (debug_slim): New function to handle trees.
6031 Call DEFINE_DEBUG_VEC for trees.
6032 Call DEFINE_DEBUG_HASH_SET for trees.
6033 (debug (vec<tree, va_gc>) &): Remove.
6034 (debug (<vec<tree, va_gc>) *): Remove.
6035 * print-rtl.c (debug_slim): New function to handle const_rtx.
6036 Call DEFINE_DEBUG_VEC for rtx_def.
6037 Call DEFINE_DEBUG_VEC for rtx_insn.
6038 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6039 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6040 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6041 (debug (vec<rtx_insn *> *ptr): Remove.
6042 (debug_insn_vector): Remove.
6043 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6045 2017-11-20 Tom de Vries <tom@codesourcery.com>
6047 PR rtl-optimization/82020
6048 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6049 IF_THEN_ELSE condition.
6051 2017-11-19 Jeff Law <law@redhat.com>
6053 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6054 of degenerates resulting from ignoring an edge.
6056 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6059 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6061 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6064 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6065 for uninitialied counts.
6067 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6070 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6072 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6075 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6077 2017-11-19 Tom de Vries <tom@codesourcery.com>
6079 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6082 2017-11-19 Tom de Vries <tom@codesourcery.com>
6084 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6087 2017-11-19 Tom de Vries <tom@codesourcery.com>
6089 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6092 2017-11-19 Tom de Vries <tom@codesourcery.com>
6094 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6096 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6098 2017-11-19 Tom de Vries <tom@codesourcery.com>
6100 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6102 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6103 semicolon after MCORE_EXPORT_NAME call.
6105 2017-11-19 Tom de Vries <tom@codesourcery.com>
6108 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6109 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6111 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6113 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6114 (cgraph_edge::clone): Cleanup updating of profile.
6115 * ipa-cp.c (update_profiling_info): Likewise.
6116 * ipa-inline-transform.c (inline_transform): Likewise.
6117 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6118 * ipa-split.c (execute_split_functions): Do not split when function
6120 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6121 * profile-count.c (profile_count::dump): Add global0.
6122 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6124 (profile_count::to_sreal_scale): Likewise.
6125 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6126 (profile_count::combine_with_ipa_count): New function.
6127 * profile-count.h (profile_guessed_global0adjusted): New.
6128 (profile_count::adjusted_zero): New.
6129 (profile_count::global0adjusted): New.
6130 (profile_count::combine_with_ipa_count): New.
6131 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6132 correct profile of return block of split functions.
6133 (copy_cfg_body): Remove unused profile_count.
6134 (copy_body): Likewise.
6135 (expand_call_inline): Update.
6136 (tree_function_versioning): Update.
6138 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6140 * hash-set.h (hash_set::empty): New.
6141 * tree-ssa-threadbackward.h: Delete.
6142 * tree-ssa-threadbackward.c (class thread_jumps): New.
6143 Move max_threaded_paths into class.
6144 (fsm_find_thread_path): Remove arguments that are now in class.
6145 (profitable_jump_thread_path): Rename to...
6146 (thread_jumps::profitable_jump_thread_path): ...this.
6147 (convert_and_register_jump_thread_path): Rename to...
6148 (thread_jumps::convert_and_register_current_path): ...this.
6149 (check_subpath_and_update_thread_path): Rename to...
6150 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6151 (register_jump_thread_path_if_profitable): Rename to...
6152 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6153 (handle_phi): Rename to...
6154 (thread_jumps::handle_phi): ...this.
6155 (handle_assignment): Rename to...
6156 (thread_jumps::handle_assignment): ...this.
6157 (fsm_find_control_statement_thread_paths): Rename to...
6158 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6159 (find_jump_threads_backwards): Rename to...
6160 (thread_jumps::find_jump_threads_backwards): ...this.
6161 Initialize path local data.
6162 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6163 from within thread_jumps class.
6164 (pass_early_thread_jumps::execute): Same.
6166 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6168 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6170 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6172 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6173 to canonical location.
6175 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6178 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6180 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6182 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6184 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6185 rcrt1.o%s/grcrt1.o%s for -static-pie.
6187 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6189 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6190 ix86_shift_rotate_cost): Break out from ...
6191 (ix86_rtx_costs): ... here.
6192 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6195 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6197 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6198 when entry block was promoted unlikely.
6199 (estimate_bb_frequencies): Increase frequency scale.
6200 * profile-count.h (profile_count): Export precision info.
6202 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6204 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6205 disturbin profile of entry block.
6207 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6209 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6212 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6214 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6215 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6217 2017-11-17 Jeff Law <law@redhat.com>
6219 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6220 from evrp_dom_walker class. Various methods moved into new class.
6221 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6222 (evrp_range_analyzer::enter): New method.
6223 (evrp_range_analyzer::leave): New method.
6224 (evrp_dom_walker): Remove delegators no longer needed by this class.
6225 Replace vr_values data member with evrp_range_analyzer
6227 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6228 method extracted from evrp_dom_walker::before_dom_children.
6229 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6230 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6232 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6233 Add private copy constructor and move assignment operators.
6234 Privatize methods and class data where trivially possible.
6235 (evrp_dom_walker::cleanup): New function, extracted from
6236 execute_early_vrp. Simplify access to class data.
6238 * vr-values.h (get_output_for_vrp): Prototype.
6239 * vr-values.c (get_output_for_vrp): New function extracted from
6240 vrp_visit_assignment_or_call and extract_range_from_stmt.
6241 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6243 2017-11-17 Luis Machado <luis.machado@linaro.org>
6245 * config/aarch64/aarch64.c
6246 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6247 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6248 tune_params::AUTOPREFETCHER_WEAK.
6250 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6253 * config/arm/arm.c (arm_valid_target_attribute_rec):
6254 Parse "arch=" and "+<ext>".
6255 (arm_valid_target_attribute_tree): Re-init global options.
6256 (arm_option_override): Make non-static.
6257 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6258 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6259 (__ARM_FEATURE_CRC32): Support undef.
6260 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6261 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6263 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6265 * gdbinit.in (break-on-diagnostic): New command.
6267 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6269 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6270 return to indirect jump for EH return if control-flow
6271 protection is enabled. Change explicit 'false' argument in
6272 pro_epilogue_adjust_stack with a value of
6274 * config/i386/i386.md (simple_return_indirect_internal):
6275 Remove SImode restriction to support 64-bit.
6277 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6279 * combine.c (added_notes_insn): New.
6280 (try_combine): Handle added_notes_insn like added_links_insn.
6281 Rewrite return value code.
6282 (distribute_notes): Set added_notes_insn to the earliest insn we added
6285 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6287 PR rtl-optimization/82621
6288 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6289 dest of one of those SETs is unused.
6291 2017-11-17 Richard Biener <rguenther@suse.de>
6294 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6295 * tree-pretty-print.c (dump_generic_node): Handle
6296 annot_expr_parallel_kind.
6297 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6298 * gimplify.c (gimple_boolify): Likewise.
6300 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6302 * config.gcc (extra_headers): Add cet.h for x86 targets.
6303 * config/i386/cet.h: New file.
6304 * doc/install.texi: Add --enable-cet/--disable-cet.
6306 2017-11-17 Richard Biener <rguenther@suse.de>
6308 PR tree-optimization/83017
6309 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6310 (gen_parallel_loop): Properly count iterations.
6311 (parallelize_loops): Handle loop->can_be_parallel independent
6312 of flag_loop_parallelize_all. Make static profitability test match
6314 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6315 * invoke.texi (parloops-min-per-thread): Document.
6317 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6319 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6320 upstreaming review comments.
6322 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6324 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6325 with fast unaligned access.
6326 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6328 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6330 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6331 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6333 2017-11-17 Richard Biener <rguenther@suse.de>
6335 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6336 folding of references.
6338 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6341 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6342 of replacing call to strncmp with corresponding call to strcmp when
6345 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6347 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6348 option prefer-avx256 for skylake-avx512 configuration.
6349 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6350 (get_builtin_code_for_version): Ditto.
6352 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6353 Monk Chiang <sh.chiang04@gmail.com>
6355 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6356 (FIXED_REGISTERS): Reserve more register numbers.
6357 (CALL_USED_REGISTERS): Likewise.
6358 (REG_ALLOC_ORDER): Likewise.
6359 (REG_CLASS_CONTENTS): Likewise.
6360 (REGISTER_NAMES): Likewise.
6362 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6363 Kito Cheng <kito.cheng@gmail.com>
6365 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6367 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6369 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6371 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6372 parameter list for vec_splats.
6374 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6376 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6378 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6380 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6383 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6384 with the required version of automake 1.11.6.
6386 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6388 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6390 (DU_C2_3_power9): Correct reservation combinations.
6391 (FP_DIV_power9, VEC_DIV_power9): New.
6392 (power9-alu): Split out rotate/shift...
6393 (power9-rot): ...to here, correct dispatch resource.
6394 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6396 (power9-fp): Correct latency.
6397 (power9-sdiv): Add div/sqrt resource.
6398 (power9-ddiv): Correct latency, add div/sqrt resource.
6399 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6400 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6402 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6404 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6406 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6407 switch statement mapping KF built-ins to TF built-ins if we don't
6408 have the proper ISA 3.0 assembler support.
6410 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6412 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6413 (lower_emutls_function_body): Do not compute it.
6415 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6417 * ipa-split.c (split_bb_info): Turn time to sreal.
6418 (split_point): Likewise.
6419 (dump_split_point): Likewise.
6420 (fine_split_points): Likewise.
6421 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6423 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6425 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6428 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6430 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6432 (expensive_function_p): Remove useless assert.
6433 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6435 2017-11-16 Martin Sebor <msebor@redhat.com>
6437 PR tree-optimization/82588
6438 PR tree-optimization/82583
6439 * tree-vrp.c (check_array_ref): Handle flexible array members,
6440 string literals, and inner indices.
6441 (search_for_addr_array): Add detail to diagnostics.
6443 2017-11-16 Nathan Sidwell <nathan@acm.org>
6447 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6448 (SET_DECL_ASSEMBLER_NAME): Forward to
6449 overwrite_decl_assembler_name.
6450 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6451 (overwrite_decl_assembler_name): Declare.
6452 * tree.c (overwrite_decl_assembler_name): New.
6453 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6454 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6455 (LANG_HOOKS_INITIALIZER): Add it.
6456 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6457 * langhooks.c (lhd_set_decl_assembler_name): Use
6458 SET_DECL_ASSEMBLER_NAME.
6459 (lhd_overwrite_decl_assembler_name): Default implementation.
6461 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6462 Jackson Woodruff <jackson.woodruff@arm.com>
6464 PR tree-optimization/71026
6465 * match.pd: Canonicalize constant multiplies in division.
6467 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6469 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6470 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6471 Remove OPT_fomit_frame_pointer entry.
6472 * common/config/arc/arc-common.c: Likewise.
6473 * common/config/arm/arm-common.c: Likewise.
6474 * common/config/avr/avr-common.c: Likewise.
6475 * common/config/c6x/c6x-common.c: Likewise.
6476 * common/config/cr16/cr16-common.c: Likewise.
6477 * common/config/cris/cris-common.c: Likewise.
6478 * common/config/epiphany/epiphany-common.c: Likewise.
6479 * common/config/fr30/fr30-common.c: Likewise.
6480 * common/config/frv/frv-common.c: Likewise.
6481 * common/config/ia64/ia64-common.c: Likewise.
6482 * common/config/iq2000/iq2000-common.c: Likewise.
6483 * common/config/lm32/lm32-common.c: Likewise.
6484 * common/config/m32r/m32r-common.c: Likewise.
6485 * common/config/mcore/mcore-common.c: Likewise.
6486 * common/config/microblaze/microblaze-common.c: Likewise.
6487 * common/config/mips/mips-common.c: Likewise.
6488 * common/config/mmix/mmix-common.c: Likewise.
6489 * common/config/mn10300/mn10300-common.c: Likewise.
6490 * common/config/nios2/nios2-common.c: Likewise.
6491 * common/config/pa/pa-common.c: Likewise.
6492 * common/config/pdp11/pdp11-common.c: Likewise.
6493 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6494 * common/config/riscv/riscv-common.c: Likewise.
6495 * common/config/rs6000/rs6000-common.c: Likewise.
6496 * common/config/rx/rx-common.c: Likewise.
6497 * common/config/s390/s390-common.c: Likewise.
6498 * common/config/sh/sh-common.c: Likewise.
6499 * common/config/sparc/sparc-common.c: Likewise.
6500 * common/config/tilegx/tilegx-common.c: Likewise.
6501 * common/config/tilepro/tilepro-common.c: Likewise.
6502 * common/config/v850/v850-common.c: Likewise.
6503 * common/config/visium/visium-common.c: Likewise.
6504 * common/config/xstormy16/xstormy16-common.c: Likewise.
6505 * common/config/xtensa/xtensa-common.c: Likewise.
6506 * invoke.texi (-fomit-frame-pointer): Update documentation.
6508 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6510 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6512 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6514 * cfg.c (scale_bbs_frequencies_int,
6515 cale_bbs_frequencies_gcov_type): Remove.
6516 * cfg.h (scale_bbs_frequencies_int,
6517 cale_bbs_frequencies_gcov_type): Remove.
6519 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6521 * tree-ssa-loop-manip.c
6522 (scale_dominated_blocks_in_loop): Update to profile counts.
6523 (tree_transform_and_unroll_loop): Likewise.
6525 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6527 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6528 scale_bbs_frequencies_int.
6530 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6532 * final.c (compute_alignments): Use counts rather than frequencies.
6534 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6536 * cfgloopanal.c: Include sreal.h
6537 (average_num_loop_insns): Use counts and sreal for accounting.
6539 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6541 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6544 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6545 Kito Cheng <kito.cheng@gmail.com>
6547 * config/nds32/constraints.md: Provide more constraints.
6548 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6549 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6550 support constraints usage.
6552 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6554 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6556 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6557 Kito Cheng <kito.cheng@gmail.com>
6559 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6560 * config/nds32/nds32.opt: Refine the layout.
6561 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6562 TARGET_EXT_STRING): Support new options.
6563 * config/nds32/nds32.h: Likewise.
6564 * config/nds32/nds32.md: Likewise.
6565 * config/nds32/nds32-predicates.c: Likewise.
6566 * config/nds32/constraints.md: Likewise.
6567 * common/config/nds32/nds32-common.c: Likewise.
6569 2017-11-16 Julia Koval <julia.koval@intel.com>
6572 * config/i386/gfniintrin.h: Add sse check.
6573 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6575 2017-11-16 Julia Koval <julia.koval@intel.com>
6577 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6578 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6579 (ix86_handle_option): Handle -mavx512vbmi2.
6580 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6581 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6582 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6583 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6584 (ix86_valid_target_attribute_inner_p): Ditto.
6585 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6586 * config/i386/i386.opt (mavx512vbmi2): New option.
6587 * doc/invoke.texi: Add new option.
6589 2017-11-16 Julia Koval <julia.koval@intel.com>
6591 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6592 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6593 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6594 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6595 _mm512_gf2p8mul_epi8): New intrinsics.
6596 * config/i386/i386-builtin-types.def
6597 (V64QI_FTYPE_V64QI_V64QI): New type.
6598 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6599 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6600 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6601 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6602 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6603 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6605 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6607 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6608 explicitly as a stream of bytes.
6610 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6612 * config/rs6000/altivec.h (vec_xst_be): New #define.
6613 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6614 and externalize from *altivec_vperm_<mode>_internal.
6615 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6617 (XL_BE_V8HI): Likewise.
6618 (XL_BE_V4SI): Likewise.
6619 (XL_BE_V4SI): Likewise.
6620 (XL_BE_V2DI): Likewise.
6621 (XL_BE_V4SF): Likewise.
6622 (XL_BE_V2DF): Likewise.
6623 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6624 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6625 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6626 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6627 VSX_BUILTIN_VEC_XST_BE.
6628 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6629 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6631 (altivec_init_builtins): Replace conditional calls to def_builtin
6632 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6633 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6634 with unconditional calls. Remove calls to def_builtin for
6635 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6636 __builtin_vec_xst_be.
6637 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6638 to define_expand, and add alternate RTL generation for P8.
6639 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6640 vsx_ld_elemrev_v8hi.
6641 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6642 add alternate RTL generation for P8.
6643 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6644 vsx_ld_elemrev_v16qi.
6645 (vsx_st_elemrev_v8hi): Convert define_insn
6646 to define_expand, and add alternate RTL generation for P8.
6647 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6648 vsx_st_elemrev_v8hi.
6649 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6650 add alternate RTL generation for P8.
6651 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6652 vsx_st_elemrev_v16qi.
6654 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6657 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6658 TARGET_AVX512ER check.
6659 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6660 neither -mzeroupper nor -mno-zeroupper is used and
6661 TARGET_EMIT_VZEROUPPER is set.
6662 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6663 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6665 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6667 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6668 folding of vector compares.
6669 (fold_build_vec_cmp): New helper function.
6670 (fold_compare_helper): New helper function.
6671 (builtin_function_type): Add compare builtins to the list of functions
6672 having unsigned arguments. Cosmetic updates to comment indentation.
6673 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6674 the not+eq combination.
6676 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6678 PR tree-optimization/82726
6679 PR tree-optimization/70754
6680 * tree-predcom.c (order_drefs_by_pos): New function.
6681 (combine_chains): Move code setting has_max_use_after to...
6682 (try_combine_chains): ...here. New parameter. Sort combined chains
6683 according to position information.
6684 (tree_predictive_commoning_loop): Update call to above function.
6685 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6687 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6689 PR tree-optimization/82726
6691 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6693 PR tree-optimization/70754
6694 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6695 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6696 combined stmt before it if not NULL.
6697 (combine_chains): Process refs reversely and compute dominance point
6701 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6703 PR tree-optimization/79663
6704 * tree-predcom.c (combine_chains): Process refs in reverse order
6705 only for ZERO length chains, and add explaining comment.
6707 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6709 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6710 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6711 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6713 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6715 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6717 2017-11-15 Martin Liska <mliska@suse.cz>
6719 * tree-cfg.c (pass_warn_function_return::execute):
6720 Compare warn_return_type for greater than zero.
6722 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6726 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6727 to return true on Xeon and not on Xeon Phi.
6728 (ix86_check_avx256_register): Changed to ...
6729 (ix86_check_avx_upper_register): ... this. Add extra check for
6730 VALID_AVX512F_REG_OR_XI_MODE.
6731 (ix86_avx_u128_mode_needed): Changed
6732 ix86_check_avx256_register to ix86_check_avx_upper_register.
6733 (ix86_check_avx256_stores): Changed to ...
6734 (ix86_check_avx_upper_stores): ... this. Changed
6735 ix86_check_avx256_register to ix86_check_avx_upper_register.
6736 (ix86_avx_u128_mode_after): Changed
6737 avx_reg256_found to avx_upper_reg_found. Changed
6738 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6739 (ix86_avx_u128_mode_entry): Changed
6740 ix86_check_avx256_register to ix86_check_avx_upper_register.
6741 (ix86_avx_u128_mode_exit): Ditto.
6742 * config/i386/i386.h: (host_detect_local_cpu): New define.
6744 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6746 * config/arm/xgene1.md (xgene1): Split into automatons
6747 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6748 (xgene1_f_load): Adjust reservations and/or types.
6749 (xgene1_f_store): Likewise.
6750 (xgene1_load_pair): Likewise.
6751 (xgene1_store_pair): Likewise.
6752 (xgene1_fp_load1): Likewise.
6753 (xgene1_load1): Likewise.
6754 (xgene1_store1): Likewise.
6755 (xgene1_move): Likewise.
6756 (xgene1_alu): Likewise.
6757 (xgene1_simd): Likewise.
6758 (xgene1_bfm): Likewise.
6759 (xgene1_neon_load1): Likewise.
6760 (xgene1_neon_store1): Likewise.
6761 (xgene1_neon_logic): Likewise.
6762 (xgene1_neon_st1): Likewise.
6763 (xgene1_neon_ld1r): Likewise.
6764 (xgene1_alu_cond): Added.
6765 (xgene1_shift_reg): Likwise.
6766 (xgene1_bfx): Likewise.
6767 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6769 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6772 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6774 (can_widen_mult_without_libcall): New function.
6775 (expand_mul_overflow): If only checking unsigned mul overflow,
6776 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6777 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6778 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6779 (expand_DIVMOD): Formatting fix.
6780 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6781 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6782 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6784 PR tree-optimization/82977
6785 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6786 constructed temporary to strlen_to_stridx.put.
6788 2017-11-15 Martin Liska <mliska@suse.cz>
6790 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6791 * configure: Regenerate.
6793 2017-11-15 Martin Liska <mliska@suse.cz>
6796 * config/sh/sh-mem.cc: Use proper probability for
6799 2017-11-14 Jeff Law <law@redhat.com>
6801 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6802 the red zone for stack_clash_protection_final_dynamic_probe targets
6803 when the total dynamic stack size is zero bytes.
6805 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6806 blocks is post order.
6808 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6810 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6811 * final.c (rest_of_clean_state): Set it for the
6812 -fcompare-debug dump.
6813 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6814 class when TDF_COMPARE_DEBUG is set.
6816 * dwarf2out.c (gen_producer_string): Discard
6819 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6822 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6823 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6824 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6825 (__TGMATH_REAL_2_3): Remove macros.
6826 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6827 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6829 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6831 (remquo): Define using __TGMATH_REAL_3.
6833 2017-11-14 Jeff Law <law@redhat.com>
6835 * vr-values.c: New file with contents extracted from tree-vrp.c.
6836 * Makefile.in (OBJS): Add vr-values.o
6837 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6838 (set_value_range, set_and_canonicalize_value_range): Likewise.
6839 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6840 (value_range_constant_singleton, symbolic_range_p): Likewise.
6841 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6842 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6843 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6844 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6845 (find_case_label_range, find_case_label_index): Likewise.
6846 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6847 (range_int_cst_singleton_p, value_inside_range): Likewise.
6848 (get_single_symbol): Likewise.
6849 (switch_update): Move structure definition here.
6850 (to_remove_edges, to_update_switch_stmts): Provide externs.
6851 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6852 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6853 (vrp_val_is_min, set_value_range): Likewise.
6854 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6855 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6856 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6857 (range_is_nonnull, range_int_cst_p): Likewwise.
6858 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6859 (get_single_symbol, operand_less_p): Likewise
6860 (compare_values_warnv, compare_values): Likewise.
6861 (value_inside_range, value_range_constant_singleton): Likewise.
6862 (zero_nonzero_bitgs_from_vr): Likewise.
6863 (extract_range_from_binary_expr_1): Likewise.
6864 (overflow_comparison_p): Likewise.
6865 (to_remove_edges, to_update_switch_stmts): Likewise.
6866 (find_case_label-index, find_case_label_range): Likewise.
6867 (switch_update, set_value_range_to_nonnegative): Remove.
6868 (set_value_range_to_truthvalue): Likewise.
6869 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6870 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6871 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6872 (find_case_label_ranges, test_for_singularity): Likewise.
6873 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6874 (x_vr_values): Move to its remaining use site.
6876 2017-11-10 Jeff Law <law@redhat.com>
6878 * vr-values.h (VR_INITIALIZER): Move #define here.
6879 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6880 * Makefile.in (OBJS): Add tree-evrp.o
6881 * tree-vrp.h (assert_info): Move structure definition here.
6882 (set_value_range_to_varying): Prototype.
6883 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6884 (infer_value_range, register_edge_assert_for): Likewise.
6885 (stmt_interesting_for_vrp): Likewise.
6886 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6887 (set_value_range_to_varying): No longer static.
6888 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6889 (infer_value_range, register_edge_assert_for): Likewise.
6891 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6893 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6894 generate the XXBRD instruction.
6896 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6897 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6898 as being compatible if -mabi=ieeelongdouble.
6899 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6900 to setup float128 built-ins with hardware support.
6901 (BU_FLOAT128_HW_2): Likewise.
6902 (BU_FLOAT128_HW_3): Likewise.
6903 (BU_FLOAT128_HW_VSX_1): Likewise.
6904 (BU_FLOAT128_HW_VSX_2): Likewise.
6905 (scalar_extract_expq): Change float128 built-in functions to
6906 accommodate having both KFmode and TFmode functions. Use the
6907 KFmode variant as the default.
6908 (scalar_extract_sigq): Likewise.
6909 (scalar_test_neg_qp): Likewise.
6910 (scalar_insert_exp_q): Likewise.
6911 (scalar_insert_exp_qp): Likewise.
6912 (scalar_test_data_class_qp): Likewise.
6913 (sqrtf128_round_to_odd): Delete processing the round to odd
6914 built-in functions as special built-in functions, and define them
6915 as float128 built-ins. Use the KFmode variant as the default.
6916 (truncf128_round_to_odd): Likewise.
6917 (addf128_round_to_odd): Likewise.
6918 (subf128_round_to_odd): Likewise.
6919 (mulf128_round_to_odd): Likewise.
6920 (divf128_round_to_odd): Likewise.
6921 (fmaf128_round_to_odd): Likewise.
6922 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6923 support for KFmode and TFmode xststdcqp calls.
6924 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6925 point, switch the built-in handlers for the get/set float128
6926 exponent, get float128 mantissa, float128 test built-ins, and the
6927 float128 round to odd built-in functions. Eliminate creating the
6928 float128 round to odd built-in functions as special built-ins.
6929 (rs6000_init_builtins): Eliminate special creation of the float128
6930 round to odd built-in functions.
6931 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6932 function insns to support both TFmode and KFmode variants.
6933 (xsxsigqp_<mode>): Likewise.
6934 (xsiexpqpf_<mode>): Likewise.
6935 (xsiexpqp_<mode>): Likewise.
6936 (xststdcqp_<mode>): Likewise.
6937 (xststdcnegqp_<mode>): Likewise.
6938 (xststdcqp_<mode>): Likewise.
6940 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6942 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6943 unused path_in_freq_ptr parameter.
6944 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6946 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6948 * ipa-inline.c (edge_badness): Dump sreal frequency.
6949 (compute_inlined_call_time): Match natural implementaiton ...
6950 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6951 forgotten division by CGRAPH_FREQ_BASE.
6953 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6955 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6956 Solaris 11. Update comment.
6957 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6959 * config/sol2.h (STARTFILE_SPEC): Likewise.
6960 * configure: Regenerate.
6962 2017-11-14 Carl Love <cel@us.ibm.com>
6964 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6965 le_ and be_ prefixes to swap* variables. Remove
6966 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6969 2017-11-14 Jason Merrill <jason@redhat.com>
6971 Support GTY((cache)) on hash_map.
6972 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6973 (ggc_cache_remove): Override it instead of ggc_mx.
6974 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6975 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6976 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6977 (simple_cache_map_traits): Override maybe_mx.
6978 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6979 (hash_map): Friend gt_cleare_cache.
6980 (gt_cleare_cache): New.
6981 * tree.h (tree_cache_traits): New hash_map traits class.
6982 (tree_cache_map): New typedef.
6984 2017-11-14 Richard Biener <rguenther@suse.de>
6986 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6987 paramter and handling.
6988 (cleanup_control_flow_bb): Likewise.
6989 (cleanup_control_flow_pre): New helper performing a DFS walk
6990 to call cleanup_control_flow_bb in PRE order.
6991 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6992 via cleanup_control_flow_pre.
6994 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6996 * config/aarch64/aarch64-simd.md
6997 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6998 (*aarch64_simd_bsl<mode>_alt): Likewise.
6999 (aarch64_simd_bsldi_internal): New.
7000 (aarch64_simd_bsldi_alt): Likewise.
7002 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7004 * tracer.c (better_p): Do not compare frequencies.
7005 * reg-stack.c (better_edge): Likewise.
7006 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7009 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7011 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7012 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7013 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7014 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7015 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7016 * predict.c (maybe_hot_frequency_p): Inline to ...
7017 (maybe_hot_count_p): ... here; rewrite to counts.
7018 (counts_to_freqs): Rename to ...
7019 (update_max_bb_count): ... this one.
7020 (expensive_function_p): Use counts.
7021 (estimate_bb_frequencies): Update.
7022 (rebuild_frequencies): Update.
7023 * predict.h (counts_to_freqs): Rename to ...
7024 (update_max_bb_count): ... this one.
7025 * profile.c (compute_branch_probabilities): Add debug info
7026 * tree-inline.c (expand_call_inline): Update debug info.
7027 (optimize_inline_calls): Use update_max_bb_count..
7028 (tree_function_versioning): Use update_max_bb_count..
7029 * value-prof.c (gimple_value_profile_transformations):
7030 Do not use update_max_bb_count.
7032 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7034 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7035 always use frequencies.
7037 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7039 * bb-reorder.c: Remove frequencies from comments.
7040 (better_edge_p): Use profile counts.
7041 (find_traces): Dump profile counts.
7042 (rotate_loop): Use profile counts.
7043 (find_traces_1_round): Likewise.
7044 (connect_better_edge_p): Use counts instead of probabilities for
7046 (copy_bb_p): Drop early check for non-0 frequency.
7047 (sanitize_hot_paths): Update comments.
7049 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7051 * ipa-split.c (struct split_point): Add count.
7052 (consider_split): Do not compute incoming frequency; compute incoming
7053 count and store it to split_point.
7054 (split_function): Set count of the call to split part correctly.
7056 2017-11-13 Carl Love <cel@us.ibm.com>
7058 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7060 2017-11-13 Tom Tromey <tom@tromey.com>
7062 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7064 2017-11-13 Carl Love <cel@us.ibm.com>
7066 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7067 Add support for builtins:
7068 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7069 vector {un,}signed {char,int,short},
7070 vector {un,}signed {char,int,short}) arguments.
7071 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7072 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7073 Add BU_P9V_AV_2 expansions for the builtins.
7074 * config/rs6000/altivec.h (vec_first_match_index,
7075 vec_first_mismatch_index, vec_first_match_or_eos_index,
7076 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7077 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7078 new extern declaration.
7079 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7080 * config/rs6000/vsx.md (first_match_index_<mode>,
7081 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7082 first_mismatch_or_eos_index_<mode>): Add define expand.
7083 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7084 * doc/extend.texi: Update the built-in documenation file for the new
7087 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7089 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7090 into the min/max operations for _Float<N> and _Float<N>X types.
7092 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7095 * dwarf2out.c (do_eh_frame): New static variable.
7096 (dwarf2out_begin_prologue): Set it.
7097 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7099 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7101 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7103 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7105 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7108 * debug.h (dwarf2out_do_eh_frame): Declare.
7109 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7110 (dwarf2out_do_frame): Use it.
7111 (dwarf2out_do_cfi_asm): Likewise.
7112 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7113 (dwarf2out_assembly_start): Likewise.
7114 (dwarf2out_begin_prologue): Fix comment.
7115 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7116 if the target needs either debug or unwind DWARF2 info.
7117 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7118 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7119 -ftrapping-math, -ftrapv and -fwrapv.
7121 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7123 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7124 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7125 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7126 (estimate_edge_size_and_time): Likewise.
7127 (ipa_merge_fn_summary_after_inlining): Likewise.
7128 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7129 (compute_uninlined_call_time): Use sreal_frequency.
7130 (compute_inlined_call_time): Likewise.
7131 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7132 * profile-count.c: Include sreal.h.
7133 (profile_count::to_sreal_scale): New.
7134 * profile-count.h: Forward declare sreal.
7135 (profile_count::to_sreal_scale): Declare.
7137 2017-11-13 Nathan Sidwell <nathan@acm.org>
7139 * diagnostic.c (maybe_line_and_column): New.
7140 (diagnostic_get_location_text): Use it.
7141 (diagnostic_report_current_module): Likewise.
7142 (test_diagnostic_get_location_text): Add tests.
7144 2017-11-13 Luis Machado <luis.machado@linaro.org>
7146 * doc/md.texi (Specifying processor pipeline description): Fix
7147 incorrect latency for the div instruction example.
7149 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7151 PR tree-optimization/78821
7152 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7153 that bit_not_p is the same.
7154 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7155 (split_group): Count precisely bit_not_p bits in each statement.
7156 (invert_op): New function.
7157 (imm_store_chain_info::output_merged_store): Use invert_op to
7158 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7159 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7161 2017-11-13 Martin Liska <mliska@suse.cz>
7163 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7164 (struct source_info): Likewise.
7165 (add_branch_counts): Likewise.
7166 (add_line_counts): Likewise.
7167 (function_summary): Likewise.
7168 (output_intermediate_line): Likewise.
7169 (generate_results): Likewise.
7171 2017-11-13 Martin Liska <mliska@suse.cz>
7173 * gcov.c (struct block_info): Remove typedef for block_t.
7174 (struct line_info): Likewise.
7175 (line_info::has_block): Likewise.
7176 (EXIT_BLOCK): Likewise.
7177 (unblock): Likewise.
7178 (circuit): Likewise.
7179 (get_cycles_count): Likewise.
7180 (process_file): Likewise.
7181 (read_graph_file): Likewise.
7182 (solve_flow_graph): Likewise.
7183 (find_exception_blocks): Likewise.
7184 (add_line_counts): Likewise.
7185 (accumulate_line_info): Likewise.
7186 (output_line_details): Likewise.
7188 2017-11-13 Martin Liska <mliska@suse.cz>
7190 * gcov.c (struct arc_info): Remove typedef for arc_t.
7191 (struct line_info): Likewise.
7192 (add_branch_counts): Likewise.
7193 (output_branch_count): Likewise.
7194 (function_info::~function_info): Likewise.
7195 (circuit): Likewise.
7196 (output_intermediate_line): Likewise.
7197 (read_graph_file): Likewise.
7198 (solve_flow_graph): Likewise.
7199 (find_exception_blocks): Likewise.
7200 (add_line_counts): Likewise.
7201 (accumulate_line_info): Likewise.
7202 (output_line_details): Likewise.
7203 (output_function_details): Likewise.
7205 2017-11-13 Martin Liska <mliska@suse.cz>
7207 * gcov.c (struct function_info): Remove typedef for function_t.
7208 (struct source_info): Likewise.
7209 (source_info::get_functions_at_location): Likewise.
7210 (solve_flow_graph): Likewise.
7211 (find_exception_blocks): Likewise.
7212 (add_line_counts): Likewise.
7213 (output_intermediate_file): Likewise.
7214 (process_file): Likewise.
7215 (generate_results): Likewise.
7216 (release_structures): Likewise.
7217 (read_graph_file): Likewise.
7218 (read_count_file): Likewise.
7219 (accumulate_line_counts): Likewise.
7220 (output_lines): Likewise.
7222 2017-11-13 Martin Liska <mliska@suse.cz>
7224 * gcov.c (function_info::function_info): Remove num_counts
7225 and add vector<gcov_type>.
7226 (function_info::~function_info): Use the vector.
7227 (process_file): Likewise.
7228 (read_graph_file): Likewise.
7229 (read_count_file): Likewise.
7230 (solve_flow_graph): Likewise.
7232 2017-11-13 Martin Liska <mliska@suse.cz>
7234 * gcov.c (function_info::is_artificial): New function.
7235 (process_file): Erase all artificial early.
7236 (generate_results): Skip as all artificial are already
7239 2017-11-13 Martin Liska <mliska@suse.cz>
7241 * gcov.c (read_graph_file): Store to global vector of functions.
7242 (read_count_file): Iterate the vector.
7243 (process_file): Likewise.
7244 (generate_results): Likewise.
7245 (release_structures): Likewise.
7247 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7249 PR tree-optimization/82954
7250 * gimple-ssa-store-merging.c
7251 (imm_store_chain_info::coalesce_immediate_stores): If
7252 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7254 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7256 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7257 Upddate call to ENDIAN_LANE_N.
7258 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7259 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7260 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7261 and use aarch64_endian_lane_rtx.
7262 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7264 2017-11-12 Tom de Vries <tom@codesourcery.com>
7266 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7268 2017-11-12 Tom de Vries <tom@codesourcery.com>
7270 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7272 2017-11-12 Tom de Vries <tom@codesourcery.com>
7274 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7276 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7277 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7278 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7280 2017-11-11 Martin Sebor <msebor@redhat.com>
7283 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7286 * prefic.c (translate_name): Replace strncpy with memcpy to
7287 avoid -Wstringop-truncation.
7289 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7291 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7293 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7295 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7297 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7299 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7302 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7304 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7305 enable generating XXBRH if the value is in a vector register.
7306 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7307 value is in a vector register.
7308 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7309 register to register bswap64's instead of doing the GPR sequence
7310 used on previous machines.
7311 (bswapdi2_xxbrd): New insn.
7312 (bswapdi2_reg): Disallow on ISA 3.0.
7313 (register to register bswap64 splitter): Do not split the insn on
7314 ISA 3.0 systems that use XXBRD.
7316 2017-11-10 Martin Sebor <msebor@redhat.com>
7319 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7321 (find_subframework_file): Same.
7323 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7325 * auto-profile.c (afdo_indirect_call): Drop frequency.
7326 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7327 (cgraph_node::create_edge): Drop frequency argument.
7328 (cgraph_node::create_indirect_edge): Drop frequency argument.
7329 (cgraph_edge::make_speculative): Drop frequency arguments.
7330 (cgraph_edge::resolve_speculation): Do not update frequencies
7331 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7332 (cgraph_node::dump): Check consistency in IPA mode.
7333 (cgraph_edge::maybe_hot_p): Use IPA counter.
7334 (cgraph_edge::verify_count_and_frequency): Rename to ...
7335 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7336 (cgraph_node::verify_node): Update.
7337 * cgraph.h (struct cgraph_edge): Drop frequency.
7338 (cgraph_edge::frequency): New function.
7339 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7341 (cgraph_edge::rebuild_edges): Likewise.
7342 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7343 (duplicate_thunk_for_node): Do not pass frequency.
7344 (cgraph_node::create_clone): Scale only counts.
7345 (cgraph_node::create_virtual_clone): Do not pass frequency.
7346 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7347 (cgraph_node::create_version_clone): Do not pass frequency.
7348 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7349 (cgraph_node::expand_thunk): Do not pass frequency.
7350 (cgraph_node::create_wrapper): Do not pass frequency.
7351 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7353 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7354 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7355 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7356 (gather_caller_stats): Use frequency function.
7357 (ipcp_cloning_candidate_p): Use frequency function.
7358 (ipcp_propagate_stage): Use frequency function.
7359 (get_info_about_necessary_edges): Use frequency function.
7360 (update_profiling_info): Update only IPA profile.
7361 (update_specialized_profile): Use frequency functoin.
7362 (perhaps_add_new_callers): Update only IPA profile.
7363 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7364 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7365 (dump_ipa_call_summary): Use frequency function.
7366 (estimate_edge_size_and_time): Use frequency function.
7367 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7368 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7369 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7370 (update_noncloned_counts): ... ths one; scale counts only.
7371 (clone_inlined_nodes): Do not scale frequency.
7372 (inline_call): Do not pass frequency.
7373 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7374 (compute_inlined_call_time): Use IPA profile.
7375 (want_inline_small_function_p): Use IPA profile.
7376 (want_inline_self_recursive_call_p): Use IPA profile.
7377 (edge_badness): Use IPA profile.
7378 (lookup_recursive_calls): Use IPA profile.
7379 (recursive_inlining): Do not pass frequency.
7380 (resolve_noninline_speculation): Do not update frequency.
7381 (inline_small_functions): Collect max of IPA profile.
7382 (dump_overall_stats): Dump IPA porfile.
7383 (dump_inline_stats): Dump IPA porfile.
7384 (ipa_inline): Collect IPA stats.
7385 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7386 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7387 (ipa_propagate_frequency): Use frequency function.
7388 (ipa_profile): Cleanup.
7389 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7390 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7391 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7392 (input_node): Do not stream frequency.
7393 (input_edge): Do not stream frequency.
7394 (merge_profile_summaries): Scale only IPA profiles.
7395 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7396 * predict.c (drop_profile): Do not recompute frequency.
7397 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7398 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7399 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7400 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7401 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7402 * tree-inline.c (copy_bb): Do not scale frequency.
7403 (expand_call_inline): Do not scale frequency.
7404 (tree_function_versioning): Do not scale frequency.
7405 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7407 2017-11-10 Julia Koval <julia.koval@intel.com>
7409 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7410 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7411 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7412 (_mm256_maskz_gf2p8affine_epi64_epi8)
7413 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7414 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7415 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7416 (__builtin_ia32_vgf2p8affineqb_v32qi)
7417 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7418 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7420 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7424 (arm_option_override): Refactor.
7425 (arm_option_reconfigure_globals): New.
7426 (arm_options_perform_arch_sanity_checks): New.
7427 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7429 (arm_options_perform_arch_sanity_checks): Likewise
7431 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7433 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7434 (power9-qpmul): New.
7435 * rs6000/rs6000.md ("type" attr): Add qmul.
7436 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7437 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7438 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7440 2017-11-10 Martin Sebor <msebor@redhat.com>
7443 * builtins.c (compute_objsize): Handle arrays that
7444 compute_builtin_object_size likes to fail for. Make extern.
7445 * builtins.h (compute_objsize): Declare.
7446 (check_strncpy_sizes): New function.
7447 (expand_builtin_strncpy): Call check_strncpy_sizes.
7448 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7449 -Wstringop-truncation.
7450 (gimple_fold_builtin_strncat): Same.
7451 * gimple.c (gimple_build_call_from_tree): Set call location.
7452 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7453 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7454 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7455 (handle_builtin_strlen): Use strlen_to_stridx.
7456 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7458 Use strlen_to_stridx.
7459 (pass_strlen::execute): Release strlen_to_stridx.
7460 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7461 (-Wstringop-truncation): Document new option.
7463 2017-11-10 Martin Liska <mliska@suse.cz>
7465 PR gcov-profile/82702
7466 * gcov.c (main): Handle intermediate files in a different
7468 (get_gcov_intermediate_filename): New function.
7469 (output_gcov_file): Remove support of intermediate files.
7470 (generate_results): Allocate intermediate file.
7471 (release_structures): Clean-up properly fn_end.
7472 (output_intermediate_file): Start iterating with line 1.
7474 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7476 PR tree-optimization/82929
7477 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7478 ops_swapped_p non-static data member.
7479 (store_immediate_info::store_immediate_info): Clear it.
7480 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7481 ops set ops_swapped_p.
7482 (count_multiple_uses): Handle ops_swapped_p.
7484 2017-11-10 Martin Liska <mliska@suse.cz>
7486 * coverage.c (coverage_init): Stream information about
7487 support of has_unexecuted_blocks.
7488 * doc/gcov.texi: Document that.
7489 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7490 * gcov.c (read_graph_file): Likewise.
7491 (output_line_beginning): Fix a small issue with
7494 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7496 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7497 reference of trivial component.
7499 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7502 * gimple-ssa-store-merging.c
7503 (pass_store_merging::terminate_all_aliasing_chains): For
7504 gimple_store_p stmts also call refs_output_dependent_p.
7506 PR rtl-optimization/82913
7507 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7510 2017-11-09 Jeff Law <law@redhat.com>
7512 * vr-values.h: New file with vr_values class.
7513 * tree-vrp.c: Include vr-values.h
7514 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7515 data objects into the vr_values class.
7516 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7517 (get_value_range): Make it a member function within vr_values class.
7518 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7519 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7520 (op_with_constant_singleton_value_range): Likewise.
7521 (extract_range_for_var_from_comparison_expr): Likewise.
7522 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7523 (extract_range_from_binary_expr): Likewise.
7524 (extract_range_from_unary_expr): Likewise.
7525 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7526 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7527 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7528 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7529 (compare_name_with_value, compare_names): Likewise.
7530 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7531 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7532 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7533 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7534 (extract_range_from_phi_node): Likewise.
7535 (simplify_truth_ops_using_ranges): Likewise.
7536 (simplify_div_or_mod_using_ranges): Likewise.
7537 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7538 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7539 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7540 (simplify_float_conversion_using_ranges): Likewise.
7541 (simplify_internal_call_using_ranges): Likewise.
7542 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7543 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7544 poitner via x_vr_values for calls into gimple folder.
7545 (vrp_initialize_lattice): Make this the vr_values ctor.
7546 (vrp_free_lattice): Make this the vr_values dtor.
7547 (set_vr_value): New function.
7548 (class vrp_prop): Add vr_values data member. Add various member
7549 functions as well as member functions that delegate to vr_values.
7550 (check_array_ref): Make a member function within vrp_prop class.
7551 (search_for_addr_array, vrp_initialize): Likewise.
7552 (vrp_finalize): Likewise. Revamp to avoid direct access to
7553 vr_value, values_propagated, etc.
7554 (check_array_bounds): Extract vrp_prop class instance pointer from
7555 walk info structure. Use it to call member functions.
7556 (check_all_array_refs): Make a member function within vrp_prop class.
7557 Smuggle class instance pointer via walk info structure.
7558 (x_vr_values): New local static.
7559 (vrp_valueize): Use x_vr_values to get class instance.
7560 (vr_valueize_1): Likewise.
7561 (class vrp_folder): Add vr_values data member. Add various member
7562 functions as well as member functions that delegate to vr_values.
7563 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7564 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7565 class instance from vr_values. Use it to call member functions.
7566 (vrp_dom_walker): Add vr_values data member.
7567 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7568 instance via x_vr_values.
7569 (identify_jump_threads): Accept vr_values as argument. Store
7570 it into the walker structure.
7571 (evrp_dom_walker): Add vr_values class data member. Add various
7573 (evrp_dom_walker::try_find_new_range): Use vr_values data
7574 member to access the memory allocator.
7575 (evrp_dom_walker::before_dom_children): Store vr_values class
7576 instance into the vrp_folder class.
7577 (evrp_dom_walker::push_value_range): Rework to avoid direct
7578 access to num_vr_values and vr_value.
7579 (evrp_dom_walker::pop_value_range): Likewise.
7580 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7581 Use vr_values to get to dump_all_value_ranges member function.
7582 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7583 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7584 Pass vr_values class instance down to identify_jump_threads.
7585 Remove call to vrp_free_lattice.
7586 (debug_all_value_ranges): Remove.
7588 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7589 (vrp_folder): Likewise.
7591 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7592 Get it from the existing bitmap instead.
7593 (vrp_intersect_ranges_1): Likewise.
7595 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7597 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7599 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7600 set bit_not_p to it.
7601 (imm_store_chain_info::coalesce_immediate_stores): Break group
7602 if bit_not_p is different.
7603 (count_multiple_uses, split_group,
7604 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7605 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7606 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7607 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7608 to store_immediate_info ctor.
7610 2017-11-09 Jim Wilson <jimw@sifive.com>
7612 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7613 (scan_prog_file): Likewise.
7615 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7617 * bb-reorder.c (max_entry_frequency): Remove.
7618 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7619 connect_traces, push_to_next_round_p): Remove prototypes.
7620 (find_traces_1_round): Use counts only.
7621 (push_to_next_round_p): Likewise.
7622 (find_traces): Likewise.
7623 (rotate_loop): Likewise.
7624 (find_traces_1_round): Likewise.
7625 (connect_traces): Likewise.
7626 (edge_order): Likewise.
7628 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7630 * config/arm/arm.c (output_return_instruction): Add comments to
7631 indicate requirement for cmse_nonsecure_entry return to account
7632 for the size of clearing instruction output here.
7633 (thumb_exit): Likewise.
7634 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7635 return in hardfloat mode.
7637 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7639 * config/rs6000/rs6000.c (machine_function): Add a bool,
7640 "toc_is_wrapped_separately".
7641 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7642 if it wasn't explicitly set or unset, we are optimizing for speed, and
7643 doing separate shrink-wrapping.
7644 (rs6000_get_separate_components): Enable the TOC component if
7645 saving the TOC register in the prologue.
7646 (rs6000_components_for_bb): Handle the TOC component.
7647 (rs6000_emit_prologue_components): Store the TOC register where needed.
7648 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7649 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7651 2017-11-09 Martin Jambor <mjambor@suse.cz>
7653 * ipa-param-manipulation.c: New file.
7654 * ipa-param-manipulation.h: Likewise.
7655 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7656 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7657 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7658 (ipa_parm_adjustment): Likewise.
7659 (ipa_parm_adjustment_vec): Likewise.
7660 (ipa_get_vector_of_formal_parms): Moved declaration to
7661 ipa-param-manipulation.h.
7662 (ipa_get_vector_of_formal_parm_types): Likewise.
7663 (ipa_modify_formal_parameters): Likewise.
7664 (ipa_modify_call_arguments): Likewise.
7665 (ipa_combine_adjustments): Likewise.
7666 (ipa_dump_param_adjustments): Likewise.
7667 (ipa_modify_expr): Likewise.
7668 (ipa_get_adjustment_candidate): Likewise.
7669 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7670 ipa-param-manipulation.c.
7671 (ipa_get_vector_of_formal_parm_types): Likewise.
7672 (ipa_modify_formal_parameters): Likewise.
7673 (ipa_modify_call_arguments): Likewise.
7674 (ipa_modify_expr): Likewise.
7675 (get_ssa_base_param): Likewise.
7676 (ipa_get_adjustment_candidate): Likewise.
7677 (index_in_adjustments_multiple_times_p): Likewise.
7678 (ipa_combine_adjustments): Likewise.
7679 (ipa_dump_param_adjustments): Likewise.
7680 * tree-sra.c: Also include ipa-param-manipulation.h
7681 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7684 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7685 Alan Hayward <alan.hayward@arm.com>
7686 David Sherwood <david.sherwood@arm.com>
7688 * doc/sourcebuild.texi (vect_masked_store): Document.
7690 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7691 Alan Hayward <alan.hayward@arm.com>
7692 David Sherwood <david.sherwood@arm.com>
7694 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7696 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7697 Alan Hayward <alan.hayward@arm.com>
7698 David Sherwood <david.sherwood@arm.com>
7700 * doc/sourcebuild.texi (vect_variable_length): Document.
7702 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7703 Alan Hayward <alan.hayward@arm.com>
7704 David Sherwood <david.sherwood@arm.com>
7706 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7708 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7709 Alan Hayward <alan.hayward@arm.com>
7710 David Sherwood <david.sherwood@arm.com>
7712 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7714 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7715 Alan Hayward <alan.hayward@arm.com>
7716 David Sherwood <david.sherwood@arm.com>
7718 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7719 previously undocumented selectors.
7720 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7722 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7724 * doc/rtl.texi (const_vector): Say that elements can be
7725 const_wide_ints too.
7726 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7727 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7728 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7729 * optabs.c (expand_vector_broadcast): Likewise.
7731 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7732 Alan Hayward <alan.hayward@arm.com>
7733 David Sherwood <david.sherwood@arm.com>
7735 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7736 scaled index even if the unscaled address was invalid.
7737 Don't increase the complexity of using a scale in that case.
7739 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7740 Alan Hayward <alan.hayward@arm.com>
7741 David Sherwood <david.sherwood@arm.com>
7743 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7744 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7745 UNITS_PER_WORD bytes.
7746 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7747 based on REGMODE_NATURAL_SIZE of the inner mode.
7748 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7749 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7750 * expmed.c (lowpart_bit_field_p): Divide the value up into
7751 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7752 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7753 whether something is likely to occupy more than one register.
7755 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7758 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7759 profile_count::adjust_for_ipa_scaling.
7760 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7761 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7763 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7765 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7767 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7768 (split_group): Add total_orig and total_new arguments, estimate the
7769 number of statements related to the store group without store merging
7770 and with store merging.
7771 (imm_store_chain_info::output_merged_store): Adjust split_group
7772 callers, punt if estimated number of statements with store merging
7773 is not smaller than estimated number of statements without it.
7775 (handled_load): Remove has_single_use checks.
7776 (pass_store_merging::process_store): Likewise.
7778 2017-11-09 Richard Biener <rguenther@suse.de>
7780 PR tree-optimization/82902
7781 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7783 2017-11-09 Martin Liska <mliska@suse.cz>
7786 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7789 2017-11-09 Martin Liska <mliska@suse.cz>
7791 PR tree-optimization/82669
7792 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7794 2017-11-09 Martin Liska <mliska@suse.cz>
7796 PR gcov-profile/48463
7797 * coverage.c (coverage_begin_function): Output also end locus
7798 of a function and information whether the function is
7800 * gcov-dump.c (tag_function): Parse and print the information.
7801 * gcov.c (INCLUDE_MAP): Add include.
7802 (INCLUDE_SET): Likewise.
7803 (struct line_info): Move earlier in the source file because
7804 of vector<line_info> in function_info structure.
7805 (line_info::line_info): Likewise.
7806 (line_info::has_block): Likewise.
7807 (struct source_info): Add new member index.
7808 (source_info::get_functions_at_location): New function.
7809 (function_info::group_line_p): New function.
7810 (output_intermediate_line): New function.
7811 (output_intermediate_file): Use the mentioned function.
7812 (struct function_start): New.
7813 (struct function_start_pair_hash): Likewise.
7814 (process_file): Add code that identifies group functions.
7815 Assign lines either to global or function scope.
7816 (generate_results): Skip artificial functions.
7817 (find_source): Assign index for each source file.
7818 (read_graph_file): Read new flag artificial and end_line.
7819 (add_line_counts): Assign it either to global of function scope.
7820 (accumulate_line_counts): Isolate core of the function to
7821 accumulate_line_info and call it for both function and global
7823 (accumulate_line_info): New function.
7824 (output_line_beginning): Fix GNU coding style.
7825 (print_source_line): New function.
7826 (output_line_details): Likewise.
7827 (output_function_details): Likewise.
7828 (output_lines): Iterate both source (global) scope and function
7830 (struct function_line_start_cmp): New class.
7831 * doc/gcov.texi: Reflect changes in documentation.
7833 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7836 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7837 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7838 and similarly for not instead of neg.
7840 2017-11-08 Andi Kleen <ak@linux.intel.com>
7842 * config/i386/i386.opt: Add -mforce-indirect-call.
7843 * config/i386/predicates.md: Check for flag_force_indirect_call.
7844 * doc/invoke.texi: Document -mforce-indirect-call
7846 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7848 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7850 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7851 config/riscv/riscv.c (predict.h): New include.
7852 (riscv_slow_unaligned_access_p): No longer static.
7853 (riscv_block_move_straight): Add require.
7854 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7856 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7859 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7860 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7861 nonimmediate_operand predicate for operand 1 instead of
7864 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7866 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7868 * config/aarch64/constraints.md (Uml): New constraint.
7869 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7872 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7874 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7875 of two vec_duplicates into a vec_concat.
7877 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7879 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7880 Simplify vec_merge of vec_duplicate and vec_concat.
7881 * config/aarch64/constraints.md (Utq): New constraint.
7882 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7885 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7887 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7888 Simplify vec_merge of vec_duplicate and const_vector.
7889 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7891 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7892 mode iterator. Update predicate on operand 1 to
7893 handle non-const_vec constants. Delete constraints.
7894 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7896 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7898 PR tree-optimization/78821
7899 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7901 (store_operand_info::store_operand_info): Initialize it to false.
7902 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7903 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7904 store in the group, and if chain_info is non-NULL, to ignore altogether
7906 (compatible_load_p): Fail if bit_not_p does not match.
7907 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7908 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7909 (pass_store_merging::process_store): Adjust
7910 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7911 call terminate_all_aliasing_chains newly when adding a store into
7912 a chain with non-NULL chain_info.
7914 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7916 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7918 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7920 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7922 (aarch64_layout_frame): Initialise emit_frame_chain.
7923 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7924 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7926 2017-11-08 Martin Liska <mliska@suse.cz>
7928 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7929 of gimple_call_internal_p.
7931 2017-11-07 Tom de Vries <tom@codesourcery.com>
7933 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7935 2017-11-07 Tom de Vries <tom@codesourcery.com>
7937 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7940 2017-11-08 Martin Liska <mliska@suse.cz>
7943 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7945 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7947 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7949 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7952 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7953 (*cmp<mode>_ccz_1): New insn with $k alternative.
7956 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7957 RTX_COMM_COMPARE as commutative as well.
7958 (ix86_binary_operator_ok): Formatting fix.
7959 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7960 *<code><mode>3<mask_name><round_saeonly_name>,
7961 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7962 *<s>mul<mode>3_highpart<mask_name>,
7963 *vec_widen_umult_even_v16si<mask_name>,
7964 *vec_widen_umult_even_v8si<mask_name>,
7965 *vec_widen_umult_even_v4si<mask_name>,
7966 *vec_widen_smult_even_v16si<mask_name>,
7967 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7968 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7969 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7970 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7971 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7972 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7973 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7974 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7975 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7976 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7977 ix86_binary_operator_ok. Formatting fixes.
7978 (*<plusminus_insn><mode>3<mask_name><round_name>,
7979 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7982 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7984 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7985 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7986 eq and ne if TARGET_ISEL.
7987 (cmp): New code_iterator.
7988 (UNS, UNSU_, UNSIK): New code_attrs.
7989 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7990 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7992 ("eq<mode>3"): ... this.
7993 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7995 ("ne<mode>3"): ... this.
7997 2017-11-07 Julia Koval <julia.koval@intel.com>
8000 * common/config/i386/i386-common.c
8001 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8002 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8003 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8004 * config/i386/i386.opt: Ditto.
8005 * config/i386/i386.c (ix86_target_string): Ditto.
8006 (ix86_option_override_internal): Ditto.
8007 (ix86_init_mpx_builtins): Move MPX to args2.
8008 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8009 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8010 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8011 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8012 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8013 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8014 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8016 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8019 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8020 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8021 (zero-extendsidi peephole2): Remove peephole.
8023 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8026 * stor-layout.c: Include attribs.h.
8027 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8028 explicit lookup of "aligned" attribute.
8030 2017-11-07 Andrew Waterman <andrew@sifive.com>
8032 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8033 (riscv_expand_block_move): Likewise.
8034 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8036 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8037 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8038 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8039 (riscv_adjust_block_mem): Likewise.
8040 (riscv_block_move_loop): Likewise.
8041 (riscv_expand_block_move): Likewise.
8042 * config/riscv/riscv.md (movmemsi): New pattern.
8044 2017-11-07 Michael Clark <michaeljclark@mac.com>
8046 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8047 (MUSL_DYNAMIC_LINKER): Likewise.
8049 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8051 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8052 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8055 2017-11-07 Richard Biener <rguenther@suse.de>
8057 * match.pd: Fix build.
8059 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8060 Jackson Woodruff <jackson.woodruff@arm.com>
8062 PR tree-optimization/71026
8063 * match.pd: Canonicalize negate in division.
8065 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8068 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8070 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8072 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8073 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8075 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8077 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8078 non-scalar integral types.
8079 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8080 (-(A-B), -(~A)): New transformations.
8082 2017-11-07 Tom de Vries <tom@codesourcery.com>
8084 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8085 semicolon after "do {} while (0)".
8086 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8087 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8088 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8089 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8090 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8092 2017-11-07 Tom de Vries <tom@codesourcery.com>
8094 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8095 after "do {} while (0)".
8096 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8097 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8098 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8099 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8100 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8102 2017-11-07 Tom de Vries <tom@codesourcery.com>
8105 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8107 (arm_rtx_costs_internal): Add missing semicolon after
8108 HANDLE_NARROW_SHIFT_ARITH call.
8110 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8112 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8113 disable isel if it was not set explicitly.
8115 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8117 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8118 (add_type_attribute) likewise.
8120 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8122 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8123 of struct ix86_frame.
8124 (ix86_initial_elimination_offset): Likewise.
8125 (ix86_expand_split_stack_prologue): Likewise.
8127 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8129 * tree-vrp.h (enum value_range_type): Update stale comment.
8131 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8132 Alan Hayward <alan.hayward@arm.com>
8133 David Sherwood <david.sherwood@arm.com>
8135 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8136 (aarch64_expand_vec_perm_const): Take the number of units too.
8137 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8138 (aarch64_expand_vec_perm_const): Likewise.
8139 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8140 (vec_perm<mode>): Update accordingly.
8142 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8143 Alan Hayward <alan.hayward@arm.com>
8144 David Sherwood <david.sherwood@arm.com>
8146 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8147 Take the number of units too.
8148 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8149 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8150 but check for a vector mode before rather than after the call.
8151 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8152 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8153 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8154 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8155 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8156 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8157 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8158 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8159 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8160 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8161 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8162 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8163 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8164 (aarch64_sqdmull2_n<mode>): Update accordingly.
8166 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8167 Alan Hayward <alan.hayward@arm.com>
8168 David Sherwood <david.sherwood@arm.com>
8170 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8171 the number of units too.
8172 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8173 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8174 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8175 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8176 (vec_store_lanesxi<mode>): Update accordingly.
8178 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8179 Alan Hayward <alan.hayward@arm.com>
8180 David Sherwood <david.sherwood@arm.com>
8182 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8183 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8184 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8185 of units rather than the mode.
8186 * config/aarch64/iterators.md (nunits): New mode attribute.
8187 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8188 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8189 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8190 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8191 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8192 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8193 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8194 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8195 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8196 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8197 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8198 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8199 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8200 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8201 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8202 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8203 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8204 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8205 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8206 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8207 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8208 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8209 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8210 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8211 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8212 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8213 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8214 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8215 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8216 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8217 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8218 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8219 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8220 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8221 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8222 (aarch64_simd_vec_setv2di): Likewise.
8224 2017-11-06 Carl Love <cel@us.ibm.com>
8226 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8228 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8229 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8231 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8233 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8234 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8235 Add power 8 macro expansions.
8236 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8237 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8238 power 8 instructions. (VSX_XXBR): Add iterator.
8240 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8242 * config/arm/arm.md (predicable_short_it): Change default to "no",
8243 improve documentation, remove uses that are identical to the default.
8244 (enabled_for_depr_it): Rename to enabled_for_short_it.
8245 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8246 Remove default uses.
8247 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8248 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8249 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8250 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8252 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8255 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8256 float128 helper macros, which are no longer used after deleting
8257 the old 'q' built-in functions, and moving the round to odd
8258 built-in functions to being special built-in functions.
8259 (BU_FLOAT128_2): Likewise.
8260 (BU_FLOAT128_1_HW): Likewise.
8261 (BU_FLOAT128_2_HW): Likewise.
8262 (BU_FLOAT128_3_HW): Likewise.
8263 (FABSQ): Delete old 'q' built-in functions.
8264 (COPYSIGNQ): Likewise.
8265 (SQRTF128_ODD): Move round to odd built-in functions to be
8266 special built-in functions, so that we can handle
8267 -mabi=ieeelongdouble.
8268 (TRUNCF128_ODD): Likewise.
8269 (ADDF128_ODD): Likewise.
8270 (SUBF128_ODD): Likewise.
8271 (MULF128_ODD): Likewise.
8272 (DIVF128_ODD): Likewise.
8273 (FMAF128_ODD): Likewise.
8274 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8275 built-in names to 'f128'.
8276 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8277 old 'q' built-in functions, as the machine independent code for
8278 'f128' built-in functions handles this.
8279 (rs6000_expand_builtin): Add expansion for float128 round to odd
8280 functions, keying off on -mabi=ieeelongdouble of whether to use
8281 the KFmode or TFmode variant.
8282 (rs6000_init_builtins): Initialize the _Float128 round to odd
8284 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8285 _Float128 'q' built-in functions are now mapped into the new
8286 'f128' built-in functions.
8288 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8290 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8291 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8293 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8296 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8297 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8298 * toplev.c: Include "ipa-fnsummary.h".
8299 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8301 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8303 PR tree-optimization/82838
8304 * gimple-ssa-store-merging.c
8305 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8306 on a separate gimple_seq which is then appended to seq.
8308 2017-11-06 Jeff Law <law@redhat.com>
8311 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8312 (get_probe_interval): New functions.
8313 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8314 (ix86_adjust_stack_and_probe): Likewise.
8315 (output_adjust_stack_and_probe): Likewise.
8316 (ix86_emit_probe_stack_range): Likewise.
8317 (ix86_expand_prologue): Likewise.
8319 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8321 PR tree-optimization/82816
8322 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8323 if the modes of the two types are the same.
8324 (convert_plusminus_to_widen): Likewise.
8326 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8328 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8330 (usadv16qi): New define_expand.
8331 (usadv8hi): New define_expand.
8333 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8336 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8338 (inline_transform): Likewise.
8339 * predict.c (counts_to_freqs): Remove useless conditional.
8340 * profile-count.h (profile_count::apply_scale): Move sanity check.
8341 * tree-inline.c (copy_bb): Always scale.
8342 (copy_cfg_body): Likewise.
8344 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8347 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8349 (*cmp_ite0): Add enabled_for_depr_it attribute.
8350 (*cmp_ite1): Likewise.
8352 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8354 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8357 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8359 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8360 return the value by pointer.
8361 (extract_range_from_multiplicative_op_1): Update accordingly.
8362 Return as soon as an operation fails.
8364 2017-11-05 Tom de Vries <tom@codesourcery.com>
8367 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8368 (DEF_SANITIZER_BUILTIN): ... here.
8369 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8370 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8372 2017-11-05 Tom de Vries <tom@codesourcery.com>
8375 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8377 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8378 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8379 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8381 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8382 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8383 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8385 2017-11-05 Tom de Vries <tom@codesourcery.com>
8388 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8391 2017-11-04 Michael Clark <michaeljclark@mac.com>
8393 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8394 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8396 (*addsi3_extended): Likewise.
8397 (*addsi3_extended2): Likewise.
8398 (<optab>si3): Likewise.
8399 (<optab>di3): Likewise.
8400 (<optab><mode>3): Likewise.
8401 (<*optabe>si3_internal): Likewise.
8402 (zero_extendqi<SUPERQI:mode>2): Likewise.
8403 (*add<mode>hi3): Likewise.
8404 (*xor<mode>hi3): Likewise.
8405 (<optab>di3): Likewise.
8406 (*<optab>si3_extend): Likewise.
8407 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8408 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8409 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8411 2017-11-04 Andrew Waterman <andrew@sifive.com>
8413 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8414 TARGET_STRICT_ALIGN based upon -mtune argument.
8416 2017-11-04 Andrew Waterman <andrew@sifive.com>
8418 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8420 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8422 * config/i386/i386.c (choose_basereg): Use optional scratch
8423 register and add assertion.
8424 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8425 needed, and don't allocate stack.
8426 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8427 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8428 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8430 2017-11-03 Jeff Law <law@redhat.com>
8432 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8433 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8434 to probe at the start of a noreturn function.
8436 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8438 PR tree-optimization/78821
8439 * gimple-ssa-store-merging.c: Update the file comment.
8440 (MAX_STORE_ALIAS_CHECKS): Define.
8441 (struct store_operand_info): New type.
8442 (store_operand_info::store_operand_info): New constructor.
8443 (struct store_immediate_info): Add rhs_code and ops data members.
8444 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8445 and op1r arguments to the ctor, initialize corresponding data members.
8446 (struct merged_store_group): Add load_align_base and load_align
8448 (merged_store_group::merged_store_group): Initialize them.
8449 (merged_store_group::do_merge): Update them.
8450 (merged_store_group::apply_stores): Pick the constant for
8451 encode_tree_to_bitpos from one of the two operands, or skip
8452 encode_tree_to_bitpos if neither operand is a constant.
8453 (class pass_store_merging): Add process_store method decl. Remove
8454 bool argument from terminate_all_aliasing_chains method decl.
8455 (pass_store_merging::terminate_all_aliasing_chains): Remove
8456 var_offset_p argument and corresponding handling.
8457 (stmts_may_clobber_ref_p): New function.
8458 (compatible_load_p): New function.
8459 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8460 if there is overlap and rhs_code is not INTEGER_CST. For
8461 non-overlapping stores terminate group if rhs is not mergeable.
8462 (get_alias_type_for_stmts): Change first argument from
8463 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8464 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8465 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8467 (get_location_for_stmts): Change first argument from
8468 auto_vec<gimple *> & to vec<gimple *> &.
8469 (struct split_store): Remove orig_stmts data member, add orig_stores.
8470 (split_store::split_store): Create orig_stores rather than orig_stmts.
8471 (find_constituent_stmts): Renamed to ...
8472 (find_constituent_stores): ... this. Change second argument from
8473 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8474 to info structures rather than the statements.
8475 (split_group): Rename ALLOW_UNALIGNED argument to
8476 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8477 it. Adjust find_constituent_stores caller.
8478 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8479 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8480 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8481 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8482 (mem_valid_for_store_merging): New function.
8483 (handled_load): New function.
8484 (pass_store_merging::process_store): New method.
8485 (pass_store_merging::execute): Use process_store method. Adjust
8486 terminate_all_aliasing_chains caller.
8488 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8490 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8491 Return true for more constants, symbols and label references.
8492 (aarch64_valid_floating_const): Remove unused function.
8494 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8497 * config/aarch64/aarch64.c (aarch64_layout_frame):
8498 Undo forcing of LR at bottom of frame.
8500 2017-11-03 Jeff Law <law@redhat.com>
8503 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8504 for int_registers_saved.
8506 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8507 extracted from tree-ssa-dom.c.
8508 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8509 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8510 (record_equivalences_from_incoming_edge): Add additional argument
8511 to single_pred_edge_ignoring_loop_edges call.
8512 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8513 (uncprop_dom_walker::before_dom_children): Add additional argument
8514 to single_pred_edge_ignoring_loop_edges call.
8515 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8516 single_pred_edge_ignoring_loop_edges rather than open coding.
8517 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8519 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8521 * match.pd (-(-A)): Rewrite.
8523 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8525 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8526 (rs6000_emit_int_cmove): New declaration.
8527 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8528 (rs6000_emit_sISEL): Delete.
8529 (rs6000_emit_int_cmove): Make non-static.
8530 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8531 instead of rs6000_emit_sISEL.
8533 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8535 * asan.c (create_cond_insert_point): Maintain profile.
8536 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8538 * basic-block.h (struct basic_block_def): Remove frequency.
8539 (EDGE_FREQUENCY): Use to_frequency
8540 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8542 (find_traces): Update to use to_frequency.
8543 (find_traces_1_round): Likewise; use only IPA counts.
8544 (bb_to_key): Likewise.
8545 (connect_traces): Use IPA counts only.
8546 (copy_bb_p): Update to use to_frequency.
8547 (fix_up_crossing_landing_pad): Likewise.
8548 (sanitize_hot_paths): Likewise.
8549 * bt-load.c (basic_block_freq): Likewise.
8550 * cfg.c (init_flow): Set count_max to uninitialized.
8551 (check_bb_profile): Remove frequencies; check counts.
8552 (dump_bb_info): Do not dump frequencies.
8553 (update_bb_profile_for_threading): Update counts only.
8554 (scale_bbs_frequencies_int): Likewise.
8555 (MAX_SAFE_MULTIPLIER): Remove.
8556 (scale_bbs_frequencies_gcov_type): Update counts only.
8557 (scale_bbs_frequencies_profile_count): Update counts only.
8558 (scale_bbs_frequencies): Update counts only.
8559 * cfg.h (struct control_flow_graph): Add count-max.
8560 (update_bb_profile_for_threading): Update prototype.
8561 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8562 (find_many_sub_basic_blocks): Likewise.
8563 * cfgcleanup.c (try_forward_edges): Likewise.
8564 (try_crossjump_to_edge): Likewise.
8565 * cfgexpand.c (expand_gimple_cond): Likewise.
8566 (expand_gimple_tailcall): Likewise.
8567 (construct_init_block): Likewise.
8568 (construct_exit_block): Likewise.
8569 * cfghooks.c (verify_flow_info): Check consistency of counts.
8570 (dump_bb_for_graph): Do not dump frequencies.
8571 (split_block_1): Do not update frequencies.
8572 (split_edge): Do not update frequencies.
8573 (make_forwarder_block): Do not update frequencies.
8574 (duplicate_block): Do not update frequencies.
8575 (account_profile_record): Do not update frequencies.
8576 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8577 for global heuristics.
8578 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8579 (expected_loop_iterations_unbounded): Use counts only.
8580 * cfgloopmanip.c (scale_loop_profile): Simplify.
8581 (create_empty_loop_on_edge): Simplify
8583 (duplicate_loop_to_header_edge): Simplify
8584 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8585 (update_br_prob_note): Take care of removing note when profile
8587 (relink_block_chain): Do not dump frequency.
8588 (rtl_account_profile_record): Use to_frequency.
8589 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8590 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8591 (cgraph_update_edges_for_call_stmt_node): Likewise.
8592 (cgraph_edge::verify_count_and_frequency): Update.
8593 (cgraph_node::verify_node): Temporarily disable frequency verification.
8594 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8595 to_cgraph_frequency.
8596 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8597 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8599 (cgraph_node::expand_thunk): Update profile.
8600 * except.c (dw2_build_landing_pads): Do not update frequency.
8601 * final.c (compute_alignments): Use to_frequency.
8602 (dump_basic_block_info): Do not dump frequency.
8603 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8604 (dump_gimple_bb_header): Do not dump frequency.
8605 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8607 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8608 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8609 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8610 (init_before_recovery): Do not update frequency.
8611 (sched_create_recovery_edges): Do not update frequency.
8612 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8613 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8614 (ipa_cp_c_finalize): Set max_count to uninitialized.
8615 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8616 (param_change_prob): Use counts.
8617 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8619 * ipa-split.c (consider_split): Use to_frequency.
8620 (split_function): Use to_frequency.
8621 * ira-build.c (loop_compare_func): Likewise.
8622 (mark_loops_for_removal): Likewise.
8623 (mark_all_loops_for_removal): Likewise.
8624 * loop-doloop.c (doloop_modify): Do not update frequency.
8625 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8627 * lto-streamer-in.c (input_function): Update count_max.
8628 * omp-expand.c (expand_omp_taskreg): Update count_max.
8629 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8630 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8631 (maybe_hot_count_p): Use ipa counts only.
8632 (maybe_hot_bb_p): Simplify.
8633 (maybe_hot_edge_p): Simplify.
8634 (probably_never_executed): Do not take frequency argument.
8635 (probably_never_executed_bb_p): Do not pass frequency.
8636 (probably_never_executed_edge_p): Likewise.
8637 (combine_predictions_for_bb): Check that profile is nonzero.
8638 (propagate_freq): Do not set frequency.
8639 (drop_profile): Simplify.
8640 (counts_to_freqs): Simplify.
8641 (expensive_function_p): Use to_frequency.
8642 (propagate_unlikely_bbs_forward): Simplify.
8643 (determine_unlikely_bbs): Simplify.
8644 (estimate_bb_frequencies): Add hack to silence graphite issues.
8645 (compute_function_frequency): Use ipa counts.
8646 (pass_profile::execute): Update.
8647 (rebuild_frequencies): Use counts only.
8648 (force_edge_cold): Use counts only.
8649 * profile-count.c (profile_count::dump): Dump new count types.
8650 (profile_count::differs_from_p): Check compatiblity.
8651 (profile_count::to_frequency): New function.
8652 (profile_count::to_cgraph_frequency): New function.
8653 * profile-count.h (struct function): Declare.
8654 (enum profile_quality): Add profile_guessed_local and
8655 profile_guessed_global0.
8656 (class profile_proability): Decrease number of bits to 29;
8657 update from_reg_br_prob_note and to_reg_br_prob_note.
8658 (class profile_count: Update comment; decrease number of bits
8659 to 61. Check compatibility.
8660 (profile_count::compatible_p): New private member function.
8661 (profile_count::ipa_p): New member function.
8662 (profile_count::operator<): Handle global zero correctly.
8663 (profile_count::operator>): Handle global zero correctly.
8664 (profile_count::operator<=): Handle global zero correctly.
8665 (profile_count::operator>=): Handle global zero correctly.
8666 (profile_count::nonzero_p): New member function.
8667 (profile_count::force_nonzero): New member function.
8668 (profile_count::max): New member function.
8669 (profile_count::apply_scale): Handle IPA scalling.
8670 (profile_count::guessed_local): New member function.
8671 (profile_count::global0): New member function.
8672 (profile_count::ipa): New member function.
8673 (profile_count::to_frequency): Declare.
8674 (profile_count::to_cgraph_frequency): Declare.
8675 * profile.c (OVERLAP_BASE): Delete.
8676 (compute_frequency_overlap): Delete.
8677 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8678 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8679 * sched-ebb.c (rank): Use counts only.
8680 * shrink-wrap.c (handle_simple_exit): Use counts only.
8681 (try_shrink_wrapping): Use counts only.
8682 (place_prologue_for_one_component): Use counts only.
8683 * tracer.c (find_best_predecessor): Use to_frequency.
8684 (find_trace): Use to_frequency.
8685 (tail_duplicate): Use to_frequency.
8686 * trans-mem.c (expand_transaction): Do not update frequency.
8687 * tree-call-cdce.c: Do not update frequency.
8688 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8689 (gimple_merge_blocks): Likewise.
8690 (gimple_split_edge): Likewise.
8691 (gimple_duplicate_sese_region): Likewise.
8692 (gimple_duplicate_sese_tail): Likewise.
8693 (move_sese_region_to_fn): Likewise.
8694 (gimple_account_profile_record): Likewise.
8695 (insert_cond_bb): Likewise.
8696 * tree-complex.c (expand_complex_div_wide): Likewise.
8697 * tree-eh.c (lower_resx): Update profile.
8698 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8700 (initialize_cfun): Do not initialize frequencies
8701 (freqs_to_counts): Delete.
8702 (copy_cfg_body): Ignore count parameter.
8703 (copy_body): Update.
8704 (expand_call_inline): Update count_max.
8705 (optimize_inline_calls): Update count_max.
8706 (tree_function_versioning): Update count_max.
8707 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8708 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8710 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8711 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8712 (try_peel_loop): Likewise.
8713 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8715 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8716 (tree_transform_and_unroll_loop): Do not use frequencies
8717 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8718 Use reliable prediction only.
8719 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8720 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8721 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8722 probability scaling.
8723 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8724 not update frequency
8725 (any_remaining_duplicated_blocks): Likewise.
8726 (update_profile): Likewise.
8727 (estimated_freqs_path): Delete.
8728 (freqs_to_counts_path): Delete.
8729 (clear_counts_path): Delete.
8730 (ssa_fix_duplicate_block_edges): Likewise.
8731 (duplicate_thread_path): Likewise.
8732 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8733 * tree-tailcall.c (decrease_profile): Do not update frequency.
8734 (eliminate_tail_call): Likewise.
8735 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8736 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8737 (optimize_mask_stores): Likewise.
8738 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8739 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8740 (ubsan_expand_ptr_ifn): Update profile.
8741 * value-prof.c (gimple_ic): Simplify.
8742 * value-prof.h (gimple_ic): Update prototype.
8743 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8744 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8746 (want_inline_self_recursive_call_p): Likewise.
8747 (resolve_noninline_speculation): Only cummulate defined counts.
8748 (inline_small_functions): Use nonzero_p.
8749 (ipa_inline): Do not access freed node.
8751 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8753 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8754 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8756 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8758 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8759 non-legitimate address.
8761 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8763 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8764 (*lt0_<mode>di, *lt0_<mode>si): New.
8766 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8768 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8769 TARGET_PAIRED_FLOAT.
8771 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8772 Jim Wilson <jim.wilson@linaro.org>
8774 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8775 * config/aarch64/aarch64-tune.md: Regenerated.
8776 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8777 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8779 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8781 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8782 (arc_expand_prologue): Restore blink for millicode.
8783 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8785 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8788 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8789 gen_vec_duplicate after forcing the scalar into a register.
8791 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8793 * combine (try_combine): Print the insns input to try_combine to the
8796 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8799 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8800 Remove second argument from aarch64_process_target_attr call.
8801 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8803 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8805 (aarch64_handle_attr_arch): Remove second argument.
8806 (aarch64_handle_attr_cpu): Ditto.
8807 (aarch64_handle_attr_tune): Ditto.
8808 (aarch64_handle_attr_isa_flags): Ditto.
8809 (aarch64_process_one_target_attr): Ditto.
8810 (aarch64_process_target_attr): Ditto.
8811 (aarch64_option_valid_attribute_p): Remove second argument.
8812 on aarch64_process_target_attr call.
8814 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8816 * diagnostic.c: Include "selftest-diagnostic.h".
8817 (selftest::assert_location_text): New function.
8818 (selftest::test_diagnostic_get_location_text): New function.
8819 (selftest::diagnostic_c_tests): Call it.
8821 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8823 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8824 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8825 (class selftest::test_diagnostic_context): Move to...
8826 * selftest-diagnostic.c: New file.
8827 * selftest-diagnostic.h: New file.
8829 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8831 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8832 offset range for FT32B.
8833 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8834 * config/ft32/ft32.md: Add TARGET_NOPM.
8835 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8836 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8838 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8840 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8842 2017-11-02 Jeff Law <law@redhat.com>
8844 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8845 virtual keyword on FINAL OVERRIDE members.
8847 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8848 virtuals together. Add virtual destructor.
8849 (substitute_and_fold_engine): Similarly.
8851 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8853 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8855 2017-11-02 Richard Biener <rguenther@suse.de>
8857 PR tree-optimization/82795
8858 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8860 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8862 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8864 (gcc_GAS_CHECK_FEATURE): Remove.
8865 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8866 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8867 * configure: Regenerate.
8869 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8871 * config/arc/arc.c (hwloop_optimize): Account for empty
8874 2017-11-02 Richard Biener <rguenther@suse.de>
8877 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8878 Truncate ARRAY_REF index and element size.
8880 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8882 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8884 2017-11-01 Jeff Law <law@redhat.com>
8886 * tree-ssa-ccp.c (ccp_folder): New class derived from
8887 substitute_and_fold_engine.
8888 (ccp_folder::get_value): New member function.
8889 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8890 (ccp_fold_stmt): Remove prototype.
8891 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8892 * tree-ssa-copy.c (copy_folder): New class derived from
8893 substitute_and_fold_engine.
8894 (copy_folder::get_value): Renamed from get_value.
8895 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8896 * tree-vrp.c (vrp_folder): New class derived from
8897 substitute_and_fold_engine.
8898 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8899 (vrp_folder::get_value): New member function.
8900 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8901 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8902 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8903 provide a class interface to folder/substitute routines.
8904 (ssa_prop_fold_stmt_fn): Remove typedef.
8905 (ssa_prop_get_value_fn): Likewise.
8906 (subsitute_and_fold): Remove prototype.
8907 (replace_uses_in): Likewise.
8908 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8909 Renamed from replace_uses_in. Call the virtual member function
8910 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8911 (substitute_and_fold_dom_walker): Remove initialization of
8912 data member entries for calbacks. Add substitute_and_fold_engine
8913 member and initialize it.
8914 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8915 member functions for get_value, replace_phi_args_in c
8916 replace_uses_in, and fold_stmt calls.
8917 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8918 substitute_and_fold. Remove assert. Update ctor call.
8920 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8921 (ssa_prop_visit_phi_fn): Likewise.
8922 (class ssa_propagation_engine): New class to provide an interface
8924 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8926 (ssa_prop_visit_phi): Likewise.
8927 (ssa_propagation_engine::simulate_stmt): Moved into class.
8928 Call visit_phi/visit_stmt from the class rather than via
8929 file scoped static variables.
8930 (ssa_propagation_engine::simulate_block): Moved into class.
8931 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8932 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8933 set file scoped statics for the visit_stmt/visit_phi callbacks.
8934 * tree-complex.c (complex_propagate): New class derived from
8935 ssa_propagation_engine.
8936 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8937 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8938 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8940 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8941 ssa_propagation_engine.
8942 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8943 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8944 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8945 * tree-ssa-copy.c (copy_prop): New class derived from
8946 ssa_propagation_engine.
8947 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8948 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8949 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8950 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8951 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8952 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8953 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8955 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8957 PR rtl-optimization/82778
8958 PR rtl-optimization/82597
8959 * compare-elim.c (struct comparison): Add in_a_setter field.
8960 (find_comparison_dom_walker::before_dom_children): Remove killed
8961 bitmap and df_simulate_find_defs call, instead walk the defs.
8962 Compute last_setter and initialize in_a_setter. Merge definitions
8963 with first initialization for a few variables.
8964 (try_validate_parallel): Use insn_invalid_p instead of
8965 recog_memoized. Return insn rather than just the pattern.
8966 (try_merge_compare): Fix up comment. Don't uselessly test if
8967 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8969 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8972 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8973 Alan Hayward <alan.hayward@arm.com>
8974 David Sherwood <david.sherwood@arm.com>
8976 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8977 aarch64_hard_regno_nregs to get the number of registers
8980 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8981 Alan Hayward <alan.hayward@arm.com>
8982 David Sherwood <david.sherwood@arm.com>
8984 * config/aarch64/constraints.md (Upl): Rename to...
8986 * config/aarch64/aarch64.md
8987 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8990 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8991 Alan Hayward <alan.hayward@arm.com>
8992 David Sherwood <david.sherwood@arm.com>
8994 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8995 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8998 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8999 Alan Hayward <alan.hayward@arm.com>
9000 David Sherwood <david.sherwood@arm.com>
9002 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9003 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9004 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9006 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9007 of a single element.
9008 * config/aarch64/iterators.md: Add a comment above the permute
9009 unspecs to say that they are generated directly by
9010 aarch64_expand_vec_perm_const.
9011 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9013 2017-11-01 Nathan Sidwell <nathan@acm.org>
9015 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9017 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9019 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9020 and medany code models, and describe what they do.
9022 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9024 Revert accidental duplicate:
9026 * combine.c (can_change_dest_mode): Reject changes in
9027 REGMODE_NATURAL_SIZE.
9029 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9031 PR rtl-optimization/64682
9032 PR rtl-optimization/69567
9033 PR rtl-optimization/69737
9034 PR rtl-optimization/82683
9035 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9036 register mentioned in the note, drop the note, unless it came from I3,
9037 in which case it should go to I3 again.
9039 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9041 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9042 and return false if not.
9043 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9045 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9047 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9048 range is known to be empty.
9050 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9051 Alan Hayward <alan.hayward@arm.com>
9052 David Sherwood <david.sherwood@arm.com>
9054 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9055 and CONST_VECTOR_NUNITS instead of computing the number of units from
9056 the byte sizes of the vector and element.
9057 (simplify_binary_operation_1): Likewise.
9058 (simplify_const_binary_operation): Likewise.
9059 (simplify_ternary_operation): Likewise.
9061 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9062 Alan Hayward <alan.hayward@arm.com>
9063 David Sherwood <david.sherwood@arm.com>
9065 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9066 (int_mem_offset): ...this new function.
9067 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9068 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9069 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9072 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9073 Alan Hayward <alan.hayward@arm.com>
9074 David Sherwood <david.sherwood@arm.com>
9076 * lower-subreg.c (interesting_mode_p): New function.
9077 (compute_costs, find_decomposable_subregs, decompose_register)
9078 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9079 (resolve_clobber, dump_choices): Use it.
9081 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9082 Alan Hayward <alan.hayward@arm.com>
9083 David Sherwood <david.sherwood@arm.com>
9085 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9087 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9088 Alan Hayward <alan.hayward@arm.com>
9089 David Sherwood <david.sherwood@arm.com>
9091 * alias.c (find_base_value, find_base_term): Only process integer
9092 truncations. Check the precision rather than the size.
9094 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9095 Alan Hayward <alan.hayward@arm.com>
9096 David Sherwood <david.sherwood@arm.com>
9098 * machmode.h (is_narrower_int_mode): New function
9099 * optabs.c (expand_float, expand_fix): Use it.
9100 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9102 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9103 Alan Hayward <alan.hayward@arm.com>
9104 David Sherwood <david.sherwood@arm.com>
9106 * rtl.h (narrower_subreg_mode): New function.
9107 * ira-color.c (update_costs_from_allocno): Use it.
9109 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9110 Alan Hayward <alan.hayward@arm.com>
9111 David Sherwood <david.sherwood@arm.com>
9113 * optabs-query.h (convert_optab_p): New function, split out from...
9114 (convert_optab_handler): ...here.
9115 (widening_optab_handler): Delete.
9116 (find_widening_optab_handler): Remove permit_non_widening parameter.
9117 (find_widening_optab_handler_and_mode): Likewise. Provide an
9118 override that operates on mode class wrappers.
9119 * optabs-query.c (widening_optab_handler): Delete.
9120 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9121 parameter. Assert that the two modes are the same class and that
9122 the "from" mode is narrower than the "to" mode. Use
9123 convert_optab_handler instead of widening_optab_handler.
9124 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9125 instead of widening_optab_handler.
9126 * expr.c (expand_expr_real_2): Update calls to
9127 find_widening_optab_handler.
9128 * optabs.c (expand_widen_pattern_expr): Likewise.
9129 (expand_binop_directly): Take the insn_code as a parameter.
9130 (expand_binop): Only call find_widening_optab_handler for
9131 conversion optabs; use optab_handler otherwise. Update calls
9132 to find_widening_optab_handler and expand_binop_directly.
9133 Use convert_optab_handler instead of widening_optab_handler.
9134 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9135 find_widening_optab_handler and use scalar_mode rather than
9137 (convert_plusminus_to_widen): Likewise.
9139 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9140 Alan Hayward <alan.hayward@arm.com>
9141 David Sherwood <david.sherwood@arm.com>
9143 * machmode.h (fixed_size_mode): New class.
9144 * rtl.h (get_pool_mode): Return fixed_size_mode.
9145 * gengtype.c (main): Add fixed_size_mode.
9146 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9147 (get_raw_arg_mode): Likewise.
9148 * doc/tm.texi: Regenerate.
9149 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9150 * targhooks.c (default_get_reg_raw_mode): Likewise.
9151 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9152 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9153 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9154 (msp430_get_raw_result_mode): Likewise.
9155 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9156 * dbxout.c (dbxout_parms): Require fixed-size modes.
9157 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9158 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9159 * omp-low.c (lower_oacc_reductions): Likewise.
9160 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9161 (simplify_subreg): Update accordingly.
9162 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9163 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9164 that aren't fixed-size.
9165 (get_pool_mode): Return a fixed_size_mode.
9166 (output_constant_pool_2): Take a fixed_size_mode.
9168 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9169 Alan Hayward <alan.hayward@arm.com>
9170 David Sherwood <david.sherwood@arm.com>
9172 * doc/rtl.texi (vec_series): Document.
9173 (const): Say that the operand can be a vec_series.
9174 * rtl.def (VEC_SERIES): New rtx code.
9175 * rtl.h (const_vec_series_p_1): Declare.
9176 (const_vec_series_p): New function.
9177 * emit-rtl.h (gen_const_vec_series): Declare.
9178 (gen_vec_series): Likewise.
9179 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9180 (gen_vec_series): Likewise.
9181 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9182 * simplify-rtx.c (simplify_unary_operation): Handle negations
9184 (simplify_binary_operation_series): New function.
9185 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9186 (test_vector_ops_series): New function.
9187 (test_vector_ops): Call it.
9188 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9189 gen_const_vec_series.
9190 (altivec_lvsr): Likewise.
9191 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9193 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9194 Alan Hayward <alan.hayward@arm.com>
9195 David Sherwood <david.sherwood@arm.com>
9197 * doc/rtl.texi (const): Update description of address constants.
9198 Say that vector constants are allowed too.
9199 * common.md (E, F): Use CONSTANT_P instead of checking for
9201 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9202 checking for CONST_VECTOR.
9203 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9205 * expr.c (expand_expr_real_2): Check for vector modes instead
9206 of checking for CONST_VECTOR.
9207 * rtl.h (const_vec_p): New function.
9208 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9209 (unwrap_const_vec_duplicate): Handle them here too.
9211 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9212 David Malcolm <dmalcolm@redhat.com>
9213 Alan Hayward <alan.hayward@arm.com>
9214 David Sherwood <david.sherwood@arm.com>
9216 * rtl.h (vec_duplicate_p): New function.
9217 * selftest-rtl.c (assert_rtx_eq_at): New function.
9218 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9219 (assert_rtx_eq_at): Declare.
9220 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9221 * selftest-run-tests.c (selftest::run_tests): Call it.
9222 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9223 (simplify_unary_operation_1): Recursively handle vector duplicates.
9224 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9226 (simplify_subreg): Handle subregs of vector duplicates.
9227 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9228 (selftest::simplify_rtx_c_tests): New functions.
9230 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9231 Alan Hayward <alan.hayward@arm.com>
9232 David Sherwood <david.sherwood@arm.com>
9234 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9235 (gen_vec_duplicate): Likewise.
9236 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9238 (gen_const_vector): ...here.
9239 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9240 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9241 whose elements are all equal.
9242 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9243 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9244 (simplify_relational_operation): Likewise.
9245 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9247 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9248 (aarch64_expand_vector_init): Likewise.
9249 * config/arm/arm.c (neon_vdup_constant): Likewise.
9250 (neon_expand_vector_init): Likewise.
9251 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9252 (arm_block_set_unaligned_vect): Likewise.
9253 (arm_block_set_aligned_vect): Likewise.
9254 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9255 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9256 (expand_vec_perm_even_odd_pack): Likewise.
9257 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9258 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9259 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9260 gen_const_vec_duplicate.
9261 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9262 * config/mips/mips.c (mips_gen_const_int_vector): Use
9263 gen_const_vec_duplicate.
9264 (mips_expand_vector_init): Use CONST0_RTX.
9265 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9266 (define_split): Use gen_const_vec_duplicate.
9267 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9268 (define_split): Use gen_const_vec_duplicate.
9269 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9270 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9271 * config/spu/spu.c (spu_const): Likewise.
9273 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9274 Alan Hayward <alan.hayward@arm.com>
9275 David Sherwood <david.sherwood@arm.com>
9277 * combine.c (can_change_dest_mode): Reject changes in
9278 REGMODE_NATURAL_SIZE.
9280 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9282 * configure.ac (--enable-libssp): New.
9283 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9284 trying to determine target-specific default. Adjust indentation.
9285 * configure: Regenerated.
9286 * doc/install.texi (Configuration): Expand --disable-libssp
9289 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9291 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9294 2017-10-31 Martin Jambor <mjambor@suse.cz>
9297 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9299 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9301 * auto-profile.c (autofdo_source_profile::read): Use
9302 UNKNOWN_LOCATION rather than 0.
9303 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9304 (warning_at): ...this overload.
9305 (warning_at_rich_loc_n): Rename to...
9306 (warning_n): ...this overload.
9307 (error_at_rich_loc): Rename to...
9308 (error_at): ...this overload.
9309 (pedwarn_at_rich_loc): Rename to...
9310 (pedwarn): ...this overload.
9311 (permerror_at_rich_loc): Rename to...
9312 (permerror): ...this overload.
9313 (inform_at_rich_loc): Rename to...
9314 (inform): ...this overload.
9315 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9316 (diagnostic_n_impl_richloc): Rename to...
9317 (diagnostic_n_impl): ...this rich_location *-based decl.
9318 (inform_at_rich_loc): Rename to...
9319 (inform): ...this, and add an assertion.
9320 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9321 (warning_at_rich_loc): Rename to...
9322 (warning_at): ...this, and add an assertion.
9323 (warning_at_rich_loc_n): Rename to...
9324 (warning_n): ...this, and add an assertion.
9325 (warning_n): Update location_t-based implementation for removal of
9326 location_t-based diagnostic_n_impl.
9327 (pedwarn_at_rich_loc): Rename to...
9328 (pedwarn): ...this, and add an assertion.
9329 (permerror_at_rich_loc): Rename to...
9330 (permerror): ...this, and add an assertion.
9331 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9332 (error_at_rich_loc): Rename to...
9333 (error_at): ...this, and add an assertion.
9334 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9335 (driver::do_spec_on_infiles): Likewise.
9336 * substring-locations.c (format_warning_va): Update for renaming
9337 of inform_at_rich_loc.
9339 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9341 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9342 _Float<N>X built-in functions so that the variant without the
9343 "__builtin_" prefix is only enabled for the GNU C and Objective C
9344 languages when they are in non-strict ANSI/ISO mode.
9345 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9346 * target.def (floatn_builtin_p): Add a target hook to control
9347 whether _Float<N> and _Float<N>X built-in functions without the
9348 "__builtin_" prefix are enabled, and return true for C and
9349 Objective C in the default hook. Include langhooks.h in
9351 * targhooks.h (default_floatn_builtin_p): Likewise.
9352 * targhooks.c (default_floatn_builtin_p): Likewise.
9353 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9354 floatn_builtin_p target hook.
9355 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9357 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9358 Eric Botcazou <ebotcazou@adacore.com>
9360 PR rtl-optimization/81803
9361 * lra-constraints.c (curr_insn_transform): Also reload the whole
9362 register for a strict subreg no wider than a word if this is for
9363 a WORD_REGISTER_OPERATIONS target.
9365 2017-10-31 Jason Merrill <jason@redhat.com>
9367 * gdbinit.in: Skip over inlines from timevar.h.
9369 2017-10-31 Martin Liska <mliska@suse.cz>
9371 * doc/gcov.texi: Document new option.
9372 * gcov.c (print_usage): Likewise print it.
9373 (process_args): Support the argument.
9374 (format_count): New function.
9375 (format_gcov): Use the function.
9377 2017-10-31 Martin Liska <mliska@suse.cz>
9379 * gcov.c (struct name_map): do not use typedef.
9380 Define operator== and operator<.
9381 (name_search): Remove.
9382 (name_sort): Remove.
9383 (main): Do not allocate names.
9384 (process_file): Add vertical space.
9385 (generate_results): Use std::find.
9386 (release_structures): Do not release memory.
9387 (find_source): Use std::find.
9389 2017-10-31 Martin Liska <mliska@suse.cz>
9391 * gcov.c (struct line_info): Remove it's typedef.
9392 (line_info::line_info): Add proper ctor.
9393 (line_info::has_block): Do not use a typedef.
9394 (struct source_info): Do not use typedef.
9395 (circuit): Likewise.
9396 (get_cycles_count): Likewise.
9397 (output_intermediate_file): Iterate via vector iterator.
9398 (add_line_counts): Use std::vector methods.
9399 (accumulate_line_counts): Likewise.
9400 (output_lines): Likewise.
9402 2017-10-31 Martin Liska <mliska@suse.cz>
9404 * gcov.c (struct source_info): Remove typedef.
9405 (source_info::source_info): Add proper ctor.
9406 (accumulate_line_counts): Use struct, not it's typedef.
9407 (output_gcov_file): Likewise.
9408 (output_lines): Likewise.
9409 (main): Do not allocate an array.
9410 (output_intermediate_file): Use size of vector container.
9411 (process_file): Resize the vector.
9412 (generate_results): Do not preallocate, use newly added vector
9414 (release_structures): Do not release sources.
9415 (find_source): Use vector methods.
9416 (add_line_counts): Do not use typedef.
9418 2017-10-31 Martin Liska <mliska@suse.cz>
9420 * doc/gcov.texi: Document that.
9421 * gcov.c (add_line_counts): Mark lines with a non-executed
9423 (output_line_beginning): Handle such lines.
9424 (output_lines): Pass new argument.
9425 (output_intermediate_file): Print it in intermediate format.
9427 2017-10-31 Martin Liska <mliska@suse.cz>
9429 * color-macros.h: New file.
9430 * diagnostic-color.c: Factor out color related to macros to
9432 * doc/gcov.texi: Document -k option.
9433 * gcov.c (INCLUDE_STRING): Include string.h.
9434 (print_usage): Add -k option.
9435 (process_args): Parse it.
9436 (pad_count_string): New function.
9437 (output_line_beginning): Likewise.
9438 (DEFAULT_LINE_START): New macro.
9439 (output_lines): Support color output.
9441 2017-10-31 Martin Liska <mliska@suse.cz>
9443 PR gcov-profile/82633
9444 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9445 their interaction with GCOV infrastructure.
9446 * configure.ac: Add -fkeep-{inline,static}-functions to
9448 * configure: Regenerate.
9450 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9453 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9455 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9458 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9459 into a register if it does not fit into an immediate offset field.
9461 2017-10-31 Olivier Hainque <hainque@adacore.com>
9463 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9465 2017-10-31 Julia Koval <julia.koval@intel.com>
9467 * config.gcc: Add gfniintrin.h.
9468 * config/i386/gfniintrin.h: New.
9469 * config/i386/i386-builtin-types.def
9470 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9471 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9472 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9473 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9474 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9475 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9476 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9477 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9478 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9479 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9480 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9481 * config/i386/immintrin.h: Include gfniintrin.h.
9482 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9484 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9486 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9488 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9490 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9491 (arm_ashldi3_1bit): Remove pattern.
9492 (ashrdi3): Remove shift by 1 expansion.
9493 (arm_ashrdi3_1bit): Remove pattern.
9494 (lshrdi3): Remove shift by 1 expansion.
9495 (arm_lshrdi3_1bit): Remove pattern.
9496 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9497 cost of ashldi3 by 1.
9498 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9499 (<shift>di3_neon): Likewise.
9501 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9503 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9504 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9505 and (*aarch64_simd_mov<VQ:mode>).
9506 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9507 pattern alternative.
9508 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9509 attributes to match pattern alternative.
9511 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9513 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9514 * config/rs6000/emmintrin.h: New file.
9515 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9517 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9519 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9520 * (movdi_vfp_cortexa8): Remove pattern.
9522 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9524 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9526 (Specific, alpha*-dec-osf5.1): Remove.
9527 (Specific, mips-sgi-irix5): Remove.
9528 (Specific, mips-sgi-irix6): Remove.
9530 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9533 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9534 arguments to clear_bit_region_be.
9536 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9538 * gimplify.c: Include memmodel.h.
9540 2017-10-30 Martin Jambor <mjambor@suse.cz>
9542 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9543 condition whether loop should be executed at all.
9545 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9547 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9548 gimple folding of vec_madd() intrinsics.
9549 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9550 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9551 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9553 2017-10-30 Richard Biener <rguenther@suse.de>
9555 PR tree-optimization/82762
9557 2017-10-23 Richard Biener <rguenther@suse.de>
9559 PR tree-optimization/82129
9561 2017-08-01 Richard Biener <rguenther@suse.de>
9563 PR tree-optimization/81181
9564 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9565 (compute_antic): ... end of iteration here.
9567 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9569 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9570 -std=iso9899:2017 and -std=gnu17.
9571 * doc/standards.texi (C Language): Document C17 support.
9572 * doc/cpp.texi (Overview): Mention -std=c17.
9573 (Standard Predefined Macros): Document C11 and C17 values of
9574 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9575 * doc/extend.texi (Inline): Do not list individual options for
9576 standards newer than C99.
9577 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9579 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9582 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9584 * asan.c (asan_finish_file): Align asan globals array by shadow
9587 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9590 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9591 (struct store_immediate_info): Add bitregion_start and bitregion_end
9593 (store_immediate_info::store_immediate_info): Add brs and bre
9594 arguments and initialize bitregion_{start,end} from those.
9595 (struct merged_store_group): Add bitregion_start, bitregion_end,
9596 align_base and mask fields. Drop unnecessary struct keyword from
9597 struct store_immediate_info. Add do_merge method.
9598 (clear_bit_region_be): Use memset instead of loop storing zeros.
9599 (merged_store_group::do_merge): New method.
9600 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9601 stores as long as the surrounding bitregions have no gaps.
9602 (merged_store_group::merge_overlapping): Use do_merge.
9603 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9604 is byte aligned, rather than requiring that start and width are
9605 byte aligned. Drop unnecessary struct keyword from
9606 struct store_immediate_info. Allocate and populate also mask array.
9607 Make start of the arrays relative to bitregion_start rather than
9608 start and size them according to bitregion_{end,start} difference.
9609 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9610 struct store_immediate_info.
9611 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9612 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9613 struct keyword from struct store_immediate_info.
9614 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9615 between stores as long as the surrounding bitregions have no gaps.
9617 (struct split_store): Add orig non-static data member.
9618 (split_store::split_store): Initialize orig to false.
9619 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9620 if there is exactly a single original stmt. Change stmts argument
9621 to pointer from reference, if NULL, don't push anything to it. Add
9622 first argument, use it to optimize skipping over orig stmts that
9623 are known to be before bitpos already. Simplify.
9624 (split_group): Return unsigned int count how many stores are or
9625 would be needed rather than a bool. Add allow_unaligned argument.
9626 Change split_stores argument from reference to pointer, if NULL,
9627 only do a dry run computing how many stores would be produced.
9628 Rewritten algorithm to use both alignment and misalign if
9629 !allow_unaligned and handle bitfield stores with gaps.
9630 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9631 from bitregion_start instead of start. Compute allow_unaligned
9632 here, if true, do 2 split_group dry runs to compute which one
9633 produces fewer stores and prefer aligned if equal. Punt if
9634 new count is bigger or equal than original before emitting any
9635 statements, rather than during that. Remove no longer needed
9636 new_ssa_names tracking. Replace num_stmts with
9637 split_stores.length (). Use 32-bit stack allocated entries
9638 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9639 if possible. Handle bitfields with gaps.
9640 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9641 Compute bitregion_{start,end} for the stores and construct
9642 store_immediate_info with that. Formatting fixes.
9644 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9647 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9648 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9650 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9652 * gimplify.c: Include tm_p.h.
9654 * common.opt (gcoff): Re-add as ignored option.
9655 (gcoff1, gcoff2, gcoff3): Likewise.
9657 * Makefile.in (OBJS): Delete sdbout.o.
9658 (GTFILES): Delete $(srcdir)/sdbout.c.
9659 * debug.h: Delete sdb_debug_hooks.
9660 * final.c: Delete sdbout.h include.
9661 (final_scan_insn): Delete SDB_DEBUG check.
9662 (rest_of_clean_state): Likewise.
9663 * output.h: Delete sdb_begin_function_line.
9666 * toplev.c: Delete sdbout.h include.
9667 (process_options): Delete SDB_DEBUG check.
9668 * tree-core.h (tree_type_common): Delete pointer field of
9670 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9671 TYPE_SYMTAB_POINTER.
9672 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9673 (TYPE_SYMTAB_IS_POINTER): Delete.
9674 (TYPE_SYMTAB_IS_DIE): Renumber.
9675 * xcoffout.c: Refer to former sdbout.c file.
9676 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9678 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9679 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9680 (Debugging Options): Delete -gcoff.
9681 (-gstabs): Delete SDB reference.
9683 (-gcoff@var{level}): Delete.
9684 * doc/passes.texi (Debugging information output): Delete SDB and
9685 sdbout.c references.
9686 * doc/tm.texi: Regenerate.
9687 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9688 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9690 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9691 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9692 (SDB_DEBUGGING_INFO): Delete.
9693 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9694 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9695 * target.def (output_source_filename): Delete COFF reference.
9697 * common.opt (gcoff): Delete.
9698 (gxcoff+): Update Negative chain.
9699 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9701 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9702 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9703 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9705 (expand_function_start): Change sdb reference to past tense.
9706 (expand_function_end): Change sdb reference to past tense.
9707 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9708 * opts.c (debug_type_names): Delete coff entry.
9709 (common_handle_option): Delete OPT_gcoff case.
9710 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9712 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9713 * config/cris/cris.h: Delete SDB reference in comment.
9714 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9715 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9716 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9717 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9719 (ix86_expand_prologue): Likewise.
9720 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9721 * config/ia64/ia64.h: Likewise.
9722 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9723 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9725 * config/mmix/mmix.h: Likewise.
9726 * config/nds32/nds32.c: Likewise.
9727 * config/stormy/storym16.h: Likewise.
9728 * config/visium/visium.h: Likewise.
9729 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9731 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9733 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9734 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9735 FRAME_POINTER_REGNUM point at high end of local var area.
9737 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9739 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9740 Move comment around. Do not reset best_edge for a copiable
9741 destination if the copy would cause a partition change.
9742 (better_edge_p): Remove redundant check.
9744 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9746 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9748 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9750 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9751 for math functions that have _Float<N> and _Float<N>X variants.
9752 (mathfn_built_in_2): Add support for math functions that have
9753 _Float<N> and _Float<N>X variants.
9754 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9755 (expand_builtin_mathfn_ternary): Add support for fma with
9756 _Float<N> and _Float<N>X variants.
9757 (expand_builtin): Likewise.
9758 (fold_builtin_3): Likewise.
9759 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9760 create math function _Float<N> and _Float<N>X variants as external
9762 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9763 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9764 the __builtin_ prefix and if not strict ansi, without the prefix.
9765 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9766 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9767 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9768 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9769 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9770 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9771 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9772 function signatures for fma _Float<N> and _Float<N>X variants.
9773 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9774 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9775 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9776 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9777 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9778 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9779 * gencfn-macros.c (print_case_cfn): Add support for math functions
9780 that have _Float<N> and _Float<N>X variants.
9781 (print_define_operator_list): Likewise.
9782 (fltfn_suffixes): Likewise.
9784 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9785 for math functions that have _Float<N> and _Float<N>X variants.
9786 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9787 and _Float<N>X variants.
9788 (COPYSIGN): Likewise.
9791 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9792 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9794 (integer_valued_read_call_p): Likewise.
9795 * fold-const-call.c (fold_const_call_ss): Likewise.
9796 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9797 _Float<N> and _Float<N>X variants.
9798 (fold_const_call_ssss): Add support for fma _Float<N> and
9799 _Float<N>X variants.
9800 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9801 support for copysign and fma _Float<N> and _Float<N>X variants.
9802 (backprop::process_builtin_call_use): Likewise.
9803 * tree-call-cdce.c (can_test_argument_range); Add support for
9804 sqrt _Float<N> and _Float<N>X variants.
9805 (edom_only_function): Likewise.
9806 (get_no_error_domain): Likewise.
9807 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9808 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9809 copysign _Float<N> and _Float<N>X variants.
9810 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9811 handled by machine independent code.
9812 (FMAF128): Likewise.
9813 * doc/cpp.texi (Common Predefined Macros): Document defining
9814 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9815 fma _Float<N> and _Float<N>X variants.
9817 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9820 * config/i386/i386-modes.def (CCFPU): Remove definition.
9821 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9822 (ix86_cc_modes_compatible): Ditto.
9823 (ix86_expand_carry_flag_compare): Ditto.
9824 (ix86_expand_int_movcc): Ditto.
9825 (ix86_expand_int_addcc): Ditto.
9826 (ix86_reverse_condition): Ditto.
9827 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9828 Return true/false for unordered/ordered fp comparisons.
9829 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9830 (ix86_prepare_fp_compare_args): Update for rename.
9831 (ix86_expand_fp_compare): Update for rename. Generate unordered
9832 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9833 (ix86_expand_sse_compare_and_jump): Ditto.
9834 * config/i386/predicates.md (fcmov_comparison_operator):
9835 Remove CCFPU mode handling.
9836 (ix86_comparison_operator): Ditto.
9837 (ix86_carry_flag_operator): Ditto.
9838 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9839 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9840 (*cmpu<mode>_cc_i387): Ditto.
9841 (FPCMP): Remove mode iterator.
9842 (unord): Remove mode attribute.
9843 (unord_subst): New define_subst transformation
9844 (unord): New define_subst attribute.
9846 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9847 (*cmpi<unord>xf_i387): Ditto.
9848 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9849 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9850 using unord_subst transformation.
9851 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9852 (round_saeonly): Also handle CCFP mode.
9853 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9854 Remove UNSPEC_SAHF unspec handling.
9856 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9858 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9860 2017-10-27 Jeff Law <law@redhat.com>
9862 * gimple-ssa-sprintf.c: Include domwalk.h.
9863 (class sprintf_dom_walker): New class, derived from dom_walker.
9864 (sprintf_dom_walker::before_dom_children): New function.
9865 (struct call_info): Moved into sprintf_dom_walker class
9866 (compute_formath_length, handle_gimple_call): Likewise.
9867 (sprintf_length::execute): Call the dominator walker rather
9868 than walking the statements.
9870 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9871 gimple statement locations.
9872 (check_array_bounds): Corresponding changes. Get the statement's
9873 location directly from wi->stmt.
9875 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9878 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9880 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9882 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9883 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9885 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9888 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9889 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9890 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9891 maybe_get_pool_constant.
9892 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9895 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9897 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9898 2.26 caveat. Update gas and gld versions.
9899 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9902 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9904 * cgraph.h (set_malloc_flag): Declare.
9905 * cgraph.c (set_malloc_flag_1): New function.
9906 (set_malloc_flag): Likewise.
9907 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9908 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9910 (read_ipa_call_summary): Add support for reading is_return_callee.
9911 (write_ipa_call_summary): Stream is_return_callee.
9912 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9913 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9914 ipa-prop.h, ipa-fnsummary.h.
9915 (pure_const_names): Change to static.
9916 (malloc_state_e): Define.
9917 (malloc_state_names): Define.
9918 (funct_state_d): Add field malloc_state.
9919 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9920 (check_retval_uses): New function.
9921 (malloc_candidate_p): Likewise.
9922 (analyze_function): Add support for malloc attribute.
9923 (pure_const_write_summary): Stream malloc_state.
9924 (pure_const_read_summary): Add support for reading malloc_state.
9925 (dump_malloc_lattice): New function.
9926 (propagate_malloc): New function.
9927 (warn_function_malloc): New function.
9928 (ipa_pure_const::execute): Call propagate_malloc and
9929 ipa_free_fn_summary.
9930 (pass_local_pure_const::execute): Add support for malloc attribute.
9931 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9932 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9934 2017-10-27 Martin Liska <mliska@suse.cz>
9936 PR gcov-profile/82457
9937 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9938 for fork-like functions to be properly instrumented.
9940 2017-10-27 Richard Biener <rguenther@suse.de>
9943 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9944 info when we redirected EH.
9946 2017-10-26 Michael Collison <michael.collison@arm.com>
9948 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9950 (<optab>_trunchf<GPI:mode>2: New pattern.
9951 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9952 * config/aarch64/iterators.md (wv): New mode attribute.
9953 (vf, VF): New mode attributes.
9954 (vgp, VGP): New mode attributes.
9955 (s): Update attribute with SImode and DImode prefixes.
9957 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9959 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9960 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9961 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9962 (nios2_option_overide): Initialize it. Don't allow R0-relative
9963 addressing with PIC.
9964 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9965 (nios2_symbolic_constant_p): Likewise.
9966 (nios2_legitimate_address_p): Likewise.
9967 (nios2_r0rel_section_name_p): New.
9968 (nios2_symbol_ref_in_r0rel_data_p): New.
9969 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9970 (r0rel_constant_p): New.
9971 (nios2_print_operand_address): Handle r0rel_constant_p.
9972 (nios2_cdx_narrow_form_p): Likewise.
9973 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9974 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9975 (Nios II Options): Document -mr0rel-sec.
9977 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9979 * config/nios2/nios2.c: Include xregex.h.
9980 (nios2_gprel_sec_regex): New.
9981 (nios2_option_overide): Initialize it. Don't allow GP-relative
9982 addressing with PIC.
9983 (nios2_small_section_name_p): Check for regex match.
9984 * config/nios2/nios2.opt (mgprel-sec=): New option.
9985 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9986 (Nios II Options): Document -mgprel-sec.
9988 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9990 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9992 2017-10-26 Tom de Vries <tom@codesourcery.com>
9994 PR tree-optimization/82707
9995 * gimple.c (gimple_copy): Fix unsharing of
9996 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9998 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10000 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10001 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10002 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10003 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10004 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10005 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10006 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10007 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10008 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10009 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10010 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10011 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10012 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10013 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10014 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10015 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10016 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10018 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10020 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10022 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10023 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10024 warning to rs6000.c. Remove the Undocumented flag, since it has
10026 (-mabi=ibmlongdouble): Likewise.
10027 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10028 already set, set the default format for long double.
10029 (rs6000_debug_reg_global): Print whether long double is IBM or
10031 (rs6000_option_override_internal): Rework setting long double
10032 format. Only warn if the user is changing the long double default
10033 and they did not use -Wno-psabi.
10034 * doc/invoke.texi (PowerPC options): Update the documentation for
10035 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10037 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10038 Alan Hayward <alan.hayward@arm.com>
10039 David Sherwood <david.sherwood@arm.com>
10041 * rtl.h (wider_subreg_mode): New function.
10042 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10043 rather than an unsigned int *.
10044 * ira-color.c (regno_max_ref_width): Replace with...
10045 (regno_max_ref_mode): ...this new variable.
10046 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10047 Use wider_subreg_mode.
10048 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10049 rather than an unsigned int *.
10050 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10051 (process_alt_operands): Likewise.
10052 (invariant_p): Likewise.
10053 * lra-spills.c (assign_mem_slot): Likewise.
10054 (add_pseudo_to_slot): Likewise.
10055 * lra.c (collect_non_operand_hard_regs): Likewise.
10056 (add_regs_to_insn_regno_info): Likewise.
10057 * reload1.c (regno_max_ref_width): Replace with...
10058 (regno_max_ref_mode): ...this new variable.
10059 (reload): Update accordingly. Update call to
10060 ira_sort_regnos_for_alter_reg.
10061 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10062 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10063 (scan_paradoxical_subregs): Likewise.
10065 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10067 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10068 (aarch64_frame): Add emit_frame_chain boolean.
10069 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10070 Move eh_return case to aarch64_layout_frame.
10071 (aarch64_layout_frame): Initialize emit_frame_chain.
10072 (aarch64_expand_prologue): Use emit_frame_chain.
10074 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10076 * config/aarch64/aarch64.c (aarch64_layout_frame):
10077 Ensure LR is always stored at the bottom of the callee-saves.
10078 Remove rarely used frame layout which saves callee-saves at top of
10079 frame, so the store of LR can be used as a valid probe in all cases.
10081 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10083 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10084 Improve unaligned TImode/TFmode base/offset split.
10086 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10087 Alan Hayward <alan.hayward@arm.com>
10088 David Sherwood <david.sherwood@arm.com>
10090 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10091 * combine.c (find_single_use_1): Likewise.
10092 (expand_field_assignment): Likewise.
10093 (move_deaths): Likewise.
10094 * lra-constraints.c (simplify_operand_subreg): Likewise.
10095 (curr_insn_transform): Likewise.
10096 * lra.c (collect_non_operand_hard_regs): Likewise.
10097 (add_regs_to_insn_regno_info): Likewise.
10098 * rtlanal.c (reg_referenced_p): Likewise.
10099 (covers_regno_no_parallel_p): Likewise.
10101 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10103 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10104 Don't print any bits outside the precision of the value.
10105 * wide-int.cc (test_printing): Add some new tests.
10107 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10109 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10110 supports -xbrace_comment option.
10111 * configure: Regenerate.
10112 * config.in: Regenerate.
10113 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10114 (ASM_CPU_SPEC): Use it.
10116 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10118 * target.def (static_rtx_alignment): New hook.
10119 * targhooks.h (default_static_rtx_alignment): Declare.
10120 * targhooks.c (default_static_rtx_alignment): New function.
10121 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10122 * doc/tm.texi: Regenerate.
10123 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10124 instead of targetm.constant_alignment. Remove call to
10125 set_mem_attributes.
10126 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10127 (cris_preferred_mininum_alignment): New function, split out from...
10128 (cris_constant_alignment): ...here.
10129 (cris_static_rtx_alignment): New function.
10130 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10132 (ix86_constant_alignment): ...here.
10133 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10134 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10135 (mmix_static_rtx_alignment): New function.
10136 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10137 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10139 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10142 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10143 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10145 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10147 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10148 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10149 mark as a sign-extending load.
10150 (local_pic_load_u): Define.
10152 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10154 PR middle-end/82062
10155 * fold-const.c (operand_equal_for_comparison_p): Also return true
10156 if ARG0 is a simple variant of ARG1 with narrower precision.
10157 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10159 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10161 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10163 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10164 scatter_static, scatter_per_elt.
10165 * x86-tune-costs.h: Add new cost entries.
10167 2017-10-25 Richard Biener <rguenther@suse.de>
10169 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10170 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10171 class pass_fre): Move to ...
10172 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10173 class pass_fre): ... here and adjust for statistics.
10175 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10178 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10179 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10180 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10181 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10184 2017-10-25 Richard Biener <rguenther@suse.de>
10186 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10187 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10188 eliminate_push_avail, eliminate_insert): Move inside...
10189 (class eliminate_dom_walker): ... this class in preparation
10191 (fini_eliminate): Remove by merging with ...
10192 (eliminate): ... this function. Adjust for class changes.
10193 (pass_pre::execute): Remove fini_eliminate call.
10194 (pass_fre::execute): Likewise.
10196 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10199 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10200 (VPERMI2, VPERMI2I): New mode iterators.
10201 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10202 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10204 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10205 mode iterator. Remove 3 old define_insn patterns.
10206 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10207 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10208 VPERMI2 mode iterator, remove the other two expanders.
10209 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10210 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10211 instructions, remove the other two patterns.
10212 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10213 mode iterator, remove the other two patterns.
10214 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10215 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10216 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10217 and adjust argument order accordingly.
10218 (ix86_expand_vec_perm): Adjust caller.
10219 (expand_vec_perm_1): Likewise.
10220 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10221 (expand_vec_perm_vpermt2_vpshub2): ... this.
10222 (ix86_expand_vec_perm_const_1): Adjust caller.
10223 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10226 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10227 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10228 (vec_shl_<mode>): Remove unused expander.
10229 (avx512bw_<shift_insn><mode>3): New define_insn.
10230 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10231 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10233 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10236 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10239 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10241 PR rtl-optimization/82396
10242 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10243 (autopref_multipass_init): Simplify initialization.
10244 (autopref_rank_data): Simplify sort order.
10245 * gcc/sched-int.h (autopref_multipass_data_): Remove
10246 multi_mem_insn_p, min_offset and max_offset.
10248 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10250 PR middle-end/60580
10251 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10252 Check special value of flag_omit_frame_pointer.
10253 (aarch64_can_eliminate): Likewise.
10254 (aarch64_override_options_after_change_1): Simplify handling of
10255 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10257 2017-10-24 Richard Biener <rguenther@suse.de>
10259 PR tree-optimization/82697
10260 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10261 zero for conditional load and unconditional store.
10263 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10265 * doc/install.texi: Document bootstrap-cet.
10267 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10270 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10271 ENDBR instruction at function entrance if function is only
10274 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10277 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10278 patterns to better describe from which operation the CF is computed.
10279 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10280 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10281 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10282 is 0, use _0 suffixed expanders instead of emitting a comparison
10285 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10287 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10288 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10290 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10292 PR middle-end/82569
10293 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10294 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10295 * loop-iv.c (iv_get_reaching_def): Likewise.
10296 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10297 variable is promoted and the partition contains undefined values.
10299 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10301 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10303 (nios2_address_cost): Define.
10304 (nios2_legitimize_address): Recognize (exp + constant) directly.
10305 (TARGET_ADDRESS_COST): Define.
10307 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10309 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10310 (nios2_symbolic_memory_operand_p): Declare.
10311 (nios2_split_large_constant): Declare.
10312 (nios2_split_symbolic_memory_operand): Declare.
10313 * config/nios2/nios2.c: Adjust includes.
10314 (nios2_symbolic_constant_allowed): New.
10315 (nios2_symbolic_constant_p): New.
10316 (nios2_plus_symbolic_constant_p): New.
10317 (nios2_valid_addr_expr_p): Recognize addresses involving
10318 symbolic constants.
10319 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10320 (nios2_symbolic_memory_operand_p): New.
10321 (nios2_large_constant_p): New.
10322 (nios2_split_large_constant): New.
10323 (nios2_split_plus_large_constant): New.
10324 (nios2_split_symbolic_memory_operand): New.
10325 (nios2_legitimize_address): Code refactoring. Handle addresses
10326 involving symbolic constants.
10327 (nios2_emit_move_sequence): Likewise.
10328 (nios2_print_operand): Improve error output.
10329 (nios2_print_operand_address): Handle LO_SUM.
10330 (nios2_cdx_narrow_form_p): Likewise.
10331 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10332 operands involving symbolic constants.
10333 (movhi_internal, movsi_internal): Likewise.
10334 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10335 (extendhisi2, extendqi<mode>2): Likewise.
10337 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10339 * tree-pass.h (PROP_rtl_split_insns): Define.
10340 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10342 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10344 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10346 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10349 * target.def (const_not_ok_for_debug_p): Default to
10350 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10351 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10352 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10353 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10354 which targetm.const_not_ok_for_debug_p returned true.
10355 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10357 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10359 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10360 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10361 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10363 (ix86_const_not_ok_for_debug_p): New function.
10364 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10365 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10367 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10370 * system.h: Conditionally include "unique-ptr.h" if
10371 INCLUDE_UNIQUE_PTR is defined.
10372 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10373 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10375 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10377 * config/rl78/rl78.md: New define_expand "subdi3".
10379 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10382 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10383 DF_REF_INSN if DF_REF_INSN_INFO is false.
10385 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10387 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10388 xmm_move instead of sse_move.
10389 (sse_store_index): New function.
10390 (ix86_register_move_cost): Be more sensible about mismatch stall;
10391 model AVX moves correctly; make difference between sse->integer and
10393 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10394 moves; make difference between SSE and AVX.
10395 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10396 and zmm_move. Increase size of sse load and store tables;
10397 add unaligned load and store tables; add ssemmx_to_integer.
10398 * x86-tune-costs.h: Update all entries according to real
10399 move latencies from Agner Fog's manual and chip documentation.
10401 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10404 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10405 * config/i386/constraints.md (Wf): New constraint.
10406 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10407 (cmp<dwi>_doubleword): Removed.
10408 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10409 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10410 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10411 expand with cmp<dwi>_doubleword. For LTU and GEU use
10412 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10414 * common.opt (gcolumn-info): Enable by default.
10415 * doc/invoke.texi (gcolumn-info): Document new default.
10417 2017-10-23 Richard Biener <rguenther@suse.de>
10419 PR tree-optimization/82672
10420 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10421 Fold the stmt if we propagated into it.
10423 2017-10-23 Richard Biener <rguenther@suse.de>
10425 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10426 (bitmap_remove_expr_from_set): ... this. All callers call this
10427 for non-constant values.
10428 (bitmap_set_subtract): Rename to...
10429 (bitmap_set_subtract_expressions): ... this. Adjust and
10431 (bitmap_set_contains_value): Remove superfluous check.
10432 (bitmap_set_replace_value): Inline into single caller ...
10433 (bitmap_value_replace_in_set): ... here and simplify.
10434 (dependent_clean): Merge into ...
10435 (clean): ... this using an overload. Adjust.
10436 (prune_clobbered_mems): Adjust.
10437 (compute_antic_aux): Likewise.
10438 (compute_partial_antic_aux): Likewise.
10440 2017-10-23 Richard Biener <rguenther@suse.de>
10442 PR tree-optimization/82129
10444 2017-08-01 Richard Biener <rguenther@suse.de>
10446 PR tree-optimization/81181
10447 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10448 (compute_antic): ... end of iteration here.
10450 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10452 * target.def (starting_frame_offset): New hook.
10453 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10454 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10455 * doc/tm.texi.in: Regenerate.
10456 * hooks.h (hook_hwi_void_0): Declare.
10457 * hooks.c (hook_hwi_void_0): New function.
10458 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10459 STARTING_FRAME_OFFSET.
10460 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10461 * reload1.c (reload): Likewise.
10462 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10463 instead of STARTING_FRAME_OFFSET.
10464 * function.c (try_fit_stack_local): Likewise.
10465 (assign_stack_local_1): Likewise
10466 (instantiate_virtual_regs): Likewise.
10467 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10468 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10469 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10470 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10471 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10472 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10473 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10474 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10475 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10476 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10477 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10478 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10479 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10480 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10481 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10482 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10483 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10484 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10485 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10486 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10487 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10488 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10489 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10490 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10491 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10492 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10493 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10494 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10495 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10496 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10497 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10498 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10499 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10500 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10501 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10502 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10503 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10504 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10505 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10506 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10507 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10508 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10509 return a HOST_WIDE_INT.
10510 (avr_builtin_setjmp_frame_value): Use it instead of
10511 STARTING_FRAME_OFFSET.
10512 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10513 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10514 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10516 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10517 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10518 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10519 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10520 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10521 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10522 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10523 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10524 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10525 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10526 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10527 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10529 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10530 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10531 * config/mips/mips.c (mips_compute_frame_info): Refer to
10532 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10533 (mips_starting_frame_offset): New function.
10534 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10535 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10536 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10537 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10538 and return a HOST_WIDE_INT.
10539 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10540 (mmix_initial_elimination_offset): Refer to
10541 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10542 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10543 * config/pa/pa.c (pa_starting_frame_offset): New function.
10544 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10545 (pa_expand_prologue): Likewise.
10546 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10547 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10548 !FRAME_GROWS_DOWNWARD handling to...
10549 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10550 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10551 !FRAME_GROWS_DOWNWARD handling to...
10552 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10553 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10554 !FRAME_GROWS_DOWNWARD handling to...
10555 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10556 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10558 (rs6000_starting_frame_offset): New function.
10559 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10560 !FRAME_GROWS_DOWNWARD handling to...
10561 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10562 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10563 !FRAME_GROWS_DOWNWARD handling to...
10564 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10565 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10566 !FRAME_GROWS_DOWNWARD handling to...
10567 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10568 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10569 (rs6000_starting_frame_offset): New function.
10570 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10571 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10572 * config/vax/vax.c (vax_starting_frame_offset): New function.
10573 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10574 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10575 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10576 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10577 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10578 * system.h (STARTING_FRAME_OFFSET): Poison.
10580 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10582 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10583 SCALAR_TYPE_MODE instead of TYPE_MODE.
10585 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10586 Alan Hayward <alan.hayward@arm.com>
10587 David Sherwood <david.sherwood@arm.com>
10589 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10591 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10592 Alan Hayward <alan.hayward@arm.com>
10593 David Sherwood <david.sherwood@arm.com>
10595 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10597 2017-10-23 Richard Biener <rguenther@suse.de>
10599 PR tree-optimization/82129
10600 * tree-ssa-pre.c (bitmap_set_and): Remove.
10601 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10602 canonicalizing expressions in the set to those with lowest
10603 ID rather than taking that from the first edge.
10605 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10607 * combine.c (rtx_equal_for_field_assignment_p): Use
10608 byte_lowpart_offset.
10610 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10611 Alan Hayward <alan.hayward@arm.com>
10612 David Sherwood <david.sherwood@arm.com>
10614 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10615 to a SUBREG_PROMOTED_VAR.
10617 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10618 Alan Hayward <alan.hayward@arm.com>
10619 David Sherwood <david.sherwood@arm.com>
10621 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10622 (expand_debug_source_expr): Likewise.
10623 * combine.c (combine_simplify_rtx): Likewise.
10624 * cse.c (fold_rtx): Likewise.
10625 * optabs.c (expand_float): Likewise.
10626 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10627 (simplify_binary_operation_1): Likewise.
10629 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10630 Alan Hayward <alan.hayward@arm.com>
10631 David Sherwood <david.sherwood@arm.com>
10633 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10634 (record_promoted_value): Likewise.
10635 * expr.c (expand_expr_real_2): Likewise.
10636 * ree.c (update_reg_equal_equiv_notes): Likewise.
10637 (combine_set_extension): Likewise.
10638 * rtlanal.c (low_bitmask_len): Likewise.
10639 * simplify-rtx.c (neg_const_int): Likewise.
10640 (simplify_binary_operation_1): Likewise.
10642 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10643 Alan Hayward <alan.hayward@arm.com>
10644 David Sherwood <david.sherwood@arm.com>
10646 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10647 * regcprop.c (maybe_mode_change): Likewise.
10648 * reload1.c (alter_reg): Likewise.
10650 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10652 * inchash.h (inchash::hash::add_wide_int): New function.
10653 * lto-streamer-out.c (hash_tree): Use it.
10655 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10657 * inchash.h (inchash::hash::add_wide_int): Rename to...
10658 (inchash::hash::add_hwi): ...this.
10659 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10660 (polymorphic_call_target_hasher::hash): Likewise.
10661 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10662 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10663 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10664 * lto-streamer-out.c (hash_tree): Likewise.
10665 * optc-save-gen.awk: Likewise.
10666 * tree.c (add_expr): Likewise.
10668 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10671 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10672 for ordered inequality comparisons even with TARGET_IEEE_FP.
10674 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10677 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10678 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10679 Expand with cmp<dwi>_doubleword.
10681 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10683 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10684 List CET intrinsics.
10685 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10686 specific to -fcf-protection option.
10688 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10690 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10691 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10692 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10693 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10694 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10695 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10696 (extra_objs): Add cet.o for Linux/x86 targets.
10697 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10698 * config/i386/cet.c: New file.
10699 * config/i386/cetintrin.h: Likewise.
10700 * config/i386/t-cet: Likewise.
10701 * config/i386/cpuid.h (bit_SHSTK): New.
10702 (bit_IBT): Likewise.
10703 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10704 pass IBT and SHSTK bits.
10705 * config/i386/i386-builtin-types.def
10706 (VOID_FTYPE_UNSIGNED_PVOID): New.
10707 (VOID_FTYPE_UINT64_PVOID): Likewise.
10708 * config/i386/i386-builtin.def: Add CET intrinsics.
10709 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10710 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10711 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10712 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10714 * config/i386/i386.c (rest_of_insert_endbranch): New.
10715 (pass_data_insert_endbranch): Likewise.
10716 (pass_insert_endbranch): Likewise.
10717 (make_pass_insert_endbranch): Likewise.
10718 (ix86_notrack_prefixed_insn_p): Likewise.
10719 (ix86_target_string): Add -mibt, -mshstk flags.
10720 (ix86_option_override_internal): Add flag_cf_protection
10722 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10723 (ix86_print_operand): Add 'notrack' prefix output.
10724 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10725 (ix86_expand_builtin): Expand CET intrinsics.
10726 (x86_output_mi_thunk): Add 'endbranch' instruction.
10727 * config/i386/i386.h (TARGET_IBT): New.
10728 (TARGET_IBT_P): Likewise.
10729 (TARGET_SHSTK): Likewise.
10730 (TARGET_SHSTK_P): Likewise.
10731 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10732 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10733 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10734 (builtin_setjmp_setup): New pattern.
10735 (builtin_longjmp): Likewise.
10736 (rdssp<mode>): Likewise.
10737 (incssp<mode>): Likewise.
10738 (saveprevssp): Likewise.
10739 (rstorssp): Likewise.
10740 (wrss<mode>): Likewise.
10741 (wruss<mode>): Likewise.
10742 (setssbsy): Likewise.
10743 (clrssbsy): Likewise.
10744 (nop_endbr): Likewise.
10745 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10747 * config/i386/immintrin.h: Include <cetintrin.h>.
10748 * config/i386/linux-common.h
10749 (file_end_indicate_exec_stack_and_cet): New prototype.
10750 (TARGET_ASM_FILE_END): New.
10752 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10754 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10755 latencies instead of having separate table; make difference between
10756 integer and float costs.
10757 * i386.h (processor_costs): Remove scalar_stmt_cost,
10758 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10759 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10761 * x86-tune-costs.h: Remove entries which has been removed in
10762 procesor_costs from all tables; make cond_taken_branch_cost
10763 and cond_not_taken_branch_cost COST_N_INSNS based.
10765 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10767 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10769 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10771 * config/i386/i386.md (isa): Remove fma_avx512f.
10772 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10773 <avx512>_fmadd_<mode>_mask3<round_name>,
10774 <avx512>_fmsub_<mode>_mask<round_name>,
10775 <avx512>_fmsub_<mode>_mask3<round_name>,
10776 <avx512>_fnmadd_<mode>_mask<round_name>,
10777 <avx512>_fnmadd_<mode>_mask3<round_name>,
10778 <avx512>_fnmsub_<mode>_mask<round_name>,
10779 <avx512>_fnmsub_<mode>_mask3<round_name>,
10780 <avx512>_fmaddsub_<mode>_mask<round_name>,
10781 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10782 <avx512>_fmsubadd_<mode>_mask<round_name>,
10783 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10784 (*vec_widen_umult_even_v16si<mask_name>,
10785 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10786 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10788 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10790 * extend.texi: Add 'nocf_check' documentation.
10791 * gimple.texi: Add second parameter to
10792 gimple_build_call_from_tree.
10793 * invoke.texi: Add -fcf-protection documentation.
10794 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10796 2017-10-20 Richard Biener <rguenther@suse.de>
10798 PR tree-optimization/82473
10799 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10800 the largest input type.
10802 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10804 * c-attribs.c (handle_nocf_check_attribute): New function.
10805 (c_common_attribute_table): Add 'nocf_check' handling.
10806 * gimple-parser.c: Add second argument NULL to
10807 gimple_build_call_from_tree.
10808 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10809 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10811 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10812 * common.opt: Add fcf-protection flag.
10813 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10814 * flag-types.h: Add enum cf_protection_level.
10815 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10816 Add 'nocf_check' attribute propagation to gimple call.
10817 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10818 (gimple_build_call_from_tree): Update prototype.
10819 (gimple_call_nocf_check_p): New function.
10820 (gimple_call_set_nocf_check): Likewise.
10821 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10822 * ipa-icf.c: Add nocf_check attribute in statement hash.
10823 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10824 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10825 * toplev.c (process_options): Add flag_cf_protection handling.
10827 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10829 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10831 2017-10-20 Richard Biener <rguenther@suse.de>
10833 PR tree-optimization/82603
10834 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10835 remove false predicated stores.
10837 2017-10-20 Richard Biener <rguenther@suse.de>
10839 * graphite-isl-ast-to-gimple.c
10840 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10841 Remove return value and simplify, dump copied stmt after lhs
10843 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10844 Reduce dump verbosity.
10845 (gsi_insert_earliest): Likewise.
10846 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10847 * graphite.c (print_global_statistics): Adjust dumping.
10848 (print_graphite_scop_statistics): Likewise.
10849 (print_graphite_statistics): Do not dump loops here.
10850 (graphite_transform_loops): But here.
10852 2017-10-20 Nicolas Roche <roche@adacore.com>
10854 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10855 * configure: Regenerate.
10857 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10860 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10861 functions when optimizing replace GIMPLE_RETURN stmts with
10862 calls to __builtin_unreachable ().
10865 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10866 for -fsanitize=thread link of executables.
10867 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10868 link of executables.
10871 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10872 New mode iterators.
10873 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10874 define_insns for logical vector shifts to use VI248_AVX512BW
10875 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10876 condition, useless isa and prefix attributes. Change the first
10877 2 of these define_insns to ...
10878 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10879 define_insn for avx512vl.
10880 (<shift_insn><mode>3): ... and this, new define_insn without
10881 masking for non-avx512vl.
10884 * config/i386/sse.md (*andnot<mode>3,
10885 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10886 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10887 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10888 not applied use empty suffix even for TARGET_AVX512VL.
10889 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10890 is applied, supply evex,evex or evex,evex,evex instead of just
10893 2017-10-20 Julia Koval <julia.koval@intel.com>
10895 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10896 (OPTION_MASK_ISA_GFNI_UNSET): New.
10897 (ix86_handle_option): Handle OPT_mgfni.
10898 * config/i386/cpuid.h (bit_GFNI): New.
10899 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10900 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10901 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10902 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10903 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10904 * config/i386/i386.opt: Add mgfni.
10906 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10908 * config/msp430/msp430.c (msp430_option_override): Disable
10909 -fdelete-null-pointer-checks.
10910 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10912 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10914 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10915 of x87 and SSE instructions.
10917 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10919 * asan.c (create_cond_insert_point): Do not update edge count.
10920 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10921 (afdo_propagate_circuit): Likewise.
10922 (afdo_calculate_branch_prob): Likewise.
10923 (afdo_annotate_cfg): Likewise.
10924 * basic-block.h (struct edge_def): Remove count.
10925 (edge_def::count): New accessor.
10926 * bb-reorder.c (rotate_loop): Update.
10927 (find_traces_1_round): Update.
10928 (connect_traces): Update.
10929 (sanitize_hot_paths): Update.
10930 * cfg.c (unchecked_make_edge): Update.
10931 (make_single_succ_edge): Update.
10932 (check_bb_profile): Update.
10933 (dump_edge_info): Update.
10934 (update_bb_profile_for_threading): Update.
10935 (scale_bbs_frequencies_int): Update.
10936 (scale_bbs_frequencies_gcov_type): Update.
10937 (scale_bbs_frequencies_profile_count): Update.
10938 (scale_bbs_frequencies): Update.
10939 * cfganal.c (connect_infinite_loops_to_exit): Update.
10940 * cfgbuild.c (compute_outgoing_frequencies): Update.
10941 (find_many_sub_basic_blocks): Update.
10942 * cfgcleanup.c (try_forward_edges): Update.
10943 (try_crossjump_to_edge): Update
10944 * cfgexpand.c (expand_gimple_cond): Update
10945 (expand_gimple_tailcall): Update
10946 (construct_exit_block): Update
10947 * cfghooks.c (verify_flow_info): Update
10948 (redirect_edge_succ_nodup): Update
10949 (split_edge): Update
10950 (make_forwarder_block): Update
10951 (duplicate_block): Update
10952 (account_profile_record): Update
10953 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10954 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10955 * cfgloopmanip.c (scale_loop_profile): Update.
10957 (lv_adjust_loop_entry_edge): Update.
10958 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10959 (force_nonfallthru_and_redirect): Update.
10960 (purge_dead_edges): Update.
10961 (rtl_flow_call_edges_add): Update.
10962 * cgraphunit.c (init_lowered_empty_function): Update.
10963 (cgraph_node::expand_thunk): Update.
10964 * gimple-pretty-print.c (dump_probability): Update.
10965 (dump_edge_probability): Update.
10966 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10967 * haifa-sched.c (sched_create_recovery_edges): Update.
10968 * hsa-gen.c (convert_switch_statements): Update.
10969 * ifcvt.c (dead_or_predicable): Update.
10970 * ipa-inline-transform.c (inline_transform): Update.
10971 * ipa-split.c (split_function): Update.
10972 * ipa-utils.c (ipa_merge_profiles): Update.
10973 * loop-doloop.c (add_test): Update.
10974 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10975 * lto-streamer-in.c (input_cfg): Update.
10976 (input_function): Update.
10977 * lto-streamer-out.c (output_cfg): Update.
10978 * modulo-sched.c (sms_schedule): Update.
10979 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10980 * predict.c (maybe_hot_edge_p): Update.
10981 (unlikely_executed_edge_p): Update.
10982 (probably_never_executed_edge_p): Update.
10983 (dump_prediction): Update.
10984 (drop_profile): Update.
10985 (propagate_unlikely_bbs_forward): Update.
10986 (determine_unlikely_bbs): Update.
10987 (force_edge_cold): Update.
10988 * profile.c (compute_branch_probabilities): Update.
10989 * reg-stack.c (better_edge): Update.
10990 * shrink-wrap.c (handle_simple_exit): Update.
10991 * tracer.c (better_p): Update.
10992 * trans-mem.c (expand_transaction): Update.
10993 (split_bb_make_tm_edge): Update.
10994 * tree-call-cdce.c: Update.
10995 * tree-cfg.c (gimple_find_sub_bbs): Update.
10996 (gimple_split_edge): Update.
10997 (gimple_duplicate_sese_region): Update.
10998 (gimple_duplicate_sese_tail): Update.
10999 (gimple_flow_call_edges_add): Update.
11000 (insert_cond_bb): Update.
11001 (execute_fixup_cfg): Update.
11002 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11003 * tree-complex.c (expand_complex_div_wide): Update.
11004 * tree-eh.c (lower_resx): Update.
11005 (unsplit_eh): Update.
11006 (cleanup_empty_eh_move_lp): Update.
11007 * tree-inline.c (copy_edges_for_bb): Update.
11008 (freqs_to_counts): Update.
11009 (copy_cfg_body): Update.
11010 * tree-ssa-dce.c (remove_dead_stmt): Update.
11011 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11012 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11013 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11014 (unloop_loops): Update.
11015 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11016 * tree-ssa-loop-split.c (connect_loops): Update.
11017 (split_loop): Update.
11018 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11019 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11020 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11021 * tree-ssa-reassoc.c (branch_fixup): Update.
11022 * tree-ssa-tail-merge.c (replace_block_by): Update.
11023 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11024 (compute_path_counts): Update.
11025 (update_profile): Update.
11026 (recompute_probabilities): Update.
11027 (update_joiner_offpath_counts): Update.
11028 (estimated_freqs_path): Update.
11029 (freqs_to_counts_path): Update.
11030 (clear_counts_path): Update.
11031 (ssa_fix_duplicate_block_edges): Update.
11032 (duplicate_thread_path): Update.
11033 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11034 (case_bit_test_cmp): Update.
11035 (collect_switch_conv_info): Update.
11036 (gen_inbound_check): Update.
11037 (do_jump_if_equal): Update.
11038 (emit_cmp_and_jump_insns): Update.
11039 * tree-tailcall.c (decrease_profile): Update.
11040 (eliminate_tail_call): Update.
11041 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11042 (vect_do_peeling): Update.
11043 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11044 * ubsan.c (ubsan_expand_null_ifn): Update.
11045 (ubsan_expand_ptr_ifn): Update.
11046 * value-prof.c (gimple_divmod_fixed_value): Update.
11047 (gimple_mod_pow2): Update.
11048 (gimple_mod_subtract): Update.
11049 (gimple_ic): Update.
11050 (gimple_stringop_fixed_value): Update.
11052 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11055 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11057 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11059 PR rtl-optimization/82395
11060 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11061 based on non_spilled_static_chain_regno_p.
11063 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11065 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11066 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11067 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11069 2017-10-19 Martin Sebor <msebor@redhat.com>
11071 PR tree-optimization/82596
11072 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11074 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11076 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11077 (get_mem_refs_of_builtin_call): Likewise.
11078 * builtins.c (expand_builtin_apply): Adjust call to
11079 allocate_dynamic_stack_space.
11080 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11081 the third argument to allocate_dynamic_stack_space, otherwise -1.
11082 (expand_builtin): Deal with all alloca variants.
11083 (is_inexpensive_builtin): Likewise.
11084 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11085 * calls.c (special_function_p): Deal with all alloca variants.
11086 (initialize_argument_information): Adjust call to
11087 allocate_dynamic_stack_space.
11088 (expand_call): Likewise.
11089 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11090 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11091 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11092 use it for the stack usage computation.
11093 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11094 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11095 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11096 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11097 (in_loop_p): Remove first argument and useless check.
11098 (pass_walloca::execute): Remove useless test and adjust call to above.
11099 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11100 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11101 (gimplify_call_expr): Deal with all alloca variants.
11102 * hsa-gen.c (gen_hsa_alloca): Likewise.
11103 (gen_hsa_insns_for_call): Likewise.
11104 * ipa-pure-const.c (special_builtin_state): Likewise.
11105 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11106 * tree-object-size.c (alloc_object_size): Likewise.
11107 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11108 (call_may_clobber_ref_p_1): Likewise.
11109 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11110 (ccp_fold_stmt): Likewise.
11111 (optimize_stack_restore): Likewise.
11112 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11113 (mark_all_reaching_defs_necessary_1): Likewise.
11114 (propagate_necessity): Likewise.
11115 (eliminate_unnecessary_stmts): Likewise.
11116 * tree.c (build_common_builtin_nodes): Build
11117 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11118 (build_alloca_call_expr): New function.
11119 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11120 (CASE_BUILT_IN_ALLOCA): Likewise.
11121 (build_alloca_call_expr): Declare.
11122 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11124 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11127 * dwarf2out.c (new_die_raw): New static inline function.
11128 (new_die): Use it to create the DIE.
11129 (add_AT_external_die_ref): Likewise.
11130 (clone_die): Likewise.
11131 (clone_as_declaration): Likewise.
11132 (dwarf2out_vms_debug_main_pointer): Likewise.
11133 (base_type_die): Likewise. Remove early return for corner cases.
11134 Do not call add_pubtype on the DIE here.
11135 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11136 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11137 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11138 native order exists for base types, attach the DIE manually and call
11139 add_pubtype on it. Do not equate a reverse order DIE to the type.
11141 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11143 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11144 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11146 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11148 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11150 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11151 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11152 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11153 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11154 * opts.c (sanitizer_opts): Add builtin.
11155 * ubsan.c (instrument_builtin): New function.
11156 (pass_ubsan::execute): Call it.
11157 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11158 * doc/invoke.texi: Document -fsanitize=builtin.
11160 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11161 builtins, store max (log2 (align), 0) into uchar field instead of
11162 align into uptr field.
11163 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11164 store uchar 0 field instead of uptr 0 field.
11165 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11166 instead of passing one address of struct with 2 locations pass
11167 two addresses of structs with 1 location each.
11168 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11169 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11170 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11171 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11172 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11173 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11174 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11175 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11177 2017-10-19 Martin Liska <mliska@suse.cz>
11180 * file-find.c (remove_prefix): Remove.
11181 * file-find.h (remove_prefix): Likewise.
11182 * gcc-ar.c: Remove smartness of lookup.
11184 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11186 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11187 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11188 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11190 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11193 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11194 (setcc + and to xor + setcc): New peephole2.
11196 2017-10-19 Tom de Vries <tom@codesourcery.com>
11198 * doc/sourcebuild.texi (Test Directives, Variants of
11199 dg-require-support): Add dg-require-stack-size.
11201 2017-10-19 Martin Liska <mliska@suse.cz>
11204 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11205 that have a large alignment.
11206 (gimplify_target_expr): Likewise.
11208 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11210 PR rtl-optimization/82602
11211 * ira.c (rtx_moveable_p): Return false for volatile asm.
11213 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11216 * config/i386/i386-modes.def (CCGZ): New CC mode.
11217 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11218 * config/i386/predicates.md (ix86_comparison_operator):
11220 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11221 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11222 with double-word subtraction.
11223 (put_condition_code): Handle CCGZmode.
11225 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11227 * wide-int.cc (debug (const wide_int &)): New.
11228 (debug (const wide_int *)): New.
11229 (debug (const widest_int &)): New.
11230 (debug (const widest_int *)): New.
11232 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11234 PR middle-end/82556
11235 * lra-constraints.c (curr_insn_transform): Use non-input operand
11236 instead of output one for matched reload.
11238 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11240 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11241 (tree-ssa-loop-ivopts.h): New header file.
11242 (struct builtin_info): New fields.
11243 (classify_builtin_1): Compute and record base and offset parts for
11244 memset builtin partition by calling strip_offset.
11245 (offset_cmp, fuse_memset_builtins): New functions.
11246 (finalize_partitions): Fuse adjacent memset partitions by calling
11248 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11249 Expose the interface.
11250 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11252 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11254 PR tree-optimization/82574
11255 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11256 that data reference must be executed exactly once per iteration
11257 against the outermost loop in nest.
11258 (classify_partition): Update call to above function.
11260 2017-10-18 Richard Biener <rguenther@suse.de>
11262 PR tree-optimization/82591
11263 * graphite.c (graphite_transform_loops): Move code gen message
11265 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11266 Here. Handle scop_to_isl_ast failing.
11267 (scop_to_isl_ast): Limit the number of ISL operations.
11269 2017-10-18 Richard Biener <rguenther@suse.de>
11271 * graphite-isl-ast-to-gimple.c
11272 (translate_isl_ast_to_gimple::set_rename): Simplify.
11273 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11274 (graphite_copy_stmts_from_block): ... here.
11275 (copy_bb_and_scalar_dependences): Simplify.
11276 (add_parameters_to_ivs_params): Canonicalize.
11277 (generate_entry_out_of_ssa_copies): Simplify.
11278 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11279 by passing in ISL dimension.
11280 (parameter_index_in_region_1): Rename to ...
11281 (parameter_index_in_region): ... this.
11282 (extract_affine): Adjust assert, pass down parameter index.
11283 (add_param_constraints): Use range-info when available.
11284 (build_scop_context): Adjust.
11285 * sese.c (new_sese_info): Adjust.
11286 (free_sese_info): Likewise.
11287 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11288 Remove unused typedefs.
11289 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11291 2017-10-18 Martin Liska <mliska@suse.cz>
11293 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11295 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11297 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11298 (s390_sched_init): Do not reset s390_sched_state if we entered the
11299 current basic block via a fallthru edge and all others are unlikely.
11301 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11303 * config/s390/s390.c (NUM_SIDES): New variable.
11304 (LONGRUNNING_THRESHOLD): New variable.
11305 (LATENCY_FACTOR): New variable.
11306 (s390_sched_score): Decrease score for long-running instructions on
11308 (s390_sched_variable_issue): Perform bookkeeping for long-running
11311 2017-10-18 Richard Biener <rguenther@suse.de>
11313 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11314 Simplify with removal of the parameter rename map.
11315 (set_rename): Likewise.
11316 (should_copy_to_new_region): Likewise.
11317 (graphite_copy_stmts_from_block): Likewise.
11318 (copy_bb_and_scalar_dependences): Remove initialization of
11319 unused copied_bb_map.
11320 (copy_def): Remove.
11321 (copy_internal_parameters): Likewise.
11322 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11323 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11324 Use INTEGRAL_TYPE_P.
11325 (parameter_index_in_region_1): Rename to ...
11326 (assign_parameter_index_in_region): ... this. Assert we have
11327 a parameter we handle.
11328 (scan_tree_for_params): Adjust.
11329 * sese.h (parameter_rename_map_t): Remove.
11330 (struct sese_info_t): Remove unused parameter_rename_map and
11331 copied_bb_map members.
11332 * sese.c (new_sese_info): Adjust.
11333 (free_sese_info): Likewise.
11335 2017-10-18 Martin Liska <mliska@suse.cz>
11338 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11339 on an abnormal edge.
11341 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11343 * doc/invoke.texi (ffunction-sections and fdata-sections):
11346 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11348 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11349 the use statement can throw internally.
11351 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11353 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11354 any RTX present on the RHS of a SET.
11355 * compare-elim.c (try_eliminate_compare): Restore comment.
11357 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11359 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11360 may be also called on tcc_type nodes.
11361 * langhooks.c (lhd_tree_size): Likewise.
11363 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11365 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11366 format_warning_at_substring.
11367 (maybe_warn): Convert source_range * param to a location_t. Pass
11368 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11369 (format_directive): Remove code to extract source_ranges and
11370 source_range * in favor of just a location_t.
11371 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11373 * substring-locations.c (format_warning_va): Convert
11374 source_range * param to a location_t.
11375 (format_warning_at_substring): Likewise.
11376 * substring-locations.h (format_warning_va): Likewise.
11377 (format_warning_at_substring): Likewise.
11379 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11381 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11383 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11384 and scatter/gather ops.
11386 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11387 vec_gather_load and vec_scatter_store.
11388 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11389 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11390 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11391 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11392 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11393 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11395 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11397 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11398 Detect FTST instruction and handle its register pops. Only pop
11399 second operand if can_pop_second_op is true.
11400 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11401 set can_pop_second_op to false in the compare_for_stack_reg call.
11403 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11404 output_fp_compare for stack register operands.
11405 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11406 instructions here. Do not emit stack register pops here. Assert
11407 that FCOMPP pops next to top stack register. Rewrite function.
11409 2017-10-17 Nathan Sidwell <nathan@acm.org>
11411 PR middle-end/82577
11412 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11413 use DECL_ASSEMBLER_NAME_RAW.
11415 PR middle-end/82546
11416 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11419 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11420 Wilco Dijkstra <wilco.dijkstra@arm.com>
11422 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11423 converstion to Pmode from the buf_addr.
11425 2017-10-17 Richard Biener <rguenther@suse.de>
11427 * graphite-dependences.c (scop_get_reads_and_writes): Change
11428 output parameters to references.
11430 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11432 PR 71026/tree-optimization
11433 * fold-const.c (distribute_real_division): Removed.
11434 (fold_binary_loc): Remove calls to distribute_real_divison.
11436 2017-10-17 Richard Biener <rguenther@suse.de>
11438 * graphite-scop-detection.c
11439 (scop_detection::stmt_has_simple_data_refs_p): Always use
11440 the full nest as region.
11441 (try_generate_gimple_bb): Likewise.
11442 * sese.c (scalar_evolution_in_region): Simplify now that
11443 SCEV can handle instantiation in regions.
11444 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11445 in the non-loop part of a function if requested.
11447 2017-10-17 Richard Biener <rguenther@suse.de>
11449 PR tree-optimization/82563
11450 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11452 (graphite_regenerate_ast_isl): Call it.
11453 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11455 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11457 PR tree-optimization/82549
11458 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11459 Formatting fixes. Instead of calling make_bit_field_ref with negative
11462 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11464 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11465 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11466 _mm_maskz_reduce_ss): New.
11467 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11468 __builtin_ia32_reducess_mask): Ditto..
11469 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11470 * config/i386/sse.md (reduces<mode>): Renamed to ...
11471 (reduces<mode><mask_scalar_name>): ... this.
11472 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11474 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11475 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11477 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11479 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11480 * selftest-run-tests.c (selftest::run_tests): Call
11481 selftest::unique_ptr_tests_cc_tests.
11482 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11483 * unique-ptr-tests.cc: New file.
11485 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11488 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11490 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11491 (make_hard_regno_born, make_hard_regno_dead): Update
11492 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11494 2017-10-16 Jeff Law <law@redhat.com>
11496 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11498 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11500 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11502 2017-10-16 Olivier Hainque <hainque@adacore.com>
11504 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11505 with_cpu if we were configured for an e500v2 target cpu name.
11507 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11509 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11510 * doc/invoke.texi: Document +nodsp as a valid extension for
11513 2017-10-16 Martin Liska <mliska@suse.cz>
11515 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11516 (test_set_range): Likewise.
11517 (test_range_functions): Rename to ...
11518 (test_bit_in_range): ... this.
11519 (sbitmap_c_tests): Add new test.
11521 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11523 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11525 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11526 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11528 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11530 * config/aarch64/aarch64-builtins.c
11531 (aarch64_types_quadopu_lane_qualifiers): New.
11532 (TYPES_QUADOPU_LANE): New.
11533 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11534 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11535 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11536 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11537 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11538 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11539 (Vdottype, DOTPROD): New.
11540 (sur): Add SDOT and UDOT.
11542 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11544 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11545 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11546 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11547 Add TARGET_DOTPROD.
11548 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11549 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11550 Enable TARGET_DOTPROD.
11551 (cortex-a75.cortex-a55): Likewise.
11552 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11554 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11556 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11557 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11559 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11561 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11562 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11563 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11564 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11565 * config/arm/types.md (neon_dot, neon_dot_q): New.
11566 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11568 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11570 * config/arm/arm.h (TARGET_DOTPROD): New.
11571 * config/arm/arm.c (arm_arch_dotprod): New.
11572 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11573 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11574 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11575 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11576 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11577 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11578 * doc/invoke.texi (armv8.2-a): Document dotprod
11580 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11582 * i386.c (ix86_vec_cost): New function.
11583 (ix86_rtx_costs): Handle vector operations better.
11584 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11585 * x86-tune-costs.h: Add new costs to all tables.
11587 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11589 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11591 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11592 divsd, sqrtss and sqrtsd
11593 * x86-tune-costs.h: Add new entries to all costs.
11594 (znver1_cost): Fix to match real instruction latencies.
11596 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11597 Michael Collison <michael.collison@arm.com>
11599 * compare-elim.c: Include emit-rtl.h.
11600 (can_merge_compare_into_arith): New function.
11601 (try_validate_parallel): Likewise.
11602 (try_merge_compare): Likewise.
11603 (try_eliminate_compare): Call the above when no previous clobber
11605 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11608 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11610 PR middle-end/62263
11611 PR middle-end/82498
11612 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11613 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11615 PR middle-end/62263
11616 PR middle-end/82498
11617 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11618 to be any operand_equal_p operands. For & (B - 1) require
11619 B to be power of 2. Recognize
11620 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11622 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11625 * optabs.c (expand_memory_blockage): Fix call of
11626 targetm.have_memory_blockage.
11628 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11631 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11632 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11633 objects to extra_objs instead of overwriting it.
11635 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11637 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11638 Use any_fp_register_operand as operand[3] predicate. Simplify
11639 equality test for operands[2] and operands[4] memory location.
11640 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11641 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11642 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11643 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11644 any_fp_register_operand as operand[1] predicate. Simplify
11645 equality test for operands[0] and operands[3] memory location.
11646 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11647 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11648 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11650 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11652 * target-insns.def: Add memory_blockage.
11653 * optabs.c (expand_memory_blockage): New function.
11654 (expand_asm_memory_barrier): Rename ...
11655 (expand_asm_memory_blockage): ... to this.
11656 (expand_mem_thread_fence): Call expand_memory_blockage
11657 instead of expand_asm_memory_barrier.
11658 (expand_mem_singnal_fence): Ditto.
11659 (expand_atomic_load): Ditto.
11660 (expand_atomic_store): Ditto.
11661 * doc/md.texi (Standard Pattern Names For Generation):
11662 Document memory_blockage instruction pattern.
11664 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11666 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11667 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11668 * config/rl78/rl78.md: New define_expand "adddi3".
11670 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11672 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11675 2017-10-13 Jeff Law <law@redhat.com>
11677 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11679 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11682 * internal-fn.c (expand_mul_overflow): If both operands have
11683 the same highpart of -1 or 0 and the topmost bit of lowpart
11684 is different, overflow is if res <= 0 rather than res < 0.
11686 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11688 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11689 TARGET_P9_VECTOR code for unaligned_load case.
11691 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11693 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11695 2017-10-13 Nathan Sidwell <nathan@acm.org>
11697 * tree-core.h (tree_contains_struct): Make bool.
11698 * tree.c (tree_contains_struct): Likewise.
11699 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11700 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11701 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11702 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11703 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11705 2017-10-13 Richard Biener <rguenther@suse.de>
11707 * graphite-isl-ast-to-gimple.c
11708 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11709 parameters and dominance check.
11710 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11711 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11712 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11713 Do not update SSA form here or do intermediate IL verification.
11714 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11715 (graphite_initialize): Remove check on the number of loops in
11716 the function and inline into graphite_transform_loops.
11717 (graphite_finalize): Inline into graphite_transform_loops.
11718 (graphite_transform_loops): Perform SSA update and IL verification
11720 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11722 2017-10-13 Richard Biener <rguenther@suse.de>
11724 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11725 graphite_expression_type_precision): Avoid global constructor
11727 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11728 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11729 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11730 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11731 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11732 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11734 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11737 * config/i386/i386.h (ix86_red_zone_size): New.
11738 * config/i386/i386.md (push peephole2s): Replace
11739 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11741 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11742 Alan Hayward <alan.hayward@arm.com>
11743 David Sherwood <david.sherwood@arm.com>
11745 * combine.c (can_change_dest_mode): Reject changes in
11746 REGMODE_NATURAL_SIZE.
11748 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11749 Alan Hayward <alan.hayward@arm.com>
11750 David Sherwood <david.sherwood@arm.com>
11752 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11753 (expand_debug_source_expr): Likewise.
11754 * combine.c (combine_simplify_rtx): Likewise.
11755 * cse.c (fold_rtx): Likewise.
11756 * fwprop.c (canonicalize_address): Likewise.
11757 * targhooks.c (default_shift_truncation_mask): Likewise.
11759 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11760 Alan Hayward <alan.hayward@arm.com>
11761 David Sherwood <david.sherwood@arm.com>
11763 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11764 (widened_mode): Likewise.
11765 (expand_unop): Likewise.
11766 * ree.c (transform_ifelse): Likewise.
11767 (merge_def_and_ext): Likewise.
11768 (combine_reaching_defs): Likewise.
11769 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11771 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11772 Alan Hayward <alan.hayward@arm.com>
11773 David Sherwood <david.sherwood@arm.com>
11775 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11776 * combine.c (gen_lowpart_for_combine): Likewise.
11777 * dwarf2out.c (rtl_for_decl_location): Likewise.
11778 * final.c (alter_subreg): Likewise.
11779 * rtlhooks.c (gen_lowpart_general): Likewise.
11780 (gen_lowpart_if_possible): Likewise.
11782 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11783 Alan Hayward <alan.hayward@arm.com>
11784 David Sherwood <david.sherwood@arm.com>
11786 * calls.c (expand_call): Use subreg_lowpart_offset.
11787 * cse.c (cse_insn): Likewise.
11788 * regcprop.c (copy_value): Likewise.
11789 (copyprop_hardreg_forward_1): Likewise.
11791 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11794 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11795 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11796 =Q constraints to +Q and into insn condition add check
11797 that operands[0] and operands[1] are equal.
11798 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11799 =Q constraints to +Q and into insn condition add check
11800 that operands[0] is equal to either operands[1] or operands[2].
11803 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11804 instead of handling MINUS_EXPR twice (once for each argument),
11805 canonicalize operand order and handle just once, use rtype where
11806 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11809 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11810 any values of __C while still being pattern recognizable as a simple
11811 rotate instruction.
11813 2017-10-13 Richard Biener <rguenther@suse.de>
11815 PR tree-optimization/82451
11817 2017-10-02 Richard Biener <rguenther@suse.de>
11819 PR tree-optimization/82355
11820 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11821 a mapping for the enclosing loop but avoid generating one for
11822 the loop tree root.
11823 (copy_bb_and_scalar_dependences): Remove premature codegen
11824 error on PHIs in blocks duplicated into multiple places.
11825 * graphite-scop-detection.c
11826 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11827 in the region use it as loop and nest to analyze the DR in.
11828 (try_generate_gimple_bb): Likewise.
11829 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11830 (add_loop_constraints): For blocks in a loop not in the region
11831 create a dimension with a single iteration.
11832 * sese.h (gbb_loop_at_index): Remove assert.
11834 * cfgloop.c (loop_preheader_edge): For the loop tree root
11835 return the single successor of the entry block.
11836 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11837 Reset the SCEV hashtable and niters.
11838 * graphite-scop-detection.c
11839 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11840 assert that we only have POLYNOMIAL_CHREC that vary in loops
11841 contained in the region.
11842 (scop_detection::graphite_can_represent_expr): Adjust.
11843 (scop_detection::stmt_has_simple_data_refs_p): For loops
11844 not in the region set loop to NULL. The nest is now the
11845 entry edge to the region.
11846 (try_generate_gimple_bb): Likewise.
11847 * sese.c (scalar_evolution_in_region): Adjust for
11848 instantiate_scev change.
11849 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11850 Make nest parameter the edge into the region.
11851 (create_data_ref): Likewise.
11852 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11853 entry edge into a region and adjust instantiate_scev calls.
11854 (create_data_ref): Likewise.
11855 (graphite_find_data_references_in_stmt): Likewise.
11856 (find_data_references_in_stmt): Pass the loop preheader edge
11857 from the nest argument.
11858 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11859 parameter the edge into the region.
11860 (instantiate_parameters): Use the loop preheader edge as entry.
11861 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11863 (get_instantiated_value_entry): Make instantiate_below parameter
11864 the edge into the region.
11865 (instantiate_scev_name): Likewise. Adjust dominance checks,
11866 when we cannot use loop-based instantiation instantiate by
11867 walking use-def chains.
11868 (instantiate_scev_poly): Adjust.
11869 (instantiate_scev_binary): Likewise.
11870 (instantiate_scev_convert): Likewise.
11871 (instantiate_scev_not): Likewise.
11872 (instantiate_array_ref): Remove.
11873 (instantiate_scev_3): Likewise.
11874 (instantiate_scev_2): Likewise.
11875 (instantiate_scev_1): Likewise.
11876 (instantiate_scev_r): Do not blindly handle N-operand trees.
11877 Do not instantiate array-refs. Handle all constants and invariants.
11878 (instantiate_scev): Make instantiate_below parameter
11879 the edge into the region.
11880 (resolve_mixers): Use the loop preheader edge for the region
11881 parameter to instantiate_scev_r.
11882 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11884 2017-10-13 Richard Biener <rguenther@suse.de>
11886 PR tree-optimization/82525
11887 * graphite-isl-ast-to-gimple.c
11888 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11890 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11891 Fail code generation when we cannot represent the isl integer.
11892 (binary_op_to_tree): Elide modulo operations that are no-ops
11893 in the type we code generate. Remove now superfluous code
11896 2017-10-13 Richard Biener <rguenther@suse.de>
11898 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11899 (scop_detection::harmful_loop_in_region): Remove premature
11900 IV type restriction.
11901 (scop_detection::graphite_can_represent_scev): We can handle
11902 pointer IVs just fine.
11904 2017-10-13 Alan Modra <amodra@gmail.com>
11906 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11907 "Clobbers and Scratch Registers". Add paragraph on
11908 alternative to clobbers for scratch registers and OpenBLAS
11911 2017-10-13 Alan Modra <amodra@gmail.com>
11913 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11914 example of a memory input for a string of known length. Move
11915 commentary out of table. Add a number of new examples
11916 covering array memory inputs.
11918 2017-10-12 Martin Liska <mliska@suse.cz>
11920 PR tree-optimization/82493
11921 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11922 (test_range_functions): New function.
11923 (sbitmap_c_tests): Likewise.
11924 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11925 * selftest.h (sbitmap_c_tests): New function.
11927 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11929 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11931 * config/rs6000/amo.h: Fix spacing issue.
11933 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11936 * config/i386/i386.md (*ashl<mode>3_mask_1,
11937 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11938 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11941 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11943 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11944 (profile_probability): Set max_probability
11945 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11947 * profile-count.c (profile_probability::differs_from_p): Do not
11948 rely on max_probaiblity == 10000
11950 2017-10-12 Jeff Law <law@redhat.com>
11952 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11955 2017-10-12 Martin Sebor <msebor@redhat.com>
11959 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11960 (handle_alias_pairs): Call it.
11961 * common.opt (-Wattribute-alias): New option.
11962 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11963 * doc/invoke.texi (-Wattribute-alias): Document.
11965 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11968 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11970 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11972 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11973 (make_hard_regno_born, make_hard_regno_dead): Update
11974 bb_killed_pseudos and bb_gen_pseudos.
11976 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11978 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11980 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11982 * config/alpha/alpha.c (alpha_split_conditional_move):
11983 Use std::swap instead of manually swapping.
11984 (alpha_stdarg_optimize_hook): Ditto.
11985 (alpha_canonicalize_comparison): Ditto.
11987 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11989 * tree-loop-distribution.c (struct builtin_info): New struct.
11990 (struct partition): Refactor fields into struct builtin_info.
11991 (partition_free): Free struct builtin_info.
11992 (build_size_arg_loc, build_addr_arg_loc): Delete.
11993 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11994 information from struct builtin_info.
11995 (find_single_drs): New function refactored from classify_partition.
11996 Also moved builtin validity checks to this function.
11997 (compute_access_range, alloc_builtin): New functions.
11998 (classify_builtin_st, classify_builtin_ldst): New functions.
11999 (classify_partition): Refactor code into functions find_single_drs,
12000 classify_builtin_st and classify_builtin_ldst.
12001 (distribute_loop): Don't do runtime alias check when distributing
12003 (find_seed_stmts_for_distribution): New function.
12004 (pass_loop_distribution::execute): Refactor code finding seed
12005 stmts into above function. Support distribution for the innermost
12006 two-level loop nest. Adjust dump information.
12008 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12010 * tree-loop-distribution.c: Adjust the general comment.
12011 (NUM_PARTITION_THRESHOLD): New macro.
12012 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12013 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12014 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12015 in call to build_partition_graph.
12016 (finalize_partitions): New parameter. Make loop distribution more
12017 conservative by fusing more partitions.
12018 (distribute_loop): Don't do runtime alias check in case of loop nest
12020 (find_seed_stmts_for_distribution): New function.
12021 (prepare_perfect_loop_nest): New function.
12022 (pass_loop_distribution::execute): Refactor code finding seed stmts
12023 and loop nest into above functions. Support loop nest distribution.
12024 Adjust dump information accordingly.
12026 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12028 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12029 and set PTYPE_SEQUENTIAL for merged partition.
12031 2017-10-12 Richard Biener <rguenther@suse.de>
12033 PR tree-optimization/69728
12035 2017-09-19 Richard Biener <rguenther@suse.de>
12037 PR tree-optimization/69728
12038 * graphite-sese-to-poly.c (schedule_error): New global.
12039 (add_loop_schedule): Handle empty domain by failing the
12041 (build_original_schedule): Handle schedule_error.
12043 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12044 domain by returning an unchanged schedule.
12046 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12048 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12049 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12051 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12053 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12056 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12059 * expr.c (store_field): Don't optimize away bitsize == 0 store
12060 from CALL_EXPR with addressable return type.
12062 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12064 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12065 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12066 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12067 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12068 TARGET_ISEL instead of TARGET_ISEL<sel>.
12070 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12072 * config/rs6000/rs6000.c
12073 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12075 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12077 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12079 (reg_or_cint_operand): Fix comment.
12080 (reg_or_zero_operand): New predicate.
12081 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12082 * config/rs6000/rs6000.c (output_isel): Delete.
12083 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12084 instead of reg_or_cint_operand. Output instruction directly (not via
12086 (isel_unsigned_<mode>): Ditto.
12087 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12088 gpc_reg_operand. Add an instruction alternative for this. Output
12089 instruction directly.
12090 (*isel_reversed_unsigned_<mode>): Ditto.
12092 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12094 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12095 (TARGET_CANONICALIZE_COMPARISON): Define.
12097 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12100 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12101 Check whether the dest is REG before adding REG_EQUIV note.
12103 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12106 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12108 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12109 (make_hard_regno_born, make_hard_regno_dead): Update
12110 bb_killed_pseudos and bb_gen_pseudos.
12112 2017-10-11 Nathan Sidwell <nathan@acm.org>
12114 * incpath.h (enum incpath_kind): Name enum, prefix values.
12115 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12116 * incpath.c (heads, tails): Use INC_MAX.
12117 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12118 (merge_include_chains, split_quote_chain,
12119 register_include_chains): Update incpath_kind names.
12120 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12121 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12123 (add_framework_path, darwin_register_objc_includes): Likewise.
12124 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12126 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12128 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12129 Do not use float_operator operator predicate.
12130 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12131 * config/i386/predicates.md (float_operator): Remove predicate.
12133 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12135 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12136 (*jccxf_i387): Ditto.
12137 (*jcc<mode>_i387): Ditto.
12138 (*jccu<mode>_i387): Ditto.
12139 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12140 (*jcc_*_i387 splitters): Remove.
12141 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12142 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12143 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12146 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12148 * profile-count.h (slow_safe_scale_64bit): New function.
12149 (safe_scale_64bit): New inline.
12150 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12151 * profile-count.c: Include wide-int.h
12152 (slow_safe_scale_64bit): New.
12154 2017-10-11 Nathan Sidwell <nathan@acm.org>
12156 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12157 HAS_DECL_ASSEMBLER_NAME_P.
12158 * gimple-expr.c (gimple_decl_printable_name: Check
12159 HAS_DECL_ASSEMBLER_NAME_P too.
12160 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12161 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12162 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12163 * passes.c (rest_of_decl_compilation): Check
12164 HAS_DECL_ASSEMBLER_NAME_P too.
12165 * recog.c (verify_changes): Likewise.
12166 * tree-pretty-print.c (dump_decl_name): Likewise.
12167 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12169 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12170 (SET_DECL_ASSEMBLER_NAME): Use it.
12171 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12172 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12173 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12175 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12177 * config.gcc (i386, x86_64): Add extra objects.
12178 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12179 (ix86_min_insn_size): Declare.
12180 (ix86_issue_rate): Declare.
12181 (ix86_adjust_cost): Declare.
12182 (ia32_multipass_dfa_lookahead): Declare.
12183 (ix86_macro_fusion_p): Declare.
12184 (ix86_macro_fusion_pair_p): Declare.
12185 (ix86_bd_has_dispatch): Declare.
12186 (ix86_bd_do_dispatch): Declare.
12187 (ix86_core2i7_init_hooks): Declare.
12188 (ix86_atom_sched_reorder): Declare.
12189 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12190 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12191 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12192 (rip_relative_addr_p): Rename to ...
12193 (ix86_rip_relative_addr_p): ... this one; export.
12194 (memory_address_length): Update.
12195 (ix86_issue_rate): Move to x86-tune-sched.c.
12196 (ix86_flags_dependent): Move to x86-tune-sched.c.
12197 (ix86_agi_dependent): Move to x86-tune-sched.c.
12198 (exact_dependency_1): Move to x86-tune-sched.c.
12199 (exact_store_load_dependency): Move to x86-tune-sched.c.
12200 (ix86_adjust_cost): Move to x86-tune-sched.c.
12201 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12202 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12203 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12204 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12205 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12206 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12207 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12208 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12209 (min_insn_size): Rename to ...
12210 (ix86_min_insn_size): ... this one; export.
12211 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12212 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12213 (core2i7_first_cycle_multipass_backtrack): Move to
12214 x86-tune-sched-core.c.
12215 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12216 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12217 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12218 (ix86_avoid_jump_mispredicts): Update.
12219 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12220 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12221 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12222 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12223 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12224 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12225 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12226 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12227 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12228 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12229 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12230 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12231 (BIG): Move to ix86-tune-sched-bd.c.
12232 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12233 (enum insn_path): Move to ix86-tune-sched-bd.c.
12234 (get_mem_group): Move to ix86-tune-sched-bd.c.
12235 (is_cmp): Move to ix86-tune-sched-bd.c.
12236 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12237 (is_branch): Move to ix86-tune-sched-bd.c.
12238 (is_prefetch): Move to ix86-tune-sched-bd.c.
12239 (init_window): Move to ix86-tune-sched-bd.c.
12240 (allocate_window): Move to ix86-tune-sched-bd.c.
12241 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12242 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12243 (process_end_window): Move to ix86-tune-sched-bd.c.
12244 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12245 (find_constant): Move to ix86-tune-sched-bd.c.
12246 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12247 (has_immediate): Move to ix86-tune-sched-bd.c.
12248 (get_insn_path): Move to ix86-tune-sched-bd.c.
12249 (get_insn_group): Move to ix86-tune-sched-bd.c.
12250 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12251 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12252 (add_insn_window): Move to ix86-tune-sched-bd.c.
12253 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12254 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12255 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12256 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12257 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12258 (do_dispatch): Move to ix86-tune-sched-bd.c.
12259 (has_dispatch): Move to ix86-tune-sched-bd.c.
12260 * config/i386/t-i386: Add new object files.
12261 * config/i386/x86-tune-costs.h: New file.
12262 * config/i386/x86-tune-sched-atom.c: New file.
12263 * config/i386/x86-tune-sched-bd.c: New file.
12264 * config/i386/x86-tune-sched-core.c: New file.
12265 * config/i386/x86-tune-sched.c: New file.
12267 2017-10-11 Liu Hao <lh_mouse@126.com>
12269 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12270 the generic version below instead.
12271 (should_colorize): Recognize Windows consoles as terminals
12273 * pretty-print.c [__MINGW32__] (write_all): New function.
12274 [__MINGW32__] (find_esc_head): Likewise.
12275 [__MINGW32__] (find_esc_terminator): Likewise.
12276 [__MINGW32__] (eat_esc_sequence): Likewise.
12277 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12279 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12282 2017-10-11 Richard Biener <rguenther@suse.de>
12284 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12285 Properly call analyze_scalar_evolution with the loop of the stmt.
12287 2017-10-11 Richard Biener <rguenther@suse.de>
12289 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12290 * tree-core.h (tree_base): Add chrec_var union member.
12291 * tree.h (CHREC_VAR): Remove.
12292 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12293 * tree-chrec.h (build_polynomial_chrec): Adjust.
12294 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12295 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12297 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12299 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12300 * match.pd: ... here.
12301 ((T) X == (T) Y): Relax condition.
12303 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12305 PR tree-optimization/82472
12306 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12308 (break_alias_scc_partitions): Update postorder number.
12310 2017-10-11 Martin Liska <mliska@suse.cz>
12313 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12315 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12317 2017-10-11 Martin Liska <mliska@suse.cz>
12319 * print-rtl.c (print_insn): Move declaration of idbuf
12320 to same scope as name.
12322 2017-10-11 Martin Liska <mliska@suse.cz>
12327 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12328 volatile arguments.
12330 2017-10-11 Martin Liska <mliska@suse.cz>
12333 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12334 volatile arguments.
12336 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12338 * config.gcc (default_gnu_indirect_function): Default to yes for
12339 arm*-*-linux* with glibc.
12341 2017-10-11 Richard Biener <rguenther@suse.de>
12343 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12344 default-defs and types we do not want to analyze.
12345 (interpret_loop_phi): Replace unreachable code with an assert.
12346 (compute_scalar_evolution_in_loop): Remove and inline ...
12347 (analyze_scalar_evolution_1): ... here, replacing condition with
12348 what makes the intent clearer. Remove handling of cases
12349 get_scalar_evolution now handles.
12351 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12353 PR rtl-optimization/81434
12354 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12355 comment for main loop. In sched_group_found if, also add checks for
12356 pass and min_cost_group.
12358 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12360 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12361 (rs6000_insn_cost): New function.
12362 * config/rs6000/rs6000.md (cost): New attribute.
12364 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12365 H.J. Lu <hongjiu.lu@intel.com>
12369 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12370 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12371 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12372 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12373 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12374 OPTION_MASK_ISA_MMX and some other bit set require both
12375 mmx and the ISAs without the mmx bit.
12376 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12377 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12378 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12379 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12380 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12381 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12382 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12383 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12384 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12385 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12386 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12387 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12388 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12389 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12390 Add OPTION_MASK_ISA_MMX.
12392 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12394 * config.gcc (armv7*-*-freebsd*): New target.
12395 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12397 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12399 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12400 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12403 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12405 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12406 a template parameter.
12407 (WIDE_INT_REF_FOR): Update accordingly.
12408 * tree.h (wi::int_traits <const_tree>): Delete.
12409 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12410 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12411 (wi::tree_to_wide_ref): New typedef.
12412 (wi::to_wide): New function.
12413 * calls.c (get_size_range): Use wi::to_wide when operating on
12414 trees as wide_ints.
12415 * cgraph.c (cgraph_node::create_thunk): Likewise.
12416 * config/i386/i386.c (ix86_data_alignment): Likewise.
12417 (ix86_local_alignment): Likewise.
12418 * dbxout.c (stabstr_O): Likewise.
12419 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12420 * expr.c (const_vector_from_tree): Likewise.
12421 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12422 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12423 (fold_negate_expr_1, int_const_binop_1, const_binop)
12424 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12425 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12426 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12427 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12428 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12429 (fold_not_const, round_up_loc): Likewise.
12430 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12431 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12432 (alloca_call_type): Likewise.
12433 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12434 * godump.c (go_output_typedef): Likewise.
12435 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12436 * internal-fn.c (get_min_precision): Likewise.
12437 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12438 * ipa-polymorphic-call.c
12439 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12440 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12441 (ipa_modify_call_arguments): Likewise.
12442 * match.pd: Likewise.
12443 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12444 * print-tree.c (print_node_brief, print_node): Likewise.
12445 * stmt.c (expand_case): Likewise.
12446 * stor-layout.c (layout_type): Likewise.
12447 * tree-affine.c (tree_to_aff_combination): Likewise.
12448 * tree-cfg.c (group_case_labels_stmt): Likewise.
12449 * tree-data-ref.c (dr_analyze_indices): Likewise.
12450 (prune_runtime_alias_test_list): Likewise.
12451 * tree-dump.c (dequeue_and_dump): Likewise.
12452 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12453 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12454 * tree-pretty-print.c (dump_generic_node): Likewise.
12455 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12456 (simple_iv_with_niters): Likewise.
12457 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12458 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12459 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12460 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12461 (refine_value_range_using_guard, number_of_iterations_ne_max)
12462 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12463 (get_cst_init_from_scev, record_nonwrapping_iv)
12464 (scev_var_range_cant_overflow): Likewise.
12465 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12466 * tree-ssa-pre.c (compute_avail): Likewise.
12467 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12468 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12469 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12470 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12471 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12472 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12473 (dump_case_nodes, try_switch_expansion): Likewise.
12474 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12475 (vect_do_peeling): Likewise.
12476 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12477 * tree-vect-stmts.c (vectorizable_load): Likewise.
12478 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12479 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12480 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12481 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12482 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12483 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12484 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12485 (evrp_dom_walker::before_dom_children): Likewise.
12486 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12487 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12488 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12489 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12490 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12491 (get_range_pos_neg): Likewise.
12492 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12493 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12494 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12495 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12496 * config/avr/avr.c (avr_fold_builtin): Likewise.
12497 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12498 * config/msp430/msp430.c (msp430_attr): Likewise.
12499 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12500 * config/powerpcspe/powerpcspe-c.c
12501 (altivec_resolve_overloaded_builtin): Likewise.
12502 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12503 (rs6000_expand_ternop_builtin): Likewise.
12504 * config/rs6000/rs6000-c.c
12505 (altivec_resolve_overloaded_builtin): Likewise.
12506 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12507 (rs6000_expand_ternop_builtin): Likewise.
12508 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12510 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12512 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12513 when copying loop nest with only one inner loop.
12515 2017-10-10 Richard Biener <rguenther@suse.de>
12517 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12518 blocks if SCEV is active.
12519 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12521 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12522 (scev_initialize): Assert we are not yet initialized.
12524 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12526 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12527 inner loop's exit stmt by making it always exit the loop, otherwise
12528 we would generate an infinite empty loop.
12530 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12532 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12533 renaming variables in new preheader if it's deleted.
12535 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12537 * tree-loop-distribution.c (struct partition): Remove unused field
12538 loops of the structure.
12539 (partition_alloc, partition_free): Ditto.
12540 (build_rdg_partition_for_vertex): Ditto.
12542 2017-10-09 Jeff Law <law@redhat.com>
12544 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12545 return type to match prototype and documentation.
12547 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12549 * config/rs6000/rs6000.c (processor_costs): Move to ...
12550 * config/rs6000/rs6000.h: ... here.
12551 (rs6000_cost): Declare.
12553 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12555 * except.c (setjmp_fn): New global variable.
12556 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12557 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12558 if DONT_USE_BUILTIN_SETJMP is defined.
12560 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12562 * target.def (insn_cost): New hook.
12563 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12564 * doc/tm.texi: Regenerate.
12565 * rtlanal.c (insn_cost): Use the new hook.
12567 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12569 * combine.c (combine_validate_cost): Compute the new insn_cost,
12570 not just pattern_cost.
12571 (try_combine): Adjust comment.
12573 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12575 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12577 * combine.c (uid_insn_cost): Adjust comment.
12578 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12580 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12581 * dse.c (find_shift_sequence): Ditto.
12582 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12583 (bb_valid_for_noce_process_p): Use pattern_cost.
12584 * rtl.h (insn_rtx_cost): Delete.
12585 (pattern_cost): New prototype.
12586 (insn_cost): New prototype.
12587 * rtlanal.c (insn_rtx_cost): Rename to...
12588 (pattern_cost): ... this.
12591 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12593 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12594 (*jcc<mode>_0_r_i387): Ditto.
12595 (*jccxf_r_i387): Ditto.
12596 (*jcc<mode>_r_i387): Ditto.
12597 (*jccu<mode>_r_i387): Ditto.
12598 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12599 (*jcc): Rename from *jcc_1.
12601 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12603 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12604 deferred rescans after the lvx/stvx recombination pre-pass.
12606 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12608 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12609 memory operation instruction support.
12610 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12611 (rs6000-ibm-aix[789]*): Likewise.
12612 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12613 Document new functions.
12615 2017-10-09 Richard Biener <rguenther@suse.de>
12617 PR tree-optimization/82397
12618 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12619 equality only for semantically equal trees.
12621 2017-10-09 Richard Biener <rguenther@suse.de>
12623 PR tree-optimization/82449
12624 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12625 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12626 allow constant addresses.
12627 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12630 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12632 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12635 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12638 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12641 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12644 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12646 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12649 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12650 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12652 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12654 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12655 (WI_BINARY_PREDICATE_RESULT): Likewise.
12656 (wi::binary_traits::operator_result): New type.
12657 (wi::binary_traits::predicate_result): Likewise.
12658 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12659 (generic_wide_int::operator==, generic_wide_int::operator!=)
12660 (generic_wide_int::operator&, generic_wide_int::and_not)
12661 (generic_wide_int::operator|, generic_wide_int::or_not)
12662 (generic_wide_int::operator^, generic_wide_int::operator+
12663 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12665 (operator~, unary operator-, operator==, operator!=, operator&)
12666 (operator|, operator^, operator+, binary operator-, operator*): New
12668 * expr.c (get_inner_reference): Use wi::bit_and_not.
12669 * fold-const.c (fold_binary_loc): Likewise.
12670 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12671 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12672 (bit_value_binop): Likewise.
12673 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12674 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12675 (extract_range_from_binary_expr_1): Likewise.
12676 (masked_increment): Likewise.
12677 (simplify_bit_ops_using_ranges): Likewise.
12679 2017-10-09 Martin Jambor <mjambor@suse.cz>
12682 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12683 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12684 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12686 (hsa_fixup_mov_insn_type): New function.
12687 (hsa_op_with_type::get_in_type): Use it.
12688 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12689 immediates in an assert.
12690 (hsa_op_with_type::extend_int_to_32bit): New method.
12691 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12692 types. Convert to dest type if necessary.
12693 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12694 (reg_for_gimple_ssa): Pass false as min32int to
12695 hsa_type_for_scalar_tree_type.
12696 (gen_hsa_addr): Fixup type when creating addresable temporary.
12697 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12698 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12699 necessary. Call hsa_fixup_mov_insn_type.
12700 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12701 extend operands and convert to dest type if necessary.
12702 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12703 to dest type if necessary.
12704 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12705 if conversion nt necessary and size matches.
12706 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12707 to dest type if necessary.
12708 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12709 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12711 (gen_hsa_clrsb): Likewise.
12712 (gen_hsa_ffs): Likewise.
12713 (gen_hsa_divmod): Extend operands and convert to dest type if
12715 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12717 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12719 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12720 Remove empty default arguments. Use a brace block as output
12722 (conditional return): Ditto.
12724 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12725 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12727 (group_ending_nop): Ditto.
12728 (doloop_end): Ditto.
12729 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12730 (splitters for those): Ditto.
12732 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12734 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12735 a conditional jump (and the compare for it) so that pc_rtx is the
12737 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12738 for the deleted and renamed ctr<mode>_internal[234] patterns.
12739 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12740 Delete second conditional return pattern.
12741 (ctr<mode>_internal2): Delete this second bdnz pattern.
12742 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12743 (ctr<mode>_internal4): Delete this second bdz pattern.
12745 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12747 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12748 (always_initialized_rtx_for_ssa_name_p): New predicate.
12749 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12750 (finish_out_of_ssa): Free new field of SA.
12751 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12752 * tree-ssa-coalesce.c: Include tree-ssa.h.
12753 (get_parm_default_def_partitions): Remove extern keyword.
12754 (get_undefined_value_partitions): New function.
12755 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12756 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12757 uninitialized bits.
12758 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12760 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12762 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12764 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12766 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12767 for targets that preffer 128bit.
12769 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12771 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12773 2017-10-08 Olivier Hainque <hainque@adacore.com>
12775 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12776 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12778 (thumb_set_return_address): Likewise.
12780 2017-10-08 Olivier Hainque <hainque@adacore.com>
12782 * common/config/arm/arm-common.c (arm_except_unwind_info):
12783 Handle DWARF2_UNWIND_INFO.
12785 2017-10-07 Michael Collison <michael.collison@arm.com>
12787 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12790 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12792 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12793 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12794 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12795 defined, force the creation of a new block for a dispatch label.
12797 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12799 * invoke.texi (Wsuggest-attribute=cold): Document.
12800 * common.opt (Wsuggest-attribute=cold): New
12801 * ipa-pure-const.c (warn_function_cold): New function.
12802 * predict.c (compute_function_frequency): Use it.
12803 * predict.h (warn_function_cold): Declare.
12805 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12807 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12810 2017-10-06 Martin Liska <mliska@suse.cz>
12812 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12813 keyword for member functions.
12814 (struct sanopt_tree_couple): New struct.
12815 (struct sanopt_tree_couple_hash): New function.
12816 (struct sanopt_ctx): Add new hash_map.
12817 (has_dominating_ubsan_ptr_check): New function.
12818 (record_ubsan_ptr_check_stmt): Likewise.
12819 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12820 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12821 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12823 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12826 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12827 aarch64_simd_valid_immediate on CONST_VECTORs.
12828 (aarch64_reg_or_bic_imm): Likewise.
12830 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12832 PR rtl-optimization/82396
12833 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12835 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12837 * graphite-dependences.c (scop_get_reads): Move code to...
12838 (scop_get_must_writes): Move code to...
12839 (scop_get_may_writes): Move code to...
12840 (scop_get_reads_and_writes): ... here.
12841 (scop_get_dependences): Call scop_get_reads_and_writes.
12843 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12845 PR tree-optimization/82434
12846 * fold-const.h (can_native_encode_type_p,
12847 can_native_encode_string_p): Remove.
12848 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12849 don't encode anything, just return what would be otherwise returned.
12850 (native_encode_fixed, native_encode_complex, native_encode_vector):
12852 (native_encode_string): Likewise. Inline by hand
12853 can_native_encode_string_p.
12854 (can_native_encode_type_p): Remove.
12855 (can_native_encode_string_p): Remove.
12856 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12857 STRING_CSTs using can_native_encode_string_p, test all
12858 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12859 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12860 argument from native_encode_expr.
12861 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12862 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12865 2017-10-06 Richard Biener <rguenther@suse.de>
12867 PR tree-optimization/82397
12868 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12869 operand_equal_p but rely on data_ref_compare_tree for detecting
12871 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12872 to match up with dr_group_sort_cmp.
12874 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12877 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12879 * config/s390/s390-builtin-types.def: Regenerate.
12881 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12884 * config/s390/s390-builtin-types.def: Regenerate.
12885 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12886 Change flag from B_VXE to B_VX.
12887 (s390_vec_min_dbl): Remove B_VXE flag.
12889 2017-10-06 Richard Biener <rguenther@suse.de>
12891 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12892 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12893 translate_isl_ast_to_gimple::is_valid_rename,
12894 translate_isl_ast_to_gimple::get_rename,
12895 translate_isl_ast_to_gimple::get_def_bb_for_const,
12896 translate_isl_ast_to_gimple::get_new_name,
12897 translate_isl_ast_to_gimple::collect_all_ssa_names,
12898 translate_isl_ast_to_gimple::copy_loop_phi_args,
12899 translate_isl_ast_to_gimple::collect_all_ssa_names,
12900 translate_isl_ast_to_gimple::copy_loop_phi_args,
12901 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12902 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12903 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12904 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12905 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12906 translate_isl_ast_to_gimple::copy_cond_phi_args,
12907 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12908 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12909 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12910 translate_isl_ast_to_gimple::rename_uses,
12911 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12912 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12913 (set_rename_for_each_def): Likewise.
12914 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12915 here. Handle rewriting SCEV analyzable uses here.
12916 (copy_bb_and_scalar_dependences): Generate code for PHI
12918 (graphite_regenerate_ast_isl): Adjust.
12919 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12920 (add_write, add_read): New functions.
12921 (build_cross_bb_scalars_def): Use it and simplify.
12922 (build_cross_bb_scalars_use): Likewise.
12923 (graphite_find_cross_bb_scalar_vars): Inline into...
12924 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12925 simulating out-of-SSA. Compute liveout and add dependencies.
12926 (build_scops): Force an empty entry block.
12927 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12929 (sese_build_liveouts): Declare.
12930 (sese_trivially_empty_bb_p): Likewise.
12931 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12932 compute liveout and debug_liveout.
12933 (sese_bad_liveouts_use): Remove.
12934 (sese_reset_debug_liveouts_bb): Likewise.
12935 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12936 (sese_build_liveouts): Build liveout and debug_liveout and store
12938 (new_sese_info): Adjust.
12939 (free_sese_info): Likewise.
12940 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12941 do not build liveout here.
12942 (move_sese_in_condition): Adjust region entry.
12943 (scev_analyzable_p): Match up with chrec_apply requirements.
12944 (sese_trivially_empty_bb_p): New.
12945 * tree-into-ssa.c (get_reaching_def): Properly support generating
12946 default-defs for incremental rewrite of anonymous names.
12948 2017-10-06 Richard Biener <rguenther@suse.de>
12950 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12951 precision do not perform modulo reduction.
12953 2017-10-06 Richard Biener <rguenther@suse.de>
12955 PR tree-optimization/82436
12956 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12957 conservatively choose the vectorization factor when checking
12958 whether we can perform the required load permutation.
12959 (vect_transform_slp_perm_load): Assert when we may not fail.
12961 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12963 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12964 message for incompatible -msdata=* and -mcall-* options.
12966 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12968 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12969 rate for post-reload scheduling.
12971 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12973 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12975 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12977 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12978 to improve monte carlo in scimark.
12980 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12982 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12983 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12984 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12985 pentium4_cost, nocona_cost): Set reassociation width to 1.
12986 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12987 width to 2 for fp operations and 1 otherwise.
12988 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12990 (atom_cost): Set reassociation width to 2.
12991 (slm_cost, generic_cost): Set fp reassociation width
12992 to 2 and 1 otherwise.
12993 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12994 (core_cost): Set fp reassociation width to 4 and vector to 2.
12995 (ix86_reassociation_width): Rewrite using cost table; special case
12996 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12997 and TARGET_AVX128_OPTIMAL.
12998 * config/i386/i386.h (processor_costs): Add
12999 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13000 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13001 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13002 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13003 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13004 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13006 2017-10-05 Nathan Sidwell <nathan@acm.org>
13008 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13010 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13012 * config/arm/arm.c (arm_test_fpu_data): New.
13013 (arm_run_selftests): Call arm_test_fpu_data.
13015 2017-10-04 Nathan Sidwell <nathan@acm.org>
13017 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13019 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13021 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13023 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13024 check type for aarch64_simd_valid_immediate.
13025 (aarch64_output_simd_mov_immediate): Update prototype.
13026 (aarch64_simd_valid_immediate): Update prototype.
13027 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13028 support for ORR-immediate.
13029 (and<mode>3): modified pattern to add support for BIC-immediate.
13030 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13031 now checks for valid immediate for BIC and ORR based on new enum
13033 (aarch64_output_simd_mov_immediate): Function now used to output
13034 BIC/ORR imm as well based on new enum argument.
13035 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13037 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13038 (aarch64_reg_or_bic_imm): Likewise.
13040 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13042 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13043 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13045 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13049 PR rtl-optimization/82396
13050 * haifa-sched.c (autopref_multipass_init): Simplify
13052 (autopref_rank_data): Simplify sort order.
13053 * sched-int.h (autopref_multipass_data_): Remove
13054 multi_mem_insn_p, min_offset and max_offset.
13056 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13058 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13060 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13062 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13063 vect_doubleint_cvt.
13065 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13067 * doc/sourcebuild.texi: Document vect_long_mult.
13069 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13071 PR tree-optimization/82413
13072 * fold-const.c (build_range_check): Use widest_int when comparing
13073 the maximum ETYPE value with HIGH.
13075 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13077 PR rtl-optimization/82396
13078 * haifa-sched.c (autopref_multipass_init): Simplify
13080 (autopref_rank_data): Simplify sort order.
13081 * sched-int.h (autopref_multipass_data_): Remove
13082 multi_mem_insn_p, min_offset and max_offset.
13084 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13086 PR tree-optimization/82381
13087 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13088 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13091 PR tree-optimization/82374
13092 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13093 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13094 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13095 current_function_decl to the new decl.
13097 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13099 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13100 helper macro for IEEE float128 hardware built-in functions.
13101 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13103 (TRUNCF128_ODD): Likewise.
13104 (ADDF128_ODD): Likewise.
13105 (SUBF128_ODD): Likewise.
13106 (MULF128_ODD): Likewise.
13107 (DIVF128_ODD): Likewise.
13108 (FMAF128_ODD): Likewise.
13109 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13110 UNSPEC_TRUNC_ROUND_TO_ODD.
13111 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13112 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13113 floating point round to odd instructions.
13114 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13115 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13116 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13117 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13118 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13119 (trunc<mode>sf2_hw): Change the truncate with round to odd
13120 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13121 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13122 to odd hardware instructions.
13123 (sub<mode>3_odd): Likewise.
13124 (mul<mode>3_odd): Likewise.
13125 (div<mode>3_odd): Likewise.
13126 (sqrt<mode>2_odd): Likewise.
13127 (fma<mode>4_odd): Likewise.
13128 (fms<mode>4_odd): Likewise.
13129 (nfma<mode>4_odd): Likewise.
13130 (nfms<mode>4_odd): Likewise.
13131 (trunc<mode>df2_odd): Change the truncate with round to odd
13132 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13134 * doc/extend.texi (PowerPC built-in functions): Update documentation
13135 for existing IEEE float128-bit built-in functions. Add built-in
13136 functions that generate the IEEE 128-bit floating point round to
13139 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13141 PR rtl-optimization/77729
13142 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13143 to (X&(C1&~C2))|C2 transformations.
13145 2017-10-03 Martin Jambor <mjambor@suse.cz>
13147 PR tree-optimization/82363
13148 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13149 mismatch, mark lacc written regardless of racc.
13151 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13153 PR tree-optimization/82381
13154 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13155 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13156 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13157 fallthrough into reassoc_stmt_dominates_stmt_p.
13160 * combine.c (combine_instructions): Don't combine in unreachable
13163 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13166 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13167 function to not use the have_cpu variable. Do not set cpu_index,
13168 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13169 or the default cpu.
13170 (rs6000_valid_attribute_p): Remove duplicate initializations of
13171 old_optimize and func_optimize.
13172 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13173 (rs6000_activate_target_options): Make global.
13174 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13177 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13179 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13180 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13181 Set *poffset to 0 and *psize and *pmax_size to -1 if
13182 *poffset + *psize overflows in HOST_WIDE_INT.
13184 PR tree-optimization/82387
13185 PR tree-optimization/82388
13186 PR tree-optimization/82389
13187 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13188 instead of live_bytes non-NULL.
13190 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13193 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13196 2017-10-02 Richard Biener <rguenther@suse.de>
13198 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13199 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13200 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13202 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13204 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13205 requested precision matches the type's.
13206 * calls.c (alloc_max_size): Calculate the new candidate size as
13207 a widest_int and use wi::to_widest when comparing it with the
13208 current candidate size.
13209 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13210 zero rather than integer_zero_node.
13211 * match.pd: Check for a no-op conversion before using wi::add
13212 rather than after. Use tree_to_uhwi when summing small shift
13213 counts into an unsigned int.
13215 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13216 Alan Hayward <alan.hayward@arm.com>
13217 David Sherwood <david.sherwood@arm.com>
13220 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13221 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13222 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13223 POINTER_AND_FP_REGS.
13225 2017-10-02 Richard Biener <rguenther@suse.de>
13227 PR tree-optimization/82355
13228 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13229 a mapping for the enclosing loop but avoid generating one for
13230 the loop tree root.
13231 (copy_bb_and_scalar_dependences): Remove premature codegen
13232 error on PHIs in blocks duplicated into multiple places.
13233 * graphite-scop-detection.c
13234 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13235 in the region use it as loop and nest to analyze the DR in.
13236 (try_generate_gimple_bb): Likewise.
13237 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13238 (add_loop_constraints): For blocks in a loop not in the region
13239 create a dimension with a single iteration.
13240 * sese.h (gbb_loop_at_index): Remove assert.
13242 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13244 * omp-expand.c (adjust_context_scope): New function.
13245 (expand_parallel_call): Call adjust_context_scope.
13247 2017-10-01 Jeff Law <law@redhat.com>
13249 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13250 dom_opt_dom_walker class with direct access to private members.
13251 Add comments. Call test_for_singularity.
13252 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13253 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13254 m_dummy_cond anymore.
13255 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13257 (pass_dominator:execute): Build the dummy_cond here and pass it
13258 to the dom_opt_dom_walker ctor.
13259 (test_for_singularity): New function.
13261 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13262 Maya Rashish <coypu@sdf.org>
13264 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13265 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13266 (alpha*-*-netbsd*) Use nbsd_tm_file.
13267 (arm*-*-netbsdelf*) Likewise.
13268 (i[34567]86-*-netbsdelf*) Likewise.
13269 (x86_64-*-netbsd*) Likewise.
13270 (mips*-*-netbsd*) Likewise.
13271 (powerpc-*-netbsd*) Likewise.
13272 (sh*-*-netbsd*) Likewise.
13273 (sparc-*-netbsdelf*) Likewise.
13274 (sparc64-*-netbsd*) Likewise.
13275 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13277 (vax-*-netbsdelf*) Likewise.
13278 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13279 (UINT_FAST8_TYPE) Likewise.
13280 (INT_FAST16_TYPE) Check CHAR_FAST16.
13281 (UINT_FAST16_TYPE) Likewise.
13283 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13286 * config/i386/i386.md
13287 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13288 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13289 *divmodsi4_zext_2): New define_insn_and_split.
13290 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13291 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13292 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13293 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13294 New define_insn_and_split.
13295 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13296 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13297 operands[1] having DImode when mode is SImode.
13299 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13300 always SImode for DIV and MOD in REG_EQUAL notes.
13302 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13304 PR middle-end/82319
13305 * match.pd: Fix handling of NaNs in pattern.
13307 2017-09-29 Jeff Law <law@redhat.com>
13309 * sbitmap.c (bitmap_bit_in_range_p): New function.
13310 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13311 * tree-ssa-dse.c (live_bytes_read): New function.
13312 (dse_classify_store): Ignore reads of dead bytes.
13314 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13315 typos and whitespace errors.
13316 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13317 * config/s390/s390.c (s390_emit_prologue): Likewise.
13319 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13322 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13323 with a symbol for LRA.
13325 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13327 PR rtl-optimization/82338
13328 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13330 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13332 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13333 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13334 (qsort_chk): Declare.
13335 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13336 (qsort_chk): New function.
13338 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13340 PR tree-optimization/82337
13341 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13342 phi definition if the PHI result appears in an abnormal PHI.
13343 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13344 basis appears in an abnormal PHI.
13346 2017-09-29 Richard Biener <rguenther@suse.de>
13348 * graphite-isl-ast-to-gimple.c
13349 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13350 (binary_op_to_tree): Use it.
13351 (get_rename_from_scev): Likewise.
13352 (copy_loop_phi_nodes): Likewise.
13353 (copy_bb_and_scalar_dependences): Likewise.
13354 (translate_pending_phi_nodes): Likewise.
13356 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13359 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13360 for movabsq $(i32 << shift), r64.
13362 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13364 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13365 index when encoding %esp as %rsp to avoid 0x67 prefix.
13367 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13369 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13370 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13372 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13374 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13375 Extensions with more than 16 double VFP registers.
13376 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13377 to_clear_mask and all code related to it. Replace the remaining
13378 entry by a sbitmap and adapt code accordingly.
13380 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13382 * brig-builtins.def: Change pure attributes to const.
13384 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13386 * config.gcc (default_gnu_indirect_function): Default to yes for
13387 sparc*-*-linux* with glibc.
13389 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13391 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13392 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13393 when creating .init_array and .fini_array sections with priority
13396 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13399 * config/aarch64/aarch64.c
13400 (aarch64_builtin_support_vector_misalignment): Always return false
13401 when misalignment is unknown.
13403 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13405 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13406 this function to return false if the definition used by the swap
13407 instruction is artificial, or if the memory address from which the
13408 constant value is loaded is not represented by a base address held
13409 in a register or if the base address register is a frame or stack
13410 pointer. Additionally, return false if the base address of the
13411 loaded constant is a SYMBOL_REF but is not considered to be a
13413 (replace_swapped_load_constant): New function.
13414 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13415 loaded constant vector with a load of a swapped constant vector.
13417 2017-09-27 Carl Love <cel@us.ibm.com>
13419 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13420 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13421 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13424 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13426 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13427 first, always call autopref_rank_data otherwise.
13429 2017-09-27 Richard Biener <rguenther@suse.de>
13431 * graphite-scop-detection.c (find_scop_parameters): Move
13432 loop bound handling ...
13433 (gather_bbs::before_dom_children): ... here, avoiding the need
13434 to build scop_info->loop_nest.
13435 (record_loop_in_sese): Remove.
13436 * sese.h (sese_info_t::loop_nest): Remove.
13437 * sese.c (new_sese_info): Do not allocate loop_nest.
13438 (free_sese_info): Do not free loop_nest.
13440 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13443 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13444 lhs from calls if the lhs has addressable type.
13446 2017-09-27 Richard Biener <rguenther@suse.de>
13448 * graphite.h (scop::max_alias_set): New member.
13449 * graphite-scop-detection.c: Remove references to non-existing
13450 --param in comments.
13451 (build_alias_sets): Record the maximum alias set used for drs.
13452 (build_scops): Support zero as unlimited for
13453 --param graphite-max-arrays-per-scop.
13454 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13455 and inline into ...
13456 (build_poly_sr_1): ... here. Compute alias set based on the
13457 maximum alias set used for drs rather than
13458 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13460 2017-09-27 Richard Biener <rguenther@suse.de>
13462 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13463 --param loop-block-tile-size=0 to disable tiling.
13465 2017-09-27 Richard Biener <rguenther@suse.de>
13467 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13468 (graphite-max-nb-scop-params): Document special value zero.
13469 * domwalk.h (dom_walker::STOP): New symbolical constant.
13470 (dom_walker::dom_walker): Add optional parameter for bb to
13472 (dom_walker::~dom_walker): Declare.
13473 (dom_walker::before_dom_children): Document STOP return value.
13474 (dom_walker::m_user_bb_to_rpo): New member.
13475 (dom_walker::m_bb_to_rpo): Likewise.
13476 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13477 mapping here if not provided by the user.
13478 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13479 provided by the user.
13480 (dom_walker::STOP): Define.
13481 (dom_walker::walk): Do not compute bb to RPO mapping here.
13482 Support STOP return value from before_dom_children to stop
13484 * graphite-optimize-isl.c (optimize_isl): If the schedule
13485 is the same still generate code if -fgraphite-identity
13486 or -floop-parallelize-all are given.
13487 * graphite-scop-detection.c: Include cfganal.h.
13488 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13490 (gather_bbs::before_dom_children): Return STOP for BBs
13492 (build_scops): Compute bb to RPO mapping and pass it to
13493 the domwalk. Treat --param graphite-max-nb-scop-params=0
13494 as not limiting the number of params.
13495 * graphite.c (graphite_initialize): Remove limit on the
13496 number of basic-blocks in a function.
13497 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13498 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13499 default value of 10.
13501 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13503 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13504 Adjust code to eliminate needing to do the shift right 32-bits
13505 operation after XSCVDPSPN.
13507 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13509 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13510 ((X / Y) != 0 -> X >= Y): Likewise.
13512 2017-09-26 Carl Love <cel@us.ibm.com>
13514 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13515 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13516 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13517 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13518 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13519 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13520 definitions and overloading.
13521 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13522 statement for P9V_BUILTIN_XST_LEN_R.
13523 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13524 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13525 define_expand and define_insn for the instructions and builtins.
13526 * doc/extend.texi: Update the built-in documentation file for the new
13527 built-in functions.
13528 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13529 define_insn for the instructions
13531 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13534 * gcc/config/netbsd-protos.h: New file.
13535 * gcc/config/netbsd.c: New file.
13536 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13537 * gcc/config/t-netbsd: New file.
13538 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13539 (tmake_file) Add t-netbsd.
13540 (extra_objs) Add netbsd.o.
13542 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13546 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13548 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13550 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13551 sign extension from a vector register to a GPR by doing a 32-bit
13552 direct move and then an EXTSW.
13553 (extendsi<mode>2 splitter): Likewise.
13554 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13555 right or vector extract after doing XSCVDPSPN. Use
13556 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13558 (movdi_from_sf_zero_ext): Likewise.
13559 (reload_gpr_from_vsxsf): Likewise.
13560 (p8_mfvsrd_4_disf): Delete, no longer used.
13561 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13562 then needing to move the SFmode to a GPR to use the XSCVDPSP
13563 instruction instead of FRSP and XSCVDPSPN.
13564 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13565 it is adjacent to the other XSCVSPDP insns.
13566 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13567 SFmode to be in traditional Altivec registers.
13568 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13569 (vsx_xscvspdpn): Likewise.
13570 (vsx_xscvspdpn_scalar): Likewise.
13572 2017-09-26 Martin Jambor <mjambor@suse.cz>
13574 * tree-sra.c (compare_access_positions): Put integral types first,
13575 stabilize sorting of integral types, remove conditions putting
13576 non-full-precision integers last.
13577 (sort_and_splice_var_accesses): Disable scalarization if a
13578 non-integert would be represented by a non-full-precision integer.
13580 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13582 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13583 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13584 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13585 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13586 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13587 conditionals inside the function instead of around it. Call
13588 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13589 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13591 2017-09-26 Richard Biener <rguenther@suse.de>
13593 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13595 (scop_detection::build_scop_breadth): ... this. Removed.
13596 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13597 (scop_detection::harmful_stmt_in_bb): Likewise.
13598 (scop_detection::graphite_can_represent_stmt): Likewise.
13599 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13600 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13601 (scop_detection::can_represent_loop_1): ... this. Removed.
13602 (scop_detection::harmful_loop_in_region): Simplify after inlining
13603 the above and remove more quadraticness.
13604 (build_scops): Adjust.
13605 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13608 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13611 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13612 REGNO (base) == SP_REG if base is a REG.
13614 PR middle-end/35691
13615 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13616 if it is different SSA_NAME.
13617 (optimize_range_tests_cmp_bitwise): New function.
13618 (optimize_range_tests): Call it.
13620 2017-09-26 Richard Biener <rguenther@suse.de>
13622 PR tree-optimization/82321
13623 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13624 for the def being inside the loop.
13626 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13628 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13630 * config/s390/s390-builtins.def: Fix constraint on op4.
13632 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13634 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13635 independent expanders.
13636 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13637 ("vec_ordered", "vec_unordered"): New expanders.
13639 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13641 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13644 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13646 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13647 vec_unpacks_lo_v16qi.
13648 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13650 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13652 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13653 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13654 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13656 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13658 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13660 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13662 ("*vec_slb<mode>"): New insn pattern.
13663 ("vec_shr_<mode>"): New expander.
13664 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13665 and force the shift count operand to V16QImode.
13666 ("vec_srb<mode>"): Set shift count mode to V16QI.
13668 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13670 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13671 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13672 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13674 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13677 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13679 2017-09-26 Richard Biener <rguenther@suse.de>
13681 PR tree-optimization/82320
13682 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13685 2017-09-25 Jeff Law <law@redhat.com>
13687 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13688 prototype for new argument.
13689 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13690 mostly extracted from rs6000_emit_allocate_stack.
13691 (rs6000_emit_probe_stack_range_stack_clash): New function.
13692 (rs6000_emit_allocate_stack): Call
13693 rs6000_emit_probe_stack_range_stack_clash as needed.
13694 (rs6000_emit_probe_stack_range): Add additional argument
13695 to call to gen_probe_stack_range{si,di}.
13696 (output_probe_stack_range): New.
13697 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13698 (output_probe_stack_range_stack_clash): New.
13699 (rs6000_emit_prologue): Emit notes into dump file as requested.
13700 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13701 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13702 Add additional operand and pass it to output_probe_stack_range.
13704 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13706 PR tree-optimization/82163
13707 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13708 (checking_verify_loop_closed_ssa): New parameter.
13709 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13710 (check_loop_closed_ssa_stmt): Delete.
13711 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13712 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13713 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13716 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13718 * brig-builtins.def: Treat HSAIL barrier builtins as
13719 setjmp/longjump style functions.
13721 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13723 * target.def (constant_alignment): New hook.
13724 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13725 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13726 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13727 * doc/tm.texi: Regenerate.
13728 * targhooks.h (default_constant_alignment): Declare.
13729 (constant_alignment_word_strings): Likewise.
13730 * targhooks.c (default_constant_alignment): New function.
13731 (constant_alignment_word_strings): Likewise.
13732 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13733 instead of CONSTANT_ALIGNMENT.
13734 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13735 (force_const_mem): Likewise.
13736 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13737 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13738 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13739 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13740 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13742 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13743 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13744 constant_alignment_word_strings.
13745 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13746 (CONSTANT_ALIGNMENT): Likewise.
13747 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13748 (arm_constant_alignment): New function.
13749 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13750 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13751 constant_alignment_word_strings.
13752 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13753 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13754 constant_alignment_word_strings.
13755 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13756 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13757 (cris_constant_alignment): New function.
13758 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13759 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13760 (epiphany_constant_alignment): New function.
13761 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13762 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13763 constant_alignment_word_strings.
13764 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13765 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13766 constant_alignment_word_strings.
13767 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13768 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13769 constant_alignment_word_strings.
13770 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13771 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13772 * config/i386/i386.c (ix86_constant_alignment): Make static.
13773 Use the same interface as the target hook.
13774 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13775 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13776 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13777 constant_alignment_word_strings.
13778 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13779 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13780 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13781 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13782 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13783 constant_alignment_word_strings.
13784 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13785 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13786 constant_alignment_word_strings.
13787 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13788 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13789 constant_alignment_word_strings.
13790 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13791 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13793 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13794 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13795 * config/mips/mips.c (mips_constant_alignment): New function.
13796 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13797 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13798 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13799 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13800 (mmix_constant_alignment): Make static. Use the same interface
13801 as the target hook.
13802 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13803 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13804 constant_alignment_word_strings.
13805 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13806 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13807 constant_alignment_word_strings.
13808 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13809 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13810 constant_alignment_word_strings.
13811 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13812 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13813 (rs6000_constant_alignment): New function.
13814 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13815 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13816 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13817 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13818 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13819 (rs6000_constant_alignment): New function.
13820 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13821 * config/s390/s390.c (s390_constant_alignment): New function.
13822 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13823 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13824 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13825 constant_alignment_word_strings.
13826 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13827 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13828 (sparc_constant_alignment): New function.
13829 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13830 * config/spu/spu.c (spu_constant_alignment): New function.
13831 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13832 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13833 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13834 constant_alignment_word_strings.
13835 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13836 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13837 constant_alignment_word_strings.
13838 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13839 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13840 constant_alignment_word_strings.
13841 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13842 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13843 (visium_constant_alignment): New function.
13844 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13845 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13846 (xtensa_constant_alignment): New function.
13847 * system.h (CONSTANT_ALIGNMENT): Poison.
13849 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13851 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13852 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13853 (rs6000_builtin_valid_without_lhs): New helper function.
13854 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13855 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13857 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13859 * target.h (vec_perm_indices): Use unsigned short rather than
13861 (auto_vec_perm_indices): Likewise.
13862 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13863 Use unsigned int rather than unsigned char.
13864 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13866 2017-09-25 Richard Biener <rguenther@suse.de>
13868 * cfgloop.h (sort_sibling_loops): Declare.
13869 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13870 (sort_sibling_loops): New function sorting the sibling loop list
13872 * graphite.c (graphite_transform_loops): Sort sibling loops.
13874 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13876 * target.def (vec_perm_const_ok): Change sel parameter to
13878 * optabs-query.c (can_vec_perm_p): Update accordingly.
13879 * doc/tm.texi: Regenerate.
13880 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13881 auto_vec_perm_indices and remove separate nelt field.
13882 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13883 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13884 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13885 (aarch64_expand_vec_perm_const): Update accordingly.
13886 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13887 to vec_perm_indices.
13888 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13889 auto_vec_perm_indices and remove separate nelt field.
13890 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13891 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13892 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13894 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13895 to vec_perm_indices.
13896 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13897 sel to vec_perm_indices.
13898 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13899 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13900 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13902 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13905 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13908 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13909 on the FUNCTION_DECL function context if it has a DIE that is a
13912 2017-09-25 Richard Biener <rguenther@suse.de>
13914 PR tree-optimization/82285
13915 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13918 2017-09-25 Tom de Vries <tom@codesourcery.com>
13922 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13925 2017-09-25 Richard Biener <rguenther@suse.de>
13927 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13928 ISL errors other than isl_error_quota happen. Dump if the
13929 schedule is the same.
13930 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13931 errors instead of aborting inside ISL.
13933 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13936 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13937 of libgcc_eh for m64.
13938 * config/i386/darwin64.h: Likewise.
13940 2017-09-25 Richard Biener <rguenther@suse.de>
13942 PR middle-end/82144
13943 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13944 attribute for incomplete types nor twice for complete ones.
13946 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13949 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13950 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13953 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13956 * config/i386/i386.opt (mprefer-avx256): Use
13957 ix86_target_flags variable.
13958 * config/i386/i386.c (ix86_target_string): Move
13959 -mprefer-avx256 to flag2_opts.
13961 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13963 PR middle-end/35691
13964 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13965 and x != -1 | y != -1 into (x & y) != -1.
13967 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13969 * config.gcc: Add new case statement to set
13970 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13971 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13972 s390x-*-linux* case statements. Added aarch64 to the list of
13973 supported architectures.
13975 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13977 PR tree-optimization/82289
13978 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13979 STMT_VINFO_RELEVANT_P.
13981 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13982 Alan Hayward <alan.hayward@arm.com>
13983 David Sherwood <david.sherwood@arm.com>
13985 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13986 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13987 (extract_range_from_binary_expr_1): Don't call
13988 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13990 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13991 Alan Hayward <alan.hayward@arm.com>
13992 David Sherwood <david.sherwood@arm.com>
13994 * target.def (preferred_vector_alignment): New hook.
13995 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13997 * doc/tm.texi: Regenerate.
13998 * targhooks.h (default_preferred_vector_alignment): Declare.
13999 * targhooks.c (default_preferred_vector_alignment): New function.
14000 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14002 (DR_TARGET_ALIGNMENT): New macro.
14003 (aligned_access_p): Update commentary.
14004 (vect_known_alignment_in_bytes): New function.
14005 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14007 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14008 Calculate the misalignment based on the target alignment rather than
14010 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14011 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14012 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14013 the target alignment, rather than masking the element misalignment
14014 with the number of elements in a vector. Also use the target
14015 alignment when calculating the maximum number of peels.
14016 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14017 instead of TYPE_ALIGN_UNIT.
14018 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14019 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14020 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14021 (vect_create_data_ref_ptr): Likewise.
14022 (vect_setup_realignment): Realign by ANDing with
14023 -DR_TARGET_MISALIGNMENT.
14024 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14025 the number of peels based on DR_TARGET_ALIGNMENT.
14026 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14027 with the guaranteed alignment boundary when deciding whether
14029 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14030 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14031 (ensure_base_align): Remove stmt_info parameter. Get the
14032 target base alignment from DR_TARGET_ALIGNMENT.
14033 (vectorizable_store): Update call accordingly. Interpret
14034 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14036 (vectorizable_load): Likewise.
14038 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14039 Alan Hayward <alan.hayward@arm.com>
14040 David Sherwood <david.sherwood@arm.com>
14042 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14043 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14044 (vect_enhance_data_refs_alignment): Likewise.
14046 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14048 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14049 error. Only quit immediately if parsing is complete.
14050 (BEGIN): Initialize fatal_err and parse_done.
14051 (begin fpu, end fpu): Check number of arguments.
14052 (begin arch, end arch): Likewise.
14053 (begin cpu, end cpu): Likewise.
14054 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14055 (optalias): Likewise.
14057 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14059 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14060 * config/arm/arm-isa.h: Delete. Move definitions to ...
14061 * arm-cpus.in: ... here. Use new feature and fgroup values.
14062 * config/arm/arm.c (arm_option_override): Use lower case for feature
14064 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14065 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14066 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14068 (print_isa_bits_for): New function.
14069 (gen_isa): New function.
14070 (gen_comm_data): Use print_isa_bits_for.
14071 (define feature): New keyword.
14072 (define fgroup): New keyword.
14073 * config/arm/t-arm (TM_H): Remove.
14074 (GTM_H): Add arm-isa.h.
14075 (arm-isa.h): Add rule to generate file.
14076 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14077 case for feature bit names.
14079 2017-09-22 Richard Biener <rguenther@suse.de>
14081 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14083 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14084 print of no dependency loops ...
14085 * graphite.c (graphite_transform_loops): ... here.
14086 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14088 (same_close_phi_node, remove_duplicate_close_phi,
14089 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14090 (canonicalize_loop_closed_ssa): ... here and simplify.
14091 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14092 (optimize_isl): Use dump_printf_loc to tell when we stopped
14093 optimizing because of an ISL timeout.
14095 2017-09-22 Richard Biener <rguenther@suse.de>
14097 PR tree-optimization/82291
14098 * tree-if-conv.c (predicate_mem_writes): Make sure to
14099 remove writes in blocks predicated with false.
14101 2017-09-22 Richard Biener <rguenther@suse.de>
14103 * sese.c: Include cfganal.h.
14104 (if_region_set_false_region): Remove.
14105 (create_if_region_on_edge): Likewise.
14106 (move_sese_in_condition): Re-implement without destroying
14109 2017-09-22 Richard Biener <rguenther@suse.de>
14111 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14112 Verify both BBs contain loop PHI nodes before dispatching to
14113 copy_loop_phi_args.
14114 (graphite_regenerate_ast_isl): Do not recompute dominators,
14115 do not verify three times. Restructure for clarity.
14116 * graphite-scop-detection.c (same_close_phi_node,
14117 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14118 defined_in_loop_p, canonicalize_loop_closed_ssa,
14119 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14120 checking and SSA rewrite, move to ...
14121 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14122 (graphite_initialize): Do not pass in ctx, do not reset the
14123 SCEV cache, compute only dominators.
14124 (graphite_transform_loops): Allocate ISL ctx after
14125 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14126 Maintain post-dominators only around build_scops.
14127 * sese.c (if_region_set_false_region): Make static. Free
14128 and recompute dominators.
14129 (move_sese_in_condition): Assert we don't get called with
14130 post-dominators computed.
14131 * sese.h (if_region_set_false_region): Remove.
14133 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14135 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14136 mode attribute for TARGET_AVX512VL.
14138 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14140 * config/i386/i386.opt (mprefer-avx256): New option.
14141 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14143 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14144 for TARGET_PREFER_AVX256.
14145 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14147 2017-09-21 Jeff Law <law@redhat.com>
14149 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14150 Fix dump output if the only stack space is for pushed registers.
14152 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14154 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14157 2017-09-21 Martin Sebor <msebor@redhat.com>
14160 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14161 code (in C++) or code that triggers warnings.
14163 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14165 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14167 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14169 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14170 * sched-rgn.c: Ditto.
14171 * sel-sched-ir.c: Ditto.
14173 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14175 * toplev.h (set_random_seed): Adjust return type.
14176 * toplev.c (init_local_tick): Move eager initialization of random_seed
14177 to get_random_seed. Adjust comment.
14178 (init_random_seed): Inline to get_random_seed, delete.
14179 (get_random_seed): Initialize random_seed lazily.
14180 (set_random_seed): Do not return previous value.
14181 (print_switch_value): Do not call get_random_seed.
14183 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14185 * cgraph.c (delete_function_version): New, broken out from...
14186 (cgraph_node::delete_function_version): ...here. Rename to
14187 cgraph_node::delete_function_version_by_decl. Update all uses.
14188 (cgraph_node::remove): Call delete_function_version.
14190 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14193 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14194 VAR_DECLs to which addressable non-volatile parameters are mapped
14195 and for id->retvar after the return value assignment. Clear
14196 id->retval and id->retbnd after inlining.
14198 2017-09-21 Richard Biener <rguenther@suse.de>
14200 PR tree-optimization/82276
14201 PR tree-optimization/82244
14202 * tree-vrp.c (build_assert_expr_for): Set
14203 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14205 (remove_range_assertions): Revert earlier change.
14207 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14210 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14212 2017-09-21 Richard Biener <rguenther@suse.de>
14214 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14215 Restore valid IL after code generation errors.
14216 * graphite.c (graphite_transform_loops): Diagnose code
14217 generation issues as MSG_MISSED_OPTIMIZATION and continue
14218 with processing SCOPs.
14220 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14221 Alan Hayward <alan.hayward@arm.com>
14222 David Sherwood <david.sherwood@arm.com>
14224 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14225 rather than choosing between plus_constant and gen_rtx_<CODE>.
14226 * expr.c (emit_push_insn): Likewise.
14227 (expand_expr_real_2): Likewise.
14229 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14230 Alan Hayward <alan.hayward@arm.com>
14231 David Sherwood <david.sherwood@arm.com>
14233 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14235 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14236 Alan Hayward <alan.hayward@arm.com>
14237 David Sherwood <david.sherwood@arm.com>
14239 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14240 calling tree_to_uhwi.
14242 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14243 Alan Hayward <alan.hayward@arm.com>
14244 David Sherwood <david.sherwood@arm.com>
14246 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14247 INTEGER_CST rather than a negative test for ADDR_EXPR.
14249 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14250 Alan Hayward <alan.hayward@arm.com>
14251 David Sherwood <david.sherwood@arm.com>
14253 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14254 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14256 2017-09-21 Richard Biener <rguenther@suse.de>
14258 PR tree-optimization/71351
14259 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14260 graphite_create_new_loop_guard): Remove, fold remaining parts
14262 (translate_isl_ast_node_for): ... here and simplify.
14264 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14267 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14268 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14269 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14270 alternative always use QI mode, for -Os imov (=R,R) alternative
14271 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14274 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14275 Jeff Law <law@redhat.com>
14277 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14278 (allocate_stack_space): New function, partially extracted from
14279 s390_emit_prologue.
14280 (s390_emit_prologue): Track offset to most recent stack probe.
14281 Code to allocate space moved into allocate_stack_space.
14282 Dump actions when no stack is allocated.
14283 (s390_prologue_plus_offset): New function.
14284 (s390_emit_stack_probe): Likewise.
14286 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14288 * common.opt (Wa, Wl, Wp, g, gz=): Add
14290 (gno-column-info): Remove.
14291 (gcolumn-info): Drop RejectNegative.
14292 (gno-): New prefix.
14293 (gno-record-gcc-switches): Remove.
14294 (grecord-gcc-switches): Drop RejectNegative.
14295 (gno-split-dwarf): Remove.
14296 (gsplit-dwarf): Drop RejectNegative.
14297 (gno-strict-dwarf): Remove.
14298 (gstrict-dwarf): Drop RejectNegative.
14299 * config/darwin.opt (gfull, gused): Add RejectNegative.
14300 * dwarf2out.c (gen_producer_string): Drop
14301 gno-record-gcc-switches handler.
14302 * optc-gen.awk: Add g to prefixes with negative forms.
14303 * opts-common.c (remapping_prefix_p): New.
14304 (find_opt): Check it.
14305 (generate_canonical_option): Test g prefix.
14306 (option_map): Add -gno- mapping.
14307 (add_misspelling_candidates): Check remapping_prefix_p.
14309 2017-09-20 Jeff Law <law@redhat.com>
14311 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14312 thinko in stack clash protection support.
14314 * explow.c (compute_stack_clash_protection_loop_data): Use
14315 CONST_INT_P instead of explicit test. Verify object is a
14316 CONST_INT_P before looking at INTVAL.
14317 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14318 instead of explicit test.
14320 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14323 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14324 address instead of to r1 and r11.
14326 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14328 * config.gcc: Support "knm".
14329 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14330 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14332 * config/i386/i386.c (m_KNM): Define.
14333 (processor_target_table): Add "knm".
14335 (ix86_option_override_internal): Add "knm".
14336 (ix86_issue_rate): Add PROCESSOR_KNM.
14337 (ix86_adjust_cost): Ditto.
14338 (ia32_multipass_dfa_lookahead): Ditto.
14339 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14340 (fold_builtin_cpu): Add M_INTEL_KNM.
14341 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14342 (processor_type): Add PROCESSOR_KNM.
14343 * config/i386/x86-tune.def: Add m_KNM.
14344 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14346 2017-09-20 Richard Biener <rguenther@suse.de>
14348 PR tree-optimization/80213
14349 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14350 are allowed in empty BBs as well.
14351 (canonicalize_loop_closed_ssa): Also look for other complex
14353 (scop_detection::get_sese): Include the loop-closed PHI block
14355 (scop_detection::merge_sese): Remove code adding extra blocks.
14356 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14357 (build_scops): Assert the final returned scop is invalid.
14359 2017-09-20 Richard Biener <rguenther@suse.de>
14361 PR tree-optimization/82264
14362 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14364 (vn_phi_lookup): Likewise.
14365 (vn_phi_insert): Likewise.
14367 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14369 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14370 that fits into uhwi or shwi, add DW_AT_const_value regardless
14371 of early_dwarf without going through RTL, using add_AT_unsigned
14374 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14375 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14376 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14377 (DEBUG_MACRO_SECTION): Likewise.
14378 (DEBUG_LTO_MACRO_SECTION): Likewise.
14379 (DEBUG_STR_DWO_SECTION): Likewise.
14380 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14381 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14382 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14383 (DEBUG_LTO_LINE_STR_SECTION): Define.
14384 (init_sections_and_labels): Initialize debug_line_str_section
14385 variable. Initialize debug_loc_section for -gdwarf-5 to
14386 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14388 2017-09-20 Richard Biener <rguenther@suse.de>
14390 * graphite-sese-to-poly.c (extract_affine): Properly handle
14391 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14393 2017-09-20 Richard Biener <rguenther@suse.de>
14395 PR tree-optimization/81373
14396 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14397 Force SESE live-out defs to be handled even if they are
14400 2017-09-19 Jeff Law <law@redhat.com>
14402 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14403 nothing for stack adjustments with REG_STACK_CHECK.
14404 * sched-deps.c (parse_add_or_inc): Reject insns with
14405 REG_STACK_CHECK from dependency breaking.
14406 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14407 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14408 * reg-notes.def (STACK_CHECK): New note.
14410 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14411 (ix86_expand_prologue): Dump stack clash info as needed.
14412 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14414 * function.c (dump_stack_clash_frame_info): New function.
14415 * function.h (dump_stack_clash_frame_info): Prototype.
14416 (enum stack_clash_probes): New enum.
14418 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14419 flag_stack_clash_protection.
14420 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14421 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14422 (arm_frame_pointer_required): Likewise.
14423 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14424 (ia64_expand_prologue): Likewise.
14425 * config/mips/mips.c (mips_expand_prologue): Likewise.
14426 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14427 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14428 (sparc_flat_expand_prologue): Likewise.
14429 * config/spu/spu.c (spu_expand_prologue): Likewise.
14431 * explow.c: Include "params.h".
14432 (anti_adjust_stack_and_probe_stack_clash): New function.
14433 (get_stack_check_protect): Likewise.
14434 (compute_stack_clash_protection_loop_data): Likewise.
14435 (emit_stack_clash_protection_loop_start): Likewise.
14436 (emit_stack_clash_protection_loop_end): Likewise.
14437 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14438 Use anti_adjust_stack_and_probe_stack_clash.
14439 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14440 (emit_stack_clash_protection_loop_start): Likewise.
14441 (emit_stack_clash_protection_loop_end): Likewise.
14442 * rtl.h (get_stack_check_protect): Prototype.
14443 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14444 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14445 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14447 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14449 * doc/tm.texi: Rebuilt.
14450 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14451 get_stack_check_protect.
14452 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14453 * config/arm/arm.c (arm_expand_prologue): Likewise.
14454 (arm_frame_pointer_required): Likewise.
14455 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14456 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14457 * config/mips/mips.c (mips_expand_prologue): Likewise.
14458 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14459 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14460 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14461 (sparc_flat_expand_prologue): Likewise.
14463 * common.opt (-fstack-clash-protection): New option.
14464 * flag-types.h (enum stack_check_type): Note difference between
14465 -fstack-check= and -fstack-clash-protection.
14466 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14467 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14468 * toplev.c (process_options): Issue warnings/errors for cases
14469 not handled with -fstack-clash-protection.
14470 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14471 (-fstack-check): Note additional problem with -fstack-check=generic.
14472 Note that -fstack-check is primarily for Ada and refer users
14473 to -fstack-clash-protection for stack-clash-protection.
14474 Document new params for stack clash protection.
14476 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14478 * config/i386/i386.c (ix86_split_long_move): Do not handle
14479 address used for LEA in a special way.
14481 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14483 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14485 2017-09-19 Martin Sebor <msebor@redhat.com>
14488 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14489 of incompatible types.
14491 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14493 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14494 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14495 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14496 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14498 2017-09-19 Richard Biener <rguenther@suse.de>
14500 PR tree-optimization/82244
14501 * tree-vrp.c (remove_range_assertions): Do not propagate
14502 a constant to abnormals but replace the assert with a copy.
14504 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14506 PR rtl-optimization/57878
14507 PR rtl-optimization/68988
14508 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14509 avoidance test involving non_reload_pseudos. Move frequency test
14510 below the general fragmentation avoidance test.
14512 2017-09-19 Richard Biener <rguenther@suse.de>
14514 PR tree-optimization/69728
14515 * graphite-sese-to-poly.c (schedule_error): New global.
14516 (add_loop_schedule): Handle empty domain by failing the
14518 (build_original_schedule): Handle schedule_error.
14520 2017-09-19 Richard Biener <rguenther@suse.de>
14522 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14523 Do not iterate to sibling loops but only to siblings of inner
14526 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14529 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14530 that the registers are different.
14532 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14534 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14535 to processor_model and "amdfam17h" to arch_names_table.
14536 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14538 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14541 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14543 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14544 Alan Hayward <alan.hayward@arm.com>
14545 David Sherwood <david.sherwood@arm.com>
14547 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14549 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14551 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14552 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14553 vect_schedule_slp_instance.
14554 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14555 Update call to vect_slp_analyze_node_operations. Simplify return
14557 (vect_slp_analyze_bb_1): Update call accordingly.
14558 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14559 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14560 (vect_schedule_slp): Update call accordingly.
14562 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14563 Alan Hayward <alan.hayward@arm.com>
14564 David Sherwood <david.sherwood@arm.com>
14566 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14567 with types that aren't in fact scalar.
14569 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14571 * tree-vect-slp.c (vect_record_max_nunits): New function,
14573 (vect_build_slp_tree_1): ...here.
14574 (vect_build_slp_tree_2): Call it for phis too.
14576 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14578 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14579 to vect_get_vec_def_for_operand when getting the mask operand.
14581 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14582 Alan Hayward <alan.hayward@arm.com>
14583 David Sherwood <david.sherwood@arm.com>
14585 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14588 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14589 Alan Hayward <alan.hayward@arm.com>
14590 David Sherwood <david.sherwood@arm.com>
14592 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14593 calculation for vector booleans.
14595 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14596 Alan Hayward <alan.hayward@arm.com>
14597 David Sherwood <david.sherwood@arm.com>
14599 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14601 (vect_transform_stmt): ...here.
14602 (vect_analyze_stmt): Use it instead of calling
14603 vectorizable_live_operation directly.
14605 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14607 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14608 non-SIMT targets in acc vector loops.
14610 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14612 * configure.ac: Add arc and check if assembler supports gdwarf2.
14613 * configure: Regenerate.
14615 2017-09-18 Richard Biener <rguenther@suse.de>
14617 PR tree-optimization/82220
14618 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14619 epilogue niters from the min_profitable_iters compute.
14621 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14624 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14625 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14626 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14628 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14631 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14632 switching to a new text section.
14634 2017-09-18 Richard Biener <rguenther@suse.de>
14636 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14638 (build_alias_set): Reject aliases with no access function.
14640 2017-09-18 Richard Biener <rguenther@suse.de>
14642 PR tree-optimization/79622
14643 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14645 (build_cross_bb_scalars_use): Likewise.
14647 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14649 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14651 2017-09-18 Alan Modra <amodra@gmail.com>
14654 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14655 stack_pointer_rtx for count 0. Update comments. Break up
14656 large rtl expression.
14658 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14660 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14661 Increase to 20 bytes.
14662 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14663 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14664 or avx version of the stub.
14666 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14669 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14670 compute the minimum stack alignment. Also update preferred stack
14671 boundary for leaf functions.
14673 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14675 PR tree-optimization/82228
14676 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14679 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14681 * common/config/nds32/nds32-common.c
14682 (nds32_option_optimization_table): Refine formatting.
14683 (nds32_option_optimization_table): Use -fsched-pressure and
14684 -fomit-frame-pointer for specific optimization level.
14686 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14688 * config/nds32/nds32.c: Refine formatting and comments.
14689 * config/nds32/nds32.h: Likewise.
14690 * config/nds32/nds32.md: Likewise.
14691 * config/nds32/nds32-cost.c: Likewise.
14692 * config/nds32/nds32-isr.c: Likewise.
14693 * config/nds32/nds32-md-auxiliary.c: Likewise.
14694 * config/nds32/nds32-multiple.md: Likewise.
14695 * config/nds32/nds32-predicates.c: Likewise.
14697 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14698 Jakub Jelinek <jakub@redhat.com>
14700 Add support for -std=c++2a.
14701 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14703 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14705 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14708 * doc/extend.texi (Common Function Attributes): Add
14709 references to ARM, AArch64, and S/390 specific attributes.
14710 (Function Specific Option Pragmas): Add AArch64 and S/390
14711 to list of back ends that support the target pragma.
14713 2017-09-15 Nathan Sidwell <nathan@acm.org>
14715 * doc/standards.texi: Fix C++17 description. Update URLs for
14718 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14720 * common.opt (Wcast-align=strict): New warning option.
14721 * doc/invoke.texi: Document -Wcast-align=strict.
14723 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14725 * cgraph.h (cgraph_thunk_info): Add comments.
14726 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14727 assert for VIRTUAL_* arguments stricter.
14729 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14731 PR tree-optimization/71026
14732 * match.pd: Move RDIV patterns from fold-const.c
14733 * fold-const.c (distribute_real_division): Removed.
14734 (fold_binary_loc): Remove calls to distribute_real_divison.
14736 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14738 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14739 c++1z and gnu++1z as deprecated. Change other references to
14740 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14741 Change -Wc++1z-compat to -Wc++17-compat.
14742 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14743 * dwarf2out.c (highest_c_language): Handle C++17.
14744 (gen_compile_unit_die): Likewise.
14746 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14748 PR rtl-optimization/82192
14749 * combine.c (make_extraction): Don't look through non-paradoxical
14750 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14753 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14754 Alan Hayward <alan.hayward@arm.com>
14755 David Sherwood <david.sherwood@arm.com>
14757 * target.def (function_arg_offset): New hook.
14758 * targhooks.h (default_function_arg_offset): Declare.
14759 * targhooks.c (default_function_arg_offset): New function.
14760 * function.c (locate_and_pad_parm): Use
14761 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14762 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14763 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14764 * doc/tm.texi: Regenerate.
14765 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14766 * config/spu/spu.c (spu_function_arg_offset): New function.
14767 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14768 * system.h (FUNCTION_ARG_OFFSET): Poison.
14770 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14771 Alan Hayard <alan.hayward@arm.com>
14772 David Sherwood <david.sherwood@arm.com>
14774 * target.def (truly_noop_truncation): New hook.
14775 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14776 than TRULY_NOOP_TRUNCATION.
14777 * hooks.h (hook_bool_uint_uint_true): Declare.
14778 * hooks.c (hook_bool_uint_uint_true): New function.
14779 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14780 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14781 * doc/tm.texi: Regenerate.
14782 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14783 rather than TRULY_NOOP_TRUNCATION in comments.
14784 (simplify_comparison): Likewise.
14785 (record_truncated_value): Likewise.
14786 * expmed.c (extract_bit_field_1): Likewise.
14787 (extract_split_bit_field): Likewise.
14788 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14789 instead of TRULY_NOOP_TRUNCATION.
14790 * function.c (assign_parm_setup_block): Likewise.
14791 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14792 * rtlhooks.c: Include target.h.
14793 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14794 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14795 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14796 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14797 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14798 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14799 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14800 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14801 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14802 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14803 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14804 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14805 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14806 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14807 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14808 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14809 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14810 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14811 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14812 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14813 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14814 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14815 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14816 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14817 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14818 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14819 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14820 rather than TRULY_NOOP_TRUNCATION in comments.
14821 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14822 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14823 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14824 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14825 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14826 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14827 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14828 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14829 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14830 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14831 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14832 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14833 rather than TRULY_NOOP_TRUNCATION in comments.
14834 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14835 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14836 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14837 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14838 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14839 TRULY_NOOP_TRUNCATION condition.
14840 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14841 (TRULY_NOOP_TRUNCATION): Delete.
14842 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14843 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14844 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14845 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14846 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14847 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14848 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14849 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14850 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14851 rather than TRULY_NOOP_TRUNCATION in comments.
14852 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14853 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14854 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14855 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14856 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14857 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14859 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14862 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14863 (*cmp_ior): Likewise.
14864 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14865 (*ior_scc_scc_cmp): Likewise.
14866 (*and_scc_scc): Likewise.
14867 (*and_scc_scc_cmp): Likewise.
14869 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14870 Alan Hayard <alan.hayward@arm.com>
14871 David Sherwood <david.sherwood@arm.com>
14873 * target.def (can_change_mode_class): New hook.
14874 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14875 (hard_regno_nregs): Likewise.
14876 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14877 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14878 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14879 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14880 (LOAD_EXTEND_OP): Update accordingly.
14881 * doc/tm.texi: Regenerate.
14882 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14883 CANNOT_CHANGE_MODE_CLASS.
14884 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14885 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14886 * combine.c (simplify_set): Update accordingly.
14887 * emit-rtl.c (validate_subreg): Likewise.
14888 * recog.c (general_operand): Likewise.
14889 * regcprop.c (mode_change_ok): Likewise.
14890 * reload1.c (choose_reload_regs): Likewise.
14891 (inherit_piecemeal_p): Likewise.
14892 * rtlanal.c (simplify_subreg_regno): Likewise.
14893 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14894 instead of CANNOT_CHANGE_MODE_CLASS.
14895 (reload_cse_simplify_operands): Likewise.
14896 * reload.c (push_reload): Use targetm.can_change_mode_class
14897 instead of CANNOT_CHANGE_MODE_CLASS.
14898 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14899 REG_CANNOT_CHANGE_MODE_P.
14900 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14901 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14902 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14903 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14904 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14905 (arm_can_change_mode_class): New function.
14906 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14907 than CANNOT_CHANGE_MODE_CLASS in comments.
14908 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14909 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14910 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14911 (ix86_can_change_mode_class): ...this new function, inverting the
14912 sense of the return value.
14913 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14914 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14915 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14916 (ia64_can_change_mode_class): New function.
14917 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14918 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14919 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14920 (m32c_can_change_mode_class): ...this new function, inverting the
14921 sense of the return value.
14922 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14923 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14924 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14925 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14926 (mips_can_change_mode_class): ...this new function, inverting the
14927 sense of the return value.
14928 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14929 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14930 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14931 (msp430_can_change_mode_class): New function.
14932 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14933 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14934 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14935 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14936 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14937 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14938 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14939 (pa_cannot_change_mode_class): Replace with...
14940 (pa_can_change_mode_class): ...this new function, inverting the
14941 sense of the return value.
14942 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14943 than CANNOT_CHANGE_MODE_CLASS in comments.
14944 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14945 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14946 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14947 (pdp11_cannot_change_mode_class): Replace with...
14948 (pdp11_can_change_mode_class): ...this new function, inverting the
14949 sense of the return value.
14950 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14951 * config/powerpcspe/powerpcspe-protos.h
14952 (rs6000_cannot_change_mode_class_ptr): Delete.
14953 * config/powerpcspe/powerpcspe.c
14954 (rs6000_cannot_change_mode_class_ptr): Delete.
14955 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14956 (rs6000_option_override_internal): Assign to
14957 targetm.can_change_mode_class instead of
14958 rs6000_cannot_change_mode_class_ptr.
14959 (rs6000_cannot_change_mode_class): Replace with...
14960 (rs6000_can_change_mode_class): ...this new function, inverting the
14961 sense of the return value.
14962 (rs6000_debug_cannot_change_mode_class): Replace with...
14963 (rs6000_debug_can_change_mode_class): ...this new function.
14964 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14965 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14966 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14967 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14968 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14970 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14971 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14972 (rs6000_option_override_internal): Assign to
14973 targetm.can_change_mode_class instead of
14974 rs6000_cannot_change_mode_class_ptr.
14975 (rs6000_cannot_change_mode_class): Replace with...
14976 (rs6000_can_change_mode_class): ...this new function, inverting the
14977 sense of the return value.
14978 (rs6000_debug_cannot_change_mode_class): Replace with...
14979 (rs6000_debug_can_change_mode_class): ...this new function.
14980 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14981 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14982 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14983 (s390_can_change_mode_class): ...this new function, inverting the
14984 sense of the return value.
14985 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14986 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14987 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14988 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14989 (sh_cannot_change_mode_class): Replace with...
14990 (sh_can_change_mode_class): ...this new function, inverting the
14991 sense of the return value.
14992 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14993 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14994 (sparc_can_change_mode_class): New function.
14995 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14996 * config/spu/spu.c (spu_can_change_mode_class): New function.
14997 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14998 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14999 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15000 (visium_can_change_mode_class): New function.
15001 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15003 2017-09-15 Richard Biener <rguenther@suse.de>
15005 PR tree-optimization/82217
15006 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15007 but not undefined case.
15009 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15012 * postreload.c (reload_cse_simplify_operands): Skip
15013 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15015 2017-09-15 Richard Biener <rguenther@suse.de>
15017 PR tree-optimization/68823
15018 * graphite-scop-detection.c (build_alias_set): If we have a
15019 possible dependence check whether we can handle them by just
15020 looking at the DRs DR_ACCESS_FNs.
15021 (build_scops): If build_alias_set fails, fail the SCOP.
15023 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15025 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15026 to support float128 built-in functions that require the ISA 3.0
15028 (BU_FLOAT128_3_HW): Likewise.
15029 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15030 built-in functions.
15031 (FMAF128): Likewise.
15033 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15034 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15035 floating point instructions.
15036 (rs6000_invalid_builtin): Likewise.
15037 (rs6000_builtin_mask_names): Likewise.
15038 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15039 (RS6000_BTM_FLOAT128_HW): Likewise.
15040 (RS6000_BTM_COMMON): Likewise.
15041 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15043 * doc/extend.texi (RS/6000 built-in functions): Document the
15044 IEEE 128-bit floating point square root and fused multiply-add
15045 built-in functions.
15047 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15049 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15050 reg (r2) isn't in the set of registers defined in the prologue.
15052 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15053 Alan Hayward <alan.hayward@arm.com>
15054 David Sherwood <david.sherwood@arm.com>
15056 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15057 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15058 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15059 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15060 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15062 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15063 max_vectorization_factor.
15064 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15066 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15067 Alan Hayward <alan.hayward@arm.com>
15068 David Sherwood <david.sherwood@arm.com>
15070 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15071 (vect_worthwhile_without_simd_p): Declare.
15072 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15073 (vectorizable_reduction): Use it.
15074 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15075 (vectorizable_operation): Likewise.
15077 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15078 Alan Hayward <alan.hayward@arm.com>
15079 David Sherwood <david.sherwood@arm.com>
15081 * tree-vectorizer.h (vect_get_num_copies): New function.
15082 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15083 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15084 (vectorizable_induction): Likewise.
15085 (vectorizable_live_operation): Likewise.
15086 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15087 (vectorizable_bswap): Likewise.
15088 (vectorizable_call): Likewise.
15089 (vectorizable_conversion): Likewise.
15090 (vectorizable_assignment): Likewise.
15091 (vectorizable_shift): Likewise.
15092 (vectorizable_operation): Likewise.
15093 (vectorizable_store): Likewise.
15094 (vectorizable_load): Likewise.
15095 (vectorizable_condition): Likewise.
15096 (vectorizable_comparison): Likewise.
15097 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15099 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15100 Alan Hayward <alan.hayward@arm.com>
15101 David Sherwood <david.sherwood@arm.com>
15103 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15104 of vect_init_vector.
15106 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15107 Alan Hayward <alan.hayward@arm.com>
15108 David Sherwood <david.sherwood@arm.com>
15110 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15111 an inline wrapper that provides a location.
15112 (gimple_build_vector): Likewise.
15113 * gimple-fold.c (gimple_build_vector_from_val): New function.
15114 (gimple_build_vector): Likewise.
15115 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15116 functions to build the initial value. Always return a gimple value.
15117 (get_initial_defs_for_reduction): Likewise. Only compute
15119 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15120 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15121 (vectorizable_induction): Use gimple_build_vector rather than
15124 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15125 Alan Hayward <alan.hayward@arm.com>
15126 David Sherwood <david.sherwood@arm.com>
15128 * target.h (vec_perm_indices): New typedef.
15129 (auto_vec_perm_indices): Likewise.
15130 * optabs-query.h: Include target.h
15131 (can_vec_perm_p): Take a vec_perm_indices *.
15132 * optabs-query.c (can_vec_perm_p): Likewise.
15133 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15134 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15135 * tree-vect-generic.c (lower_vec_perm): Likewise.
15136 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15137 (vect_grouped_load_supported): Likewise.
15138 (vect_shift_permute_load_chain): Likewise.
15139 (vect_permute_store_chain): Use auto_vec_perm_indices.
15140 (vect_permute_load_chain): Likewise.
15141 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15142 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15143 Update uses of can_vec_perm_p.
15144 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15145 mode with a number of elements. Take a vec_perm_indices *.
15146 (vect_create_epilog_for_reduction): Update accordingly.
15147 Use auto_vec_perm_indices.
15148 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15149 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15150 (vect_transform_slp_perm_load): Likewise.
15151 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15152 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15153 (vect_gen_perm_mask_checked): Likewise.
15154 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15155 (vect_gen_perm_mask_checked): Likewise.
15156 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15157 (vectorizable_store): Likewise.
15158 (vectorizable_load): Likewise.
15159 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15160 (vectorizable_bswap): Likewise.
15162 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15163 Alan Hayward <alan.hayward@arm.com>
15164 David Sherwood <david.sherwood@arm.com>
15166 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15167 * tree.c (build_vector): Likewise.
15168 (build_vector_from_ctor): Update accordingly.
15169 (build_vector_from_val): Likewise.
15170 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15171 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15172 * tree-vect-generic.c (add_rshift): Likewise.
15173 (expand_vector_divmod): Likewise.
15174 (optimize_vector_constructor): Likewise.
15175 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15176 (vect_transform_slp_perm_load): Likewise.
15177 (vect_schedule_slp_instance): Likewise.
15178 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15179 (vectorizable_call): Likewise.
15180 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15181 * expmed.c (make_tree): Likewise.
15182 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15183 a vector passed to build_vector.
15184 (fold_convert_const): Likewise.
15185 (exact_inverse): Likewise.
15186 (fold_ternary_loc): Likewise.
15187 (fold_relational_const): Likewise.
15188 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15189 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15190 (const_unop): Likewise. Store the reduction accumulator in a
15191 variable rather than an array.
15192 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15193 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15194 the new vector, rather than constructing it after the input arrays.
15195 (native_interpret_vector): Use auto_vec<tree> when building
15196 a vector passed to build_vector. Add elements in order.
15197 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15198 auto_vec<tree> when building a vector passed to build_vector.
15199 (vect_create_epilog_for_reduction): Likewise.
15200 (vectorizable_induction): Likewise.
15201 (get_initial_def_for_reduction): Likewise. Fix indentation of
15203 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15205 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15206 passed to build_vector.
15208 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15209 Alan Hayward <alan.hayward@arm.com>
15210 David Sherwood <david.sherwood@arm.com>
15212 * tree-core.h (tree_base::u): Add an "nelts" field.
15213 (tree_vector): Use VECTOR_CST_NELTS as the length.
15214 * tree.c (tree_size): Likewise.
15215 (make_vector): Initialize VECTOR_CST_NELTS.
15216 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15217 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15218 TYPE_VECTOR_SUBPARTS.
15219 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15220 as the number of units, setting it from VECTOR_CST_NELTS.
15221 (const_vector_from_tree): Likewise.
15222 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15223 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15224 (fold_negate_expr_1): Likewise.
15225 (fold_convert_const): Likewise.
15226 (const_binop): Likewise. Differentiate the number of output and
15228 (const_unop): Likewise.
15229 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15230 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15231 in cases that did the opposite.
15233 2017-09-14 Richard Biener <rguenther@suse.de>
15235 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15238 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15240 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15242 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15245 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15246 if and where to split a bb, except for splitting before debug insn
15247 sequences followed by non-label real insn. Delete debug insns
15248 in between basic blocks.
15250 * combine.c (make_compound_operation_int): Formatting fixes.
15252 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15253 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15254 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15255 * config/sol2.h (LINK_EH_SPEC): Likewise.
15256 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15257 * config/s390/linux.h (LINK_SPEC): Likewise.
15258 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15259 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15260 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15261 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15262 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15263 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15264 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15265 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15267 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15269 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15271 (ENDFILE_LINUX_SPEC): Likewise.
15272 (LINK_EH_SPEC): Likewise.
15273 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15274 (LINK_OS_LINUX_SPEC32): Likewise.
15275 (LINK_OS_LINUX_SPEC64): Likewise.
15276 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15277 (LINK_OS_LINUX_SPEC): Likewise.
15279 2017-09-13 Martin Liska <mliska@suse.cz>
15281 PR middle-end/82154
15282 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15283 CASE_HIGH is NULL_TREE.
15285 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15286 Alan Hayward <alan.hayward@arm.com>
15287 David Sherwood <david.sherwood@arm.com>
15289 * target.def (secondary_memory_needed): New hook.
15290 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15291 instead of SECONDARY_MEMORY_NEEDED.
15292 (secondary_memory_needed_mode): Likewise.
15293 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15294 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15295 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15296 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15297 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15298 * doc/tm.texi: Regenerate.
15299 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15300 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15301 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15302 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15303 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15304 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15305 mode argument first and change the reg_class arguments to reg_class_t.
15306 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15307 Make static. Update the call to inline_secondary_memory_needed.
15308 (ix86_register_move_cost): Update the call to
15309 inline_secondary_memory_needed.
15310 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15311 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15313 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15314 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15316 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15317 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15318 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15319 and match hook interface. Add comment from mips.h.
15320 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15321 * config/mmix/mmix.md (truncdfsf2): Refer to
15322 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15324 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15325 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15326 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15327 (pa_secondary_memory_needed): New function.
15328 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15329 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15330 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15331 (pdp11_secondary_memory_needed): Make static and match hook interface.
15332 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15333 * config/powerpcspe/powerpcspe-protos.h
15334 (rs6000_secondary_memory_needed_ptr): Delete.
15335 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15337 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15338 (rs6000_option_override_internal): Assign to
15339 targetm.secondary_memory_needed rather than
15340 rs6000_secondary_memory_needed_ptr.
15341 (rs6000_secondary_memory_needed): Match hook interface.
15342 (rs6000_debug_secondary_memory_needed): Likewise.
15343 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15344 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15345 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15346 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15347 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15348 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15350 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15351 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15352 (rs6000_option_override_internal): Assign to
15353 targetm.secondary_memory_needed rather than
15354 rs6000_secondary_memory_needed_ptr.
15355 (rs6000_secondary_memory_needed): Match hook interface.
15356 (rs6000_debug_secondary_memory_needed): Likewise.
15357 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15358 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15359 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15360 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15361 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15362 (sparc_secondary_memory_needed): New function.
15363 * lra-constraints.c (check_and_process_move): Refer to
15364 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15366 (curr_insn_transform): Likewise.
15367 (process_alt_operands): Use targetm.secondary_memory_needed
15368 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15369 (check_secondary_memory_needed_p): Likewise.
15370 (choose_split_class): Likewise.
15371 * reload.c: Unconditionally include code that was previously
15372 conditional on SECONDARY_MEMORY_NEEDED.
15373 (push_secondary_reload): Use targetm.secondary_memory_needed
15374 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15375 (push_reload): Likewise.
15376 * reload1.c: Unconditionally include code that was previously
15377 conditional on SECONDARY_MEMORY_NEEDED.
15378 (choose_reload_regs): Use targetm.secondary_memory_needed
15379 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15380 (gen_reload): Likewise.
15381 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15383 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15384 Alan Hayward <alan.hayward@arm.com>
15385 David Sherwood <david.sherwood@arm.com>
15387 * target.def (secondary_memory_needed_mode): New hook:
15388 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15389 * targhooks.h (default_secondary_memory_needed_mode): New function.
15390 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15391 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15392 * doc/tm.texi: Regenerate.
15393 * lra-constraints.c (check_and_process_move): Use
15394 targetm.secondary_memory_needed_mode instead of
15395 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15396 (curr_insn_transform): Likewise.
15397 * reload.c (get_secondary_mem): Likewise.
15398 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15399 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15401 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15402 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15403 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15404 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15405 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15407 * config/powerpcspe/powerpcspe-protos.h
15408 (rs6000_secondary_memory_needed_mode): Delete.
15409 * config/powerpcspe/powerpcspe.c
15410 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15411 (rs6000_secondary_memory_needed_mode): Make static.
15412 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15413 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15415 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15417 (rs6000_secondary_memory_needed_mode): Make static.
15418 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15419 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15420 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15421 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15422 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15424 (sparc_secondary_memory_needed_mode): New function.
15425 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15427 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15429 * config/aarch64/constraints.md (Umq): New constraint.
15430 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15432 (mov<mode>): Update condition.
15434 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15436 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15437 when bitposition is the same.
15439 2017-09-13 Richard Biener <rguenther@suse.de>
15441 * dwarf2out.c (output_die_symbol): Remove.
15442 (output_die): Do not output a DIEs symbol.
15444 2017-09-13 Richard Biener <rguenther@suse.de>
15446 PR middle-end/82128
15447 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15448 default-def to avoid breaking iterator update with the weird
15449 interaction with cgraph_update_edges_for_call_stmt_node.
15451 2017-09-13 Richard Biener <rguenther@suse.de>
15453 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15454 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15455 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15456 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15458 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15460 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15461 Disable pc relative literal load irrespective of
15462 TARGET_FIX_ERR_A53_84341 for default.
15464 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15466 * config/sparc/sparc.c (output_return): Output the source location of
15467 the insn in the delay slot, if any.
15468 (output_sibcall): Likewise.
15470 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15473 * common.opt (-static-pie): New alias.
15474 (shared): Negate static-pie.
15475 (-no-pie): Update help text.
15477 (static-pie): New option.
15478 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15479 -static-pie support.
15480 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15481 (LINK_EH_SPEC): Likewise.
15482 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15483 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15484 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15485 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15486 (init_gcc_specs): Likewise.
15487 (init_spec): Likewise.
15488 (display_help): Update help message for -pie.
15489 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15492 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15494 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15495 (movdi_aarch64): Likewise.
15496 (movti_aarch64): Likewise.
15498 2017-09-12 Simon Wright <simon@pushface.org>
15501 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15502 calculation of the minor version, always output as 0.
15504 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15507 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15508 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15509 on it early, rather than manual conversion late. For
15510 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15511 instead of performing manual conversion.
15513 2017-09-12 Carl Love <cel@us.ibm.com>
15515 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15516 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15518 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15519 VMULOSW): Add definitions.
15520 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15521 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15522 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15523 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15524 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15526 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15528 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15530 (movti_aarch64): Likewise.
15531 (movdf_aarch64): Likewise.
15532 (movtf_aarch64): Likewise.
15533 (load_pairdi): Likewise.
15534 (store_pairdi): Likewise.
15535 (load_pairdf): Likewise.
15536 (store_pairdf): Likewise.
15537 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15538 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15539 (ldr_got_small_<mode>): Likewise.
15540 (ldr_got_small_28k_<mode>): Likewise.
15541 (ldr_got_tiny): Likewise.
15542 * config/aarch64/iterators.md (ldst_sz): New.
15543 (ldpstp_sz): Likewise.
15544 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15546 (thunderx_load): Split load_8 to load_16.
15547 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15549 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15550 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15551 (xgene1_store_pair): Split store_8 to store_16.
15552 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15553 (falkor_st_0_st_sd): Split store_8 to store_16.
15555 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15557 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15558 and store1/2/3/4 to store_4/8/12/16.
15559 * config/aarch64/aarch64.md: Update for rename.
15560 * config/arm/arm.md: Likewise.: Likewise.
15561 * config/arm/arm.c: Likewise.
15562 * config/arm/thumb1.md: Likewise.
15563 * config/arm/thumb2.md: Likewise.
15564 * config/arm/vfp.md: Likewise.
15565 * config/arm/arm-generic.md: Likewise.
15566 * config/arm/arm1020e.md: Likewise.
15567 * config/arm/arm1026ejs.md: Likewise.
15568 * config/arm/arm1136jfs.md: Likewise.
15569 * config/arm/arm926ejs.md: Likewise.
15570 * config/arm/cortex-a15.md: Likewise.
15571 * config/arm/cortex-a17.md: Likewise.
15572 * config/arm/cortex-a5.md: Likewise.
15573 * config/arm/cortex-a53.md: Likewise.
15574 * config/arm/cortex-a57.md: Likewise.
15575 * config/arm/cortex-a7.md: Likewise.
15576 * config/arm/cortex-a8.md: Likewise.
15577 * config/arm/cortex-a9.md: Likewise.
15578 * config/arm/cortex-m4.md: Likewise.
15579 * config/arm/cortex-m7.md: Likewise.
15580 * config/arm/cortex-r4.md: Likewise.
15581 * config/arm/exynos-m1.md: Likewise.
15582 * config/arm/fa526.md: Likewise.
15583 * config/arm/fa606te.md: Likewise.
15584 * config/arm/fa626te.md: Likewise.
15585 * config/arm/fa726te.md: Likewise.
15586 * config/arm/fmp626.md: Likewise.
15587 * config/arm/iwmmxt.md: Likewise.
15588 * config/arm/ldmstm.md: Likewise.
15589 * config/arm/marvell-pj4.md: Likewise.
15590 * config/arm/xgene1.md: Likewise.
15591 * config/aarch64/thunderx.md: Likewise.
15592 * config/aarch64/thunderx2t99.md: Likewise.
15593 * config/aarch64/falkor.md: Likewise.
15595 2017-09-12 Martin Liska <mliska@suse.cz>
15597 * attribs.c (private_lookup_attribute): New function.
15598 * attribs.h (private_lookup_attribute): Declared here.
15599 (lookup_attribute): Called from this place.
15601 2017-09-12 Richard Biener <rguenther@suse.de>
15603 PR tree-optimization/82157
15604 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15605 stmts with side-effects.
15607 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15608 Alan Hayward <alan.hayward@arm.com>
15609 David Sherwood <david.sherwood@arm.com>
15611 * target.def (hard_regno_nregs): New hook.
15612 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15613 * targhooks.h (default_hard_regno_nregs): Declare.
15614 * targhooks.c (default_hard_regno_nregs): New function.
15615 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15616 (TARGET_HARD_REGNO_NREGS): ...this hook.
15617 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15618 (CLASS_MAX_NREGS): Likewise.
15619 * doc/tm.texi: Regenerate.
15620 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15621 instead of HARD_REGNO_NREGS.
15622 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15623 HARD_REGNO_NREGS in the comment.
15624 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15625 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15626 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15627 Return an unsigned int.
15628 (TARGET_HARD_REGNO_NREGS): Redefine.
15629 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15630 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15631 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15632 (arc_hard_regno_nregs): New function.
15633 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15634 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15635 (arm_hard_regno_nregs): New function.
15636 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15637 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15638 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15639 (TARGET_HARD_REGNO_NREGS): Redefine.
15640 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15641 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15642 (HARD_REGNO_NREGS): Delete.
15643 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15644 (cr16_hard_regno_nregs): New function.
15645 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15646 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15647 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15648 (cris_hard_regno_nregs): New function.
15649 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15650 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15651 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15652 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15653 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15654 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15655 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15656 (frv_hard_regno_nregs): Make static. Take and return an
15658 (frv_class_max_nregs): Remove outdated copy of documentation.
15659 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15660 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15661 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15662 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15663 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15664 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15665 (TARGET_HARD_REGNO_NREGS): Redefine.
15666 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15667 (CLASS_MAX_NREGS): Update comment.
15668 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15669 (ia64_hard_regno_nregs): New function.
15670 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15671 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15672 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15673 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15674 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15676 (m32c_hard_regno_nregs): Likewise. Make static.
15677 (TARGET_HARD_REGNO_NREGS): Redefine.
15678 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15679 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15680 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15681 (m68k_hard_regno_nregs): New function.
15682 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15683 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15684 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15685 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15686 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15687 Take and return an unsigned int.
15688 (TARGET_HARD_REGNO_NREGS): Redefine.
15689 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15690 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15691 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15692 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15693 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15694 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15695 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15696 (msp430_hard_regno_nregs): Make static. Take and return an
15698 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15699 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15700 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15701 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15702 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15703 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15704 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15705 (TARGET_HARD_REGNO_NREGS): Redefine.
15706 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15707 (PA_HARD_REGNO_NREGS): ...this.
15708 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15709 (PA_HARD_REGNO_NREGS): ...this.
15710 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15711 (pa_hard_regno_nregs): New function.
15712 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15713 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15714 (pdp11_hard_regno_nregs): New function.
15715 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15716 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15717 (rs6000_hard_regno_nregs_hook): New function.
15718 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15719 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15720 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15721 Take and return an unsigned int. Move earlier in file.
15722 (TARGET_HARD_REGNO_NREGS): Redefine.
15723 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15724 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15725 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15726 (rl78_hard_regno_nregs): Make static. Take and return an
15728 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15729 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15730 (rs6000_hard_regno_nregs_hook): New function.
15731 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15732 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15733 (TARGET_HARD_REGNO_NREGS): Redefine.
15734 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15735 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15736 instead of HARD_REGNO_NREGS.
15737 (s390_hard_regno_nregs): New function.
15738 (s390_hard_regno_mode_ok): Add comment from s390.h.
15739 (TARGET_HARD_REGNO_NREGS): Redefine.
15740 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15741 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15742 (sh_hard_regno_nregs): New function.
15743 (sh_pass_in_reg_p): Use it.
15744 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15745 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15746 (sparc_hard_regno_nregs): New function.
15747 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15748 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15749 (spu_function_arg_advance): Use it, supplying a valid register number.
15750 (TARGET_HARD_REGNO_NREGS): Redefine.
15751 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15752 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15753 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15754 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15755 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15756 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15757 (CLASS_MAX_NREGS): Remove copy of old documentation.
15758 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15759 (visium_hard_regno_nregs): New function.
15760 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15761 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15762 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15763 xtensa_hard_regno_nregs): New function.
15764 * system.h (HARD_REGNO_NREGS): Poison.
15766 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15768 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15769 hard_regno_nregs instead of HARD_REGNO_NREGS.
15770 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15771 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15772 (c6x_expand_epilogue): Likewise.
15773 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15774 (frv_read_iacc_argument): Likewise.
15775 * config/sh/sh.c: Include regs.h.
15776 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15777 (regs_used): Likewise.
15778 (output_stack_adjust): Likewise.
15779 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15780 * expmed.c: Include regs.h.
15781 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15782 * ree.c: Include regs.h.
15783 (combine_reaching_defs): Use hard_regno_nregs instead of
15785 (add_removable_extension): Likewise.
15787 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15789 * regs.h (hard_regno_nregs): Turn into a function.
15790 (end_hard_regno): Update accordingly.
15791 * caller-save.c (setup_save_areas): Likewise.
15792 (save_call_clobbered_regs): Likewise.
15793 (replace_reg_with_saved_mem): Likewise.
15794 (insert_restore): Likewise.
15795 (insert_save): Likewise.
15796 * combine.c (can_change_dest_mode): Likewise.
15797 (move_deaths): Likewise.
15798 (distribute_notes): Likewise.
15799 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15800 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15801 (rs6000_split_multireg_move): Likewise.
15802 (rs6000_register_move_cost): Likewise.
15803 (rs6000_memory_move_cost): Likewise.
15804 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15805 (rs6000_split_multireg_move): Likewise.
15806 (rs6000_register_move_cost): Likewise.
15807 (rs6000_memory_move_cost): Likewise.
15808 * cselib.c (cselib_reset_table): Likewise.
15809 (cselib_lookup_1): Likewise.
15810 * emit-rtl.c (set_mode_and_regno): Likewise.
15811 * function.c (aggregate_value_p): Likewise.
15812 * ira-color.c (setup_profitable_hard_regs): Likewise.
15813 (check_hard_reg_p): Likewise.
15814 (calculate_saved_nregs): Likewise.
15815 (assign_hard_reg): Likewise.
15816 (improve_allocation): Likewise.
15817 (calculate_spill_cost): Likewise.
15818 * ira-emit.c (modify_move_list): Likewise.
15819 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15820 (ira_hard_reg_in_set_p): Likewise.
15821 * ira.c (setup_reg_mode_hard_regset): Likewise.
15822 (clarify_prohibited_class_mode_regs): Likewise.
15823 (check_allocation): Likewise.
15824 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15825 (lra_setup_reg_renumber): Likewise.
15826 (setup_try_hard_regno_pseudos): Likewise.
15827 (spill_for): Likewise.
15828 (assign_hard_regno): Likewise.
15829 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15830 * lra-constraints.c (in_class_p): Likewise.
15831 (lra_constraint_offset): Likewise.
15832 (simplify_operand_subreg): Likewise.
15833 (lra_constraints): Likewise.
15834 (split_reg): Likewise.
15835 (split_if_necessary): Likewise.
15836 (invariant_p): Likewise.
15837 (inherit_in_ebb): Likewise.
15838 * lra-lives.c (process_bb_lives): Likewise.
15839 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15840 (get_hard_regs): Likewise.
15841 (do_remat): Likewise.
15842 * lra-spills.c (assign_spill_hard_regs): Likewise.
15843 * mode-switching.c (create_pre_exit): Likewise.
15844 * postreload.c (reload_combine_recognize_pattern): Likewise.
15845 * recog.c (peep2_find_free_register): Likewise.
15846 * regcprop.c (kill_value_regno): Likewise.
15847 (set_value_regno): Likewise.
15848 (copy_value): Likewise.
15849 (maybe_mode_change): Likewise.
15850 (find_oldest_value_reg): Likewise.
15851 (copyprop_hardreg_forward_1): Likewise.
15852 * regrename.c (check_new_reg_p): Likewise.
15853 (regrename_do_replace): Likewise.
15854 * reload.c (push_reload): Likewise.
15855 (combine_reloads): Likewise.
15856 (find_dummy_reload): Likewise.
15857 (operands_match_p): Likewise.
15858 (find_reloads): Likewise.
15859 (find_equiv_reg): Likewise.
15860 (reload_adjust_reg_for_mode): Likewise.
15861 * reload1.c (count_pseudo): Likewise.
15862 (count_spilled_pseudo): Likewise.
15863 (find_reg): Likewise.
15864 (clear_reload_reg_in_use): Likewise.
15865 (free_for_value_p): Likewise.
15866 (allocate_reload_reg): Likewise.
15867 (choose_reload_regs): Likewise.
15868 (reload_adjust_reg_for_temp): Likewise.
15869 (emit_reload_insns): Likewise.
15870 (delete_output_reload): Likewise.
15871 * rtlanal.c (subreg_get_info): Likewise.
15872 * sched-deps.c (sched_analyze_reg): Likewise.
15873 * sel-sched.c (init_regs_for_mode): Likewise.
15874 (mark_unavailable_hard_regs): Likewise.
15875 (choose_best_reg_1): Likewise.
15876 (verify_target_availability): Likewise.
15877 * valtrack.c (dead_debug_insert_temp): Likewise.
15878 * var-tracking.c (track_loc_p): Likewise.
15879 (emit_note_insn_var_location): Likewise.
15880 * varasm.c (make_decl_rtl): Likewise.
15881 * reginfo.c (choose_hard_reg_mode): Likewise.
15882 (init_reg_modes_target): Refer directly to
15883 this_target_regs->x_hard_regno_nregs.
15885 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15887 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15888 instead of hard_regno_nregs.
15890 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15892 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15893 end_hard_regno instead of hard_regno_nregs.
15894 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15895 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15896 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15897 * ira-color.c (improve_allocation): Likewise.
15898 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15899 * lra-lives.c (mark_regno_live): Likewise.
15900 (mark_regno_dead): Likewise.
15901 * lra-remat.c (operand_to_remat): Likewise.
15902 * lra.c (collect_non_operand_hard_regs): Likewise.
15903 * postreload.c (reload_combine_note_store): Likewise.
15904 (move2add_valid_value_p): Likewise.
15905 * reload.c (regno_clobbered_p): Likewise.
15907 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15909 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15911 * config/v850/v850.c (v850_reorg): Likewise.
15912 * reload.c (refers_to_regno_for_reload_p): Likewise.
15913 (find_equiv_reg): Likewise.
15914 * reload1.c (reload_reg_reaches_end_p): Likewise.
15916 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15918 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15920 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15921 * config/arm/arm.c (output_move_neon): Likewise.
15922 (arm_attr_length_move_neon): Likewise.
15923 (neon_split_vcombine): Likewise.
15924 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15925 (c6x_mark_reg_written): Likewise.
15926 (c6x_dwarf_register_span): Likewise.
15927 * config/i386/i386.c (ix86_save_reg): Likewise.
15928 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15929 (rws_access_reg): Likewise.
15930 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15931 * mode-switching.c (create_pre_exit): Likewise.
15932 * ree.c (combine_reaching_defs): Likewise.
15933 (add_removable_extension): Likewise.
15934 * regcprop.c (find_oldest_value_reg): Likewise.
15935 (copyprop_hardreg_forward_1): Likewise.
15936 * reload.c (reload_inner_reg_of_subreg): Likewise.
15937 (push_reload): Likewise.
15938 (combine_reloads): Likewise.
15939 (find_dummy_reload): Likewise.
15940 (reload_adjust_reg_for_mode): Likewise.
15941 * reload1.c (find_reload_regs): Likewise.
15942 (forget_old_reloads_1): Likewise.
15943 (reload_reg_free_for_value_p): Likewise.
15944 (reload_adjust_reg_for_temp): Likewise.
15945 (emit_reload_insns): Likewise.
15946 (delete_output_reload): Likewise.
15947 * sel-sched.c (choose_best_reg_1): Likewise.
15948 (choose_best_pseudo_reg): Likewise.
15950 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15951 Alan Hayward <alan.hayward@arm.com>
15952 David Sherwood <david.sherwood@arm.com>
15954 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15955 * target.def (slow_unaligned_access): New hook.
15956 * targhooks.h (default_slow_unaligned_access): Declare.
15957 * targhooks.c (default_slow_unaligned_access): New function.
15958 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15959 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15960 * doc/tm.texi: Regenerate.
15961 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15962 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15963 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15965 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15966 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15968 (rs6000_slow_unaligned_access): New function.
15969 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15970 (expand_block_compare): Likewise.
15971 (expand_strn_compare): Likewise.
15972 (rs6000_rtx_costs): Likewise.
15973 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15974 (riscv_slow_unaligned_access): Likewise.
15975 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15976 (riscv_slow_unaligned_access_p): ...this and make static.
15977 (riscv_option_override): Update accordingly.
15978 (riscv_slow_unaligned_access): New function.
15979 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15980 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15981 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15982 (rs6000_slow_unaligned_access): New function.
15983 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15984 (rs6000_rtx_costs): Likewise.
15985 * config/rs6000/rs6000-string.c (expand_block_compare)
15986 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15987 of SLOW_UNALIGNED_ACCESS.
15988 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15989 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15990 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15991 of SLOW_UNALIGNED_ACCESS.
15992 * expmed.c (simple_mem_bitfield_p): Likewise.
15993 * expr.c (alignment_for_piecewise_move): Likewise.
15994 (emit_group_load_1): Likewise.
15995 (emit_group_store): Likewise.
15996 (copy_blkmode_from_reg): Likewise.
15997 (emit_push_insn): Likewise.
15998 (expand_assignment): Likewise.
15999 (store_field): Likewise.
16000 (expand_expr_real_1): Likewise.
16001 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16002 * lra-constraints.c (simplify_operand_subreg): Likewise.
16003 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16004 * gimple-ssa-store-merging.c: Likewise in block comment at start
16006 * tree-ssa-strlen.c: Include target.h.
16007 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16008 of SLOW_UNALIGNED_ACCESS.
16009 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16011 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16013 PR rtl-optimization/82185
16014 * expmed.c (emit_store_flag_int): Only test tem if it has been
16017 2017-09-12 Richard Biener <rguenther@suse.de>
16019 PR middle-end/82149
16020 * match.pd ((FTYPE) N CMP CST): Fix typo.
16022 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16024 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16026 (mips_near_type_p): Add 'short_call' attribute as a synonym
16028 * doc/extend.texi (short_call): Document new function attribute.
16030 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16033 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16034 assertion check that in the condition.
16035 (get_atomic_generic_size): Likewise. Before testing if parameter
16036 has pointer type, if it has array type, call for C++
16037 default_conversion to perform array-to-pointer conversion.
16039 2017-09-12 Richard Biener <rguenther@suse.de>
16041 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16042 for operations we cannot scalarize.
16044 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16046 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16047 vectors heap vectors. Clean up comments.
16048 Make visited_bbs a reference.
16049 (profitable_jump_thread_path): Make GC
16050 vectors heap vectors. Clean up comments.
16052 (convert_and_register_jump_thread_path): Make GC vectors heap
16054 (check_subpath_and_update_thread_path): Same. Clean up comments.
16055 Make visited_bbs a reference.
16056 (handle_phi): Abstract common code to to
16057 register_jump_thread_path_if_profitable.
16058 Rename VAR_BB to DEF_BB.
16060 Make GC vectors heap vectors.
16061 Make visited_bbs a reference.
16062 (handle_assignment): Same.
16063 (register_jump_thread_path_if_profitable): New.
16064 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16066 Make GC vectors heap vectors. Clean up comments.
16067 Make visited_bbs a reference.
16068 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16069 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16072 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16075 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16076 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16078 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16080 Revert r251800 and r251799.
16082 2017-09-11 Martin Jambor <mjambor@suse.cz>
16085 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16086 arguments in advance.
16087 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16088 use it to find predecessor edges.
16089 (naive_outof_ssa): Collect vector of predecessors.
16091 2017-09-08 Jason Merrill <jason@redhat.com>
16093 PR c++/70029 - ICE with ref-qualifier and -flto
16094 * langhooks.h (struct lang_hooks_for_types): Add
16095 copy_lang_qualifiers.
16096 * attribs.c (build_type_attribute_qual_variant): Use it.
16097 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16099 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16100 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16102 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16105 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16106 (*mulsi3_sp64): New instruction.
16107 (mulsi3): New expander.
16109 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16111 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16113 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16115 * sancov.c: Include memmodel.h.
16117 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16120 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16123 2017-09-07 Carl Love <cel@us.ibm.com>
16125 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16126 the sldi instruction.
16128 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16130 * sancov.c: Include tm_p.h.
16132 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16135 * output.h (switch_to_other_text_partition): New declaration.
16136 * varasm.c (switch_to_other_text_partition): New function.
16137 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16138 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16139 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16140 to the other text partition before emitting LCL label and switch back
16141 after emitting the word after it.
16143 2017-09-07 Richard Biener <rguenther@suse.de>
16145 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16146 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16147 critical edges here, after loop init.
16148 (pass_data_pre): Remove PROP_no_crit_edges flags.
16149 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16150 for valueization of call args to avoid leaking VN_TOP.
16151 (visit_use): Assert we do not visit default defs.
16152 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16153 Use error_mark_node to more easily detect leaking VN_TOP.
16154 All default-defs are varying, not VN_TOP. Mark them visited.
16155 (run_scc_vn): Make code match comment.
16157 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16159 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16160 OPTION_MASK_FLOAT128_KEYWORD.
16161 (POWERPC_MASKS): Likewise.
16162 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16163 support for the -mfloat128-type option, and make -mfloat128
16164 default on PowerPC Linux systems. Define or undefine
16165 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16166 Define __float128 to be __ieee128 if IEEE 128-bit support is
16167 enabled, or undefine it.
16168 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16169 Delete defining __FLOAT128_TYPE__.
16170 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16171 -mfloat128-type option and make -mfloat128 default on PowerPC
16173 (TARGET_FLOAT128_TYPE): Likewise.
16174 (-mfloat128-type): Likewise.
16175 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16176 Delete the -mfloat128-type option and make -mfloat128 default on
16177 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16178 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16179 128-bit floating point is enabled. Change tests from using
16180 -mfloat128-type to -mfloat128.
16181 (rs6000_mangle_type): Use the correct mangling for the __float128
16182 type even if normal long double is restricted to 64-bits.
16183 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16185 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16186 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16187 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16189 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16190 documentation for -mfloat128.
16192 2017-09-06 Olivier Hainque <hainque@adacore.com>
16194 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16196 2017-09-06 Wish Wu <wishwu007@gmail.com>
16197 Jakub Jelinek <jakub@redhat.com>
16199 * asan.c (initialize_sanitizer_builtins): Add
16200 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16201 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16202 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16203 BT_FN_VOID_UINT64_PTR variables.
16204 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16205 (BT_FN_VOID_UINT16_UINT16): Likewise.
16206 (BT_FN_VOID_UINT32_UINT32): Likewise.
16207 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16208 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16209 (BT_FN_VOID_UINT64_PTR): Likewise.
16210 * common.opt (flag_sanitize_coverage): New variable.
16211 (fsanitize-coverage=trace-pc): Remove.
16212 (fsanitize-coverage=): Add.
16213 * flag-types.h (enum sanitize_coverage_code): New enum.
16214 * fold-const.c (fold_range_test): Disable non-short-circuit
16215 optimization if flag_sanitize_coverage.
16216 (fold_truth_andor): Likewise.
16217 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16218 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16219 (coverage_sanitizer_opts): New array.
16220 (get_closest_sanitizer_option): Add OPTS argument, handle also
16221 OPT_fsanitize_coverage_.
16222 (parse_sanitizer_options): Adjusted to also handle
16223 OPT_fsanitize_coverage_.
16224 (common_handle_option): Add OPT_fsanitize_coverage_.
16225 * sancov.c (instrument_comparison, instrument_switch): New function.
16226 (sancov_pass): Add trace-cmp support.
16227 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16228 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16229 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16230 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16231 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16232 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16233 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16234 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16235 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16236 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16238 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16240 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16241 error. Only quit immediately if parsing is complete.
16242 (BEGIN): Initialize fatal_err and parse_done.
16243 (begin fpu, end fpu): Check number of arguments.
16244 (begin arch, end arch): Likewise.
16245 (begin cpu, end cpu): Likewise.
16246 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16247 (optalias): Likewise.
16249 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16251 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16252 * config/arm/arm-isa.h: Delete. Move definitions to ...
16253 * arm-cpus.in: ... here. Use new feature and fgroup values.
16254 * config/arm/arm.c (arm_option_override): Use lower case for feature
16256 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16257 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16258 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16260 (print_isa_bits_for): New function.
16261 (gen_isa): New function.
16262 (gen_comm_data): Use print_isa_bits_for.
16263 (define feature): New keyword.
16264 (define fgroup): New keyword.
16265 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16266 (arm-isa.h): Add rule to generate file.
16267 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16268 case for feature bit names.
16270 2017-09-06 Richard Biener <rguenther@suse.de>
16272 * tree-ssa-pre.c (NECESSARY): Remove.
16273 (create_expression_by_pieces): Do not touch pass-local flags.
16274 (insert_into_preds_of_block): Likewise.
16275 (do_pre_regular_insertion): Likewise.
16276 (eliminate_insert): Likewise.
16277 (eliminate_dom_walker::before_dom_children): Likewise.
16278 (fini_eliminate): Do not look at inserted_exprs.
16279 (mark_operand_necessary): Remove.
16280 (remove_dead_inserted_code): Replace with simple work-list
16281 algorithm based on inserted_exprs and SSA uses.
16282 (pass_pre::execute): Re-order fini_eliminate and
16283 remove_dead_inserted_code.
16285 2017-09-06 Olivier Hainque <hainque@adacore.com>
16287 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16288 for VxWorks 7. Adjust surrounding comments.
16290 2017-09-06 Richard Biener <rguenther@suse.de>
16292 * gimple-ssa-strength-reduction.c
16293 (find_candidates_dom_walker::before_dom_children): Also allow
16296 2017-09-06 Richard Biener <rguenther@suse.de>
16298 PR tree-optimization/82108
16299 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16300 for gap in the non-permutation SLP case.
16302 2017-09-06 Martin Jambor <mjambor@suse.cz>
16304 PR tree-optimization/82078
16305 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16306 add_access_to_work_queue...
16307 (build_accesses_from_assign): ...here.
16308 (propagate_all_subaccesses): Make sure racc is the group
16309 representative, if there is one.
16311 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16313 PR middle-end/82095
16314 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16317 2017-09-06 Richard Biener <rguenther@suse.de>
16319 * gimple-ssa-strength-reduction.c
16320 (find_candidates_dom_walker::before_doom_children): Use a
16321 type and not a mode check.
16323 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16326 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16328 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16330 2017-09-05 Jeff Law <law@redhat.com>
16332 PR tree-optimization/64910
16333 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16334 cases where we have 3 or more operands.
16336 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16338 PR middle-end/81768
16339 * omp-low.c (lower_omp_for): Recompute tree invariant if
16340 gimple_omp_for_initial/final is ADDR_EXPR.
16342 PR middle-end/81768
16343 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16344 into gimple val before gimplification fo the COND_EXPR.
16346 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16348 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16349 REGION_COPY argument.
16350 (thread_through_all_blocks): Remove unused argument to
16351 duplicate_thread_path.
16353 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16354 Alan Hayward <alan.hayward@arm.com>
16355 David Sherwood <david.sherwood@arm.com>
16357 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16358 Take a scalar_mode rather than a machine_mode.
16359 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16360 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16361 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16362 (aarch64_gen_adjusted_ldpstp): Likewise.
16363 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16365 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16366 Alan Hayward <alan.hayward@arm.com>
16367 David Sherwood <david.sherwood@arm.com>
16369 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16370 Take a scalar_int_mode instead of a machine_mode.
16371 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16372 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16373 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16374 (aarch64_simd_attr_length_rglist): Delete.
16375 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16376 a scalar_int_mode instead of a machine_mode.
16377 (aarch64_add_offset): Likewise.
16378 (aarch64_internal_mov_immediate): Likewise
16379 (aarch64_add_constant_internal): Likewise.
16380 (aarch64_add_constant): Likewise.
16381 (aarch64_movw_imm): Likewise.
16382 (aarch64_rtx_arith_op_extract_p): Likewise.
16383 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16384 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16385 Remove assert that the mode isn't a vector.
16386 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16387 (aarch64_expand_mov_immediate): Update calls after above changes.
16388 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16389 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16390 (aarch64_move_imm): Likewise.
16391 (aarch64_can_const_movi_rtx_p): Likewise.
16392 (aarch64_strip_extend): Likewise.
16393 (aarch64_extr_rtx_p): Likewise.
16394 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16395 a CONST_INT when the mode parameter is VOIDmode.
16396 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16398 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16400 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16401 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16402 (bitwise_type_for_mode): Update accordingly.
16404 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16406 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16407 * stor-layout.c (mode_for_size_tree): Likewise.
16408 (mode_for_array): Update accordingly.
16409 (layout_decl): Likewise.
16410 (compute_record_mode): Likewise. Only set the mode once.
16412 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16414 * target.def (get_mask_mode): Change return type to opt_mode.
16416 * doc/tm.texi: Regenerate.
16417 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16418 * targhooks.c (default_get_mask_mode): Likewise.
16419 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16420 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16421 targetm.get_mask_mode.
16422 * tree.c (build_truth_vector_type): Likewise.
16424 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16426 * machmode.h (mode_for_vector): Return an opt_mode.
16427 * stor-layout.c (mode_for_vector): Likewise.
16428 (mode_for_int_vector): Update accordingly.
16429 (layout_type): Likewise.
16430 * config/i386/i386.c (emit_memmov): Likewise.
16431 (ix86_expand_set_or_movmem): Likewise.
16432 (ix86_expand_vector_init): Likewise.
16433 (ix86_get_mask_mode): Likewise.
16434 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16436 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16437 * expmed.c (extract_bit_field_1): Likewise.
16438 * expr.c (expand_expr_real_2): Likewise.
16439 * optabs-query.c (can_vec_perm_p): Likewise.
16440 (can_vec_mask_load_store_p): Likewise.
16441 * optabs.c (expand_vec_perm): Likewise.
16442 * targhooks.c (default_get_mask_mode): Likewise.
16443 * tree-vect-stmts.c (vectorizable_store): Likewise.
16444 (vectorizable_load): Likewise.
16445 (get_vectype_for_scalar_type_and_size): Likewise.
16447 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16449 * machmode.h (mode_for_int_vector): New function.
16450 * stor-layout.c (mode_for_int_vector): Likewise.
16451 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16452 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16453 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16454 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16455 (s390_expand_vcond): Likewise.
16457 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16459 * machmode.h (opt_machine_mode): New type.
16460 (opt_mode<T>): Allow construction from anything that can be
16462 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16463 (mode_for_size): Return an opt_machine_mode.
16464 * stor-layout.c (mode_for_size): Likewise.
16465 (mode_for_size_tree): Update call accordingly.
16466 (bitwise_mode_for_mode): Likewise.
16467 (make_fract_type): Likewise.
16468 (make_accum_type): Likewise.
16469 * caller-save.c (replace_reg_with_saved_mem): Update call
16471 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16472 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16473 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16474 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16475 * expmed.c (extract_bit_field_1): Likewise.
16476 * reload.c (get_secondary_mem): Likewise.
16477 * varasm.c (assemble_integer): Likewise.
16478 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16481 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16483 * machmode.h (decimal_float_mode_for_size): New function.
16484 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16485 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16486 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16487 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16488 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16489 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16491 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16493 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16494 (get_builtin_sync_mode): Likewise.
16495 (expand_ifn_atomic_compare_exchange): Likewise.
16496 (expand_builtin_atomic_clear): Likewise.
16497 (expand_builtin_atomic_test_and_set): Likewise.
16498 (fold_builtin_atomic_always_lock_free): Likewise.
16499 * calls.c (compute_argument_addresses): Likewise.
16500 (emit_library_call_value_1): Likewise.
16501 (store_one_arg): Likewise.
16502 * combine.c (combine_instructions): Likewise.
16503 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16504 * config/arm/arm.c (arm_function_value): Likewise.
16505 (aapcs_allocate_return_reg): Likewise.
16506 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16507 * config/i386/i386.c (construct_container): Likewise.
16508 (ix86_gimplify_va_arg): Likewise.
16509 (ix86_expand_sse_cmp): Likewise.
16510 (emit_memmov): Likewise.
16511 (emit_memset): Likewise.
16512 (expand_small_movmem_or_setmem): Likewise.
16513 (ix86_expand_pextr): Likewise.
16514 (ix86_expand_pinsr): Likewise.
16515 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16516 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16518 * config/mips/mips.c (mips_function_value_1) Likewise.
16519 (mips_block_move_straight): Likewise.
16520 (mips_expand_ins_as_unaligned_store): Likewise.
16521 * config/powerpcspe/powerpcspe.c
16522 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16523 (rs6000_darwin64_record_arg_flush): Likewise.
16524 * config/rs6000/rs6000.c
16525 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16526 (rs6000_darwin64_record_arg_flush): Likewise.
16527 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16528 (sparc_function_value_1): Likewise.
16529 * config/spu/spu.c (adjust_operand): Likewise.
16530 (spu_emit_branch_or_set): Likewise.
16531 (arith_immediate_p): Likewise.
16532 * emit-rtl.c (gen_lowpart_common): Likewise.
16533 * expr.c (expand_expr_real_1): Likewise.
16534 * function.c (assign_parm_setup_block): Likewise.
16535 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16536 * reload1.c (alter_reg): Likewise.
16537 * stor-layout.c (mode_for_vector): Likewise.
16538 (layout_type): Likewise.
16540 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16542 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16543 (spu_convert_move): Likewise.
16544 * lower-subreg.c (resolve_simple_move): Likewise.
16546 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16549 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16550 define_insn to a define_expand.
16551 (altivec_vsum2sws_direct): New define_insn.
16552 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16554 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16556 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16559 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16564 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16565 realign stack if stack alignment needed is less than incoming
16568 2017-09-05 Marek Polacek <polacek@redhat.com>
16571 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16574 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16576 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16578 2017-09-05 Richard Biener <rguenther@suse.de>
16580 PR tree-optimization/82084
16581 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16583 2017-09-05 Richard Biener <rguenther@suse.de>
16585 PR tree-optimization/82102
16586 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16588 2017-09-05 Martin Liska <mliska@suse.cz>
16590 PR tree-optimization/82032
16591 * tree-cfg.c (generate_range_test): New function.
16592 * tree-cfg.h (generate_range_test): Declared here.
16593 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16594 (cleanup_control_expr_graph): Use it.
16595 * tree-switch-conversion.c (try_switch_expansion): Remove
16597 (emit_case_nodes): Use generate_range_test.
16599 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16602 * config/i386/i386.md (*<btsc><mode>_mask): Add
16603 TARGET_USE_BT to insn constraint.
16604 (*btr<mode>_mask): Ditto.
16606 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16608 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16609 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16611 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16614 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16615 TARGET_NEON and TARGET_IWMMXT.
16616 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16617 TARGET_NEON and TARGET_IWMMXT.
16618 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16620 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16622 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16623 (ix86_rewrite_tls_address): Ditto.
16624 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16625 (ix86_rewrite_tls_address_1): Ditto.
16626 (ix86_rewrite_tls_address): Ditto.
16627 * config/i386/predicates.md (tls_address_pattern): New predicate.
16628 * config/i386/i386.md (TLS address splitter): New splitter.
16630 2017-09-04 Richard Biener <rguenther@suse.de>
16632 PR tree-optimization/82084
16633 * fold-const.h (can_native_encode_string_p): Declare.
16634 * fold-const.c (can_native_encode_string_p): Factor out from ...
16635 (native_encode_string): ... here.
16636 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16637 vectorizing stores from constants we later cannot handle.
16639 2017-09-04 Marek Polacek <polacek@redhat.com>
16642 * doc/invoke.texi: Update -Wtautological-compare documentation.
16644 2017-09-04 Jeff Law <law@redhat.com>
16646 PR tree-optimization/64910
16647 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16648 swap the first and last operand if the last is a constant.
16650 2017-09-04 Marek Polacek <polacek@redhat.com>
16653 * convert.c (do_narrow): When sanitizing signed integer overflows,
16654 bail out for signed types.
16655 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16657 2017-09-04 Richard Biener <rguenther@suse.de>
16659 PR tree-optimization/82060
16660 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16661 Move devirtualization after stmt folding and before EH/AB/noreturn
16662 cleanup to get the stmt refs canonicalized. Use a bool instead
16663 of gimple_modified_p since that doesn't work for NOPs. Schedule
16664 NOPs generated by folding for removal.
16666 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16667 Alan Hayward <alan.hayward@arm.com>
16668 David Sherwood <david.sherwood@arm.com>
16670 * coretypes.h (pad_direction): New enum.
16671 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16672 (FUNCTION_ARG_PADDING): Likewise.
16673 * target.def (function_arg_padding): New hook.
16674 * targhooks.h (default_function_arg_padding): Declare.
16675 * targhooks.c (default_function_arg_padding): New function.
16676 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16677 (TARGET_FUNCTION_ARG_PADDING): ...this.
16678 * doc/tm.texi: Regenerate.
16679 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16680 instead of direction.
16681 (compute_argument_addresses): Likewise.
16682 (load_register_parameters): Likewise.
16683 (emit_library_call_value_1): Likewise.
16684 (store_one_arg): Use targetm.calls.function_arg_padding instead
16685 of FUNCTION_ARG_PADDING.
16686 (must_pass_in_stack_var_size_or_pad): Likewise.
16687 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16688 (emit_group_store): Likewise.
16689 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16690 instead of FUNCTION_ARG_PADDING.
16691 (emit_push_insn): Likewise, and propagate enum change throughout
16693 * function.h (direction): Delete.
16694 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16696 * function.c (assign_parm_find_stack_rtl): Likewise.
16697 (assign_parm_setup_block_p): Likewise.
16698 (assign_parm_setup_block): Likewise.
16699 (gimplify_parameters): Likewise.
16700 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16701 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16703 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16704 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16705 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16706 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16707 (aarch64_function_arg_padding): ...this new function.
16708 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16709 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16710 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16711 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16712 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16713 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16714 (arm_pad_arg_upward): Replace with...
16715 (arm_function_arg_padding): ...this new function.
16716 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16718 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16719 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16720 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16721 (ia64_hpux_function_arg_padding): Replace with...
16722 (ia64_function_arg_padding): ...this new function. Use pad_direction
16723 instead of direction. Check for TARGET_HPUX.
16724 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16725 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16726 (iq2000_function_arg_padding): New function.
16727 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16728 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16729 (mips_function_arg_padding): ...this new function.
16730 (mips_pad_reg_upward): Update accordingly.
16731 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16732 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16733 targetm.calls.function_arg_padding.
16734 (FUNCTION_ARG_PADDING): Delete.
16735 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16736 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16737 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16738 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16739 (nios2_block_reg_padding): Return pad_direction instead of direction.
16740 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16741 instead of direction.
16742 (nios2_function_arg_padding): Likewise. Make static.
16743 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16744 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16745 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16746 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16747 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16748 (pa_function_arg_padding): Make static. Return pad_direction instead
16750 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16751 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16752 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16753 instead of direction. Use targetm.calls.function_arg_padding.
16754 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16755 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16756 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16757 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16758 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16760 (function_arg_padding): Rename to...
16761 (rs6000_function_arg_padding): ...this. Make static. Return
16762 pad_direction instead of direction.
16763 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16764 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16765 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16766 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16767 instead of direction. Use targetm.calls.function_arg_padding.
16768 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16769 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16770 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16771 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16772 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16773 (function_arg_padding): Rename to...
16774 (rs6000_function_arg_padding): ...this. Make static. Return
16775 pad_direction instead of direction.
16776 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16777 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16778 * config/s390/s390.c (s390_function_arg_padding): New function.
16779 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16780 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16781 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16782 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16783 (function_arg_padding): Rename to...
16784 (sparc_function_arg_padding): ...this. Make static. Return
16785 pad_direction instead of direction.
16786 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16787 * config/spu/spu.c (spu_function_arg_padding): New function.
16788 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16789 * system.h (FUNCTION_ARG_PADDING): Poison.
16791 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16792 Alan Hayward <alan.hayward@arm.com>
16793 David Sherwood <david.sherwood@arm.com>
16795 * target.def (modes_tieable_p): New hook.
16796 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16797 (TARGET_MODES_TIEABLE_P): ...this.
16798 * doc/tm.texi.in: Regenerate.
16799 * hooks.h (hook_bool_mode_mode_true): Declare.
16800 * hooks.c (hook_bool_mode_mode_true): New function.
16801 * combine.c (subst): Use targetm.modes_tieable_p instead of
16803 * dse.c (find_shift_sequence): Likewise.
16804 * expmed.c (extract_low_bits): Likewise.
16805 * lower-subreg.c: Include target.h.
16806 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16808 * rtlanal.c (rtx_cost): Likewise.
16809 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16810 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16811 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16812 (TARGET_MODES_TIEABLE_P): Redefine.
16813 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16814 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16815 (TARGET_MODES_TIEABLE_P): Redefine.
16816 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16817 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16818 (arc_modes_tieable_p): New function.
16819 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16820 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16821 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16822 (arm_modes_tieable_p): Make static.
16823 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16824 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16825 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16826 (TARGET_MODES_TIEABLE_P): Redefine.
16827 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16828 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16829 (TARGET_MODES_TIEABLE_P): Redefine.
16830 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16831 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16832 (cr16_modes_tieable_p): New function.
16833 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16834 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16835 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16836 (TRULY_NOOP_TRUNCATION): Update comment.
16837 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16838 (TRULY_NOOP_TRUNCATION): Update comment.
16839 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16840 (frv_modes_tieable_p): New function.
16841 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16842 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16843 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16844 (TARGET_MODES_TIEABLE_P): Redefine.
16845 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16846 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16847 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16848 (TARGET_MODES_TIEABLE_P): Redefine.
16849 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16850 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16851 (ia64_modes_tieable_p): New function.
16852 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16853 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16854 (iq2000_modes_tieable_p): New function.
16855 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16856 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16857 (lm32_modes_tieable_p): New function.
16858 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16859 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16860 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16861 (TARGET_MODES_TIEABLE_P): Redefine.
16862 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16863 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16864 (m32r_modes_tieable_p): New function.
16865 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16866 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16867 (m68k_modes_tieable_p): New function.
16868 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16869 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16870 (mcore_modes_tieable_p): New function.
16871 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16872 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16874 (TARGET_MODES_TIEABLE_P): Redefine.
16875 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16876 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16877 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16878 (TARGET_MODES_TIEABLE_P): Redefine.
16879 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16880 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16881 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16882 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16883 (mn10300_modes_tieable_p): ...this and make static.
16884 (TARGET_MODES_TIEABLE_P): Redefine.
16885 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16886 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16887 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16888 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16889 (msp430_modes_tieable_p): Make static.
16890 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16891 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16892 (TARGET_MODES_TIEABLE_P): Redefine.
16893 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16894 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16895 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16896 (TARGET_MODES_TIEABLE_P): Redefine.
16897 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16898 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16899 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16900 (TARGET_MODES_TIEABLE_P): Redefine.
16901 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16902 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16903 (pdp11_modes_tieable_p): New function.
16904 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16905 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16906 (rs6000_modes_tieable_p): New function.
16907 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16908 * config/powerpcspe/powerpcspe.md: Update comment.
16909 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16910 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16911 (TARGET_MODES_TIEABLE_P): Redefine.
16912 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16913 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16914 (rl78_modes_tieable_p): New function.
16915 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16916 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16917 (rs6000_modes_tieable_p): New function.
16918 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16919 * config/rs6000/rs6000.md: Update comment.
16920 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16921 * config/rx/rx.c (rx_modes_tieable_p): New function.
16922 (TARGET_MODES_TIEABLE_P): Redefine.
16923 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16924 * config/s390/s390.c (s390_modes_tieable_p): New function.
16925 (TARGET_MODES_TIEABLE_P): Redefine.
16926 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16927 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16928 (sh_modes_tieable_p): New function.
16929 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16930 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16931 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16932 (sparc_modes_tieable_p): Make static.
16933 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16934 * config/spu/spu.c (spu_modes_tieable_p): New function.
16935 (TARGET_MODES_TIEABLE_P): Redefine.
16936 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16937 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16938 (TARGET_MODES_TIEABLE_P): Redefine.
16939 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16940 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16941 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16942 * config/v850/v850.c (v850_modes_tieable_p): New function.
16943 (TARGET_MODES_TIEABLE_P): Redefine.
16944 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16945 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16946 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16947 (visium_modes_tieable_p): New function.
16948 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16949 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16950 (xtensa_modes_tieable_p): New function.
16951 * system.h (MODES_TIEABLE_P): Poison.
16953 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16954 Alan Hayward <alan.hayward@arm.com>
16955 David Sherwood <david.sherwood@arm.com>
16957 * target.def (hard_regno_mode_ok): New hook.
16958 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16959 (TARGET_HARD_REGNO_MODE_OK): ...this.
16960 * doc/tm.texi.in: Regenerate.
16961 * hooks.h (hook_bool_uint_mode_true): Declare.
16962 * hooks.c (hook_bool_uint_mode_true): New function.
16963 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16964 HARD_REGNO_MODE_OK.
16965 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16966 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16967 instead of HARD_REGNO_MODE_OK.
16968 * caller-save.c: Include target.h.
16969 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16970 HARD_REGNO_MODE_OK.
16971 * combine.c (can_combine_p): Likewise.
16972 (combinable_i3pat): Likewise.
16973 (can_change_dest_mode): Likewise.
16974 * expr.c (init_expr_target): Likewise.
16975 (convert_move): Likewise.
16976 (convert_modes): Likewise.
16977 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16978 (setup_prohibited_mode_move_regs): Likewise.
16979 * ira.h (target_ira): Likewise.
16980 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16981 * lra-constraints.c (process_alt_operands): Likewise.
16982 (split_reg): Likewise.
16983 * recog.c (peep2_find_free_register): Likewise.
16984 * ree.c (combine_reaching_defs): Likewise.
16985 * regcprop.c (maybe_mode_change): Likewise.
16986 * reginfo.c (init_reg_sets_1): Likewise.
16987 (choose_hard_reg_mode): Likewise.
16988 (simplifiable_subregs): Likewise.
16989 * regrename.c (check_new_reg_p): Likewise.
16990 * reload.c (find_valid_class): Likewise.
16991 (find_valid_class_1): Likewise.
16992 (reload_inner_reg_of_subreg): Likewise.
16993 (push_reload): Likewise.
16994 (combine_reloads): Likewise.
16995 (find_dummy_reload): Likewise.
16996 (find_reloads): Likewise.
16997 * reload1.c (find_reg): Likewise.
16998 (set_reload_reg): Likewise.
16999 (allocate_reload_reg): Likewise.
17000 (choose_reload_regs): Likewise.
17001 (reload_adjust_reg_for_temp): Likewise.
17002 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17003 (simplify_subreg_regno): Likewise.
17004 * sel-sched.c (init_regs_for_mode): Likewise.
17005 * varasm.c (make_decl_rtl): Likewise.
17006 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17007 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17008 HARD_REGNO_MODE_OK.
17009 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17010 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17011 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17012 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17013 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17014 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17015 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17016 (arc_mode_class): Delete.
17017 (HARD_REGNO_MODE_OK): Delete.
17018 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17019 (arc_hard_regno_mode_ok): Rename old array to...
17020 (arc_hard_regno_mode_ok_modes): ...this.
17021 (arc_conditional_register_usage): Update accordingly.
17022 (arc_mode_class): Make static.
17023 (arc_hard_regno_mode_ok): New function.
17024 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17025 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17026 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17027 (arm_hard_regno_mode_ok): Make static.
17028 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17029 HARD_REGNO_MODE_OK.
17030 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17031 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17032 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17034 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17035 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17036 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17037 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17038 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17039 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17040 * config/bfin/predicates.md (valid_reg_operand): Use
17041 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17042 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17043 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17044 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17045 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17046 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17047 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17048 (cr16_hard_regno_mode_ok): Make static and return a bool.
17049 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17050 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17051 (cris_hard_regno_mode_ok): New function.
17052 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17053 (epiphany_mode_class): Delete.
17054 (HARD_REGNO_MODE_OK): Delete.
17055 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17056 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17057 (hard_regno_mode_ok): Rename to...
17058 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17059 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17060 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17061 HARD_REGNO_MODE_OK.
17062 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17063 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17064 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17065 (frv_hard_regno_mode_ok): Make static and return a bool.
17066 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17067 HARD_REGNO_MODE_OK.
17068 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17069 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17070 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17071 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17073 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17074 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17075 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17076 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17078 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17079 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17080 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17081 (ia64_hard_regno_mode_ok): New function.
17082 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17083 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17084 (iq2000_hard_regno_mode_ok): New function.
17085 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17086 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17087 (lm32_hard_regno_mode_ok): New function.
17088 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17089 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17090 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17091 instead of HARD_REGNO_MODE_OK.
17092 (m32c_hard_regno_ok): Rename to...
17093 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17094 (m32c_cannot_change_mode_class): Update accordingly.
17095 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17096 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17097 (m32r_mode_class): Delete.
17098 (HARD_REGNO_MODE_OK): Delete.
17099 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17100 (m32r_hard_regno_mode_ok): Rename to...
17101 (m32r_hard_regno_modes): ...this.
17102 (m32r_mode_class): Make static.
17103 (m32r_hard_regno_mode_ok): New function.
17104 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17105 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17106 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17107 (m68k_hard_regno_mode_ok): Make static.
17108 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17109 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17110 (mcore_hard_regno_mode_ok): New function.
17111 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17112 (HARD_REGNO_MODE_OK): Delete.
17113 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17115 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17116 (microblaze_hard_regno_mode_ok): New function.
17117 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17118 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17119 (mips_hard_regno_mode_ok): Delete.
17120 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17121 (mips_hard_regno_mode_ok_p): ...this and make static.
17122 (mips_hard_regno_mode_ok_p): Rename to...
17123 (mips_hard_regno_mode_ok_uncached): ...this.
17124 (mips_hard_regno_mode_ok): New function.
17125 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17126 of HARD_REGNO_MODE_OK.
17127 (mips_option_override): Update after above name changes.
17128 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17129 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17130 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17131 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17132 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17133 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17134 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17135 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17136 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17137 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17138 (msp430_hard_regno_mode_ok): Make static and return a bool.
17139 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17140 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17141 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17143 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17144 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17145 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17146 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17147 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17148 (PA_HARD_REGNO_MODE_OK): ...this
17149 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17150 (PA_HARD_REGNO_MODE_OK): ...this.
17151 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17152 (pa_hard_regno_mode_ok): New function.
17153 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17154 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17155 (pdp11_hard_regno_mode_ok): New function.
17156 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17157 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17159 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17161 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17162 (rs6000_hard_regno_mode_ok): Rename to...
17163 (rs6000_hard_regno_mode_ok_uncached): ...this.
17164 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17165 (rs6000_hard_regno_mode_ok): New function.
17166 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17167 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17168 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17169 (riscv_hard_regno_mode_ok): ...this and make static.
17170 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17171 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17172 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17173 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17174 (rl78_hard_regno_mode_ok): Make static and return bool.
17175 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17176 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17178 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17179 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17180 (rs6000_hard_regno_mode_ok): Rename to...
17181 (rs6000_hard_regno_mode_ok_uncached): ...this.
17182 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17183 (rs6000_hard_regno_mode_ok): New function.
17184 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17185 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17186 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17187 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17188 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17189 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17190 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17191 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17192 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17193 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17194 (sh_hard_regno_mode_ok): Make static.
17195 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17196 instead of HARD_REGNO_MODE_OK.
17197 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17198 (sparc_mode_class): Delete.
17199 (HARD_REGNO_MODE_OK): Delete.
17200 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17201 (hard_regno_mode_classes): Make static.
17202 (sparc_mode_class): Likewise.
17203 (sparc_hard_regno_mode_ok): New function.
17204 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17205 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17206 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17208 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17209 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17210 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17211 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17212 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17213 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17214 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17215 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17216 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17217 (visium_hard_regno_mode_ok): New function.
17218 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17219 instead of HARD_REGNO_MODE_OK.
17220 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17221 (HARD_REGNO_MODE_OK): Delete.
17222 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17223 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17224 (xtensa_option_override): Update accordingly.
17225 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17226 (xtensa_hard_regno_mode_ok): New function.
17227 * system.h (HARD_REGNO_MODE_OK): Poison.
17229 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17230 Alan Hayward <alan.hayward@arm.com>
17231 David Sherwood <david.sherwood@arm.com>
17233 * target.def (hard_regno_call_part_clobbered): New hook.
17234 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17235 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17236 * doc/tm.texi: Regenerate.
17237 * hooks.h (hook_bool_uint_mode_false): Declare.
17238 * hooks.c (hook_bool_uint_mode_false): New function.
17239 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17240 * cselib.c (cselib_process_insn): Use
17241 targetm.hard_regno_call_part_clobbered instead of
17242 HARD_REGNO_CALL_PART_CLOBBERED.
17243 * ira-conflicts.c (ira_build_conflicts): Likewise.
17244 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17245 * lra-constraints.c (need_for_call_save_p): Likewise.
17246 * lra-lives.c: Include target.h.
17247 (check_pseudos_live_through_calls): Use
17248 targetm.hard_regno_call_part_clobbered instead of
17249 HARD_REGNO_CALL_PART_CLOBBERED.
17250 * regcprop.c: Include target.h.
17251 (copyprop_hardreg_forward_1): Use
17252 targetm.hard_regno_call_part_clobbered instead of
17253 HARD_REGNO_CALL_PART_CLOBBERED.
17254 * reginfo.c (choose_hard_reg_mode): Likewise.
17255 * regrename.c (check_new_reg_p): Likewise.
17256 * reload.c (find_equiv_reg): Likewise.
17257 * reload1.c (emit_reload_insns): Likewise.
17258 * sched-deps.c (deps_analyze_insn): Likewise.
17259 * sel-sched.c (init_regs_for_mode): Likewise.
17260 (mark_unavailable_hard_regs): Likewise.
17261 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17262 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17263 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17265 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17266 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17267 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17269 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17271 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17272 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17273 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17275 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17276 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17277 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17279 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17280 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17282 * config/powerpcspe/powerpcspe.c
17283 (rs6000_hard_regno_call_part_clobbered): New function.
17284 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17285 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17286 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17288 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17289 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17290 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17292 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17293 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17294 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17296 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17297 Alan Hayward <alan.hayward@arm.com>
17298 David Sherwood <david.sherwood@arm.com>
17300 * rtl.h (subreg_memory_offset): Declare.
17301 * emit-rtl.c (subreg_memory_offset): New function.
17302 * expmed.c (store_bit_field_1): Use it.
17303 * expr.c (undefined_operand_subword_p): Likewise.
17304 * simplify-rtx.c (simplify_subreg): Likewise.
17306 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17308 PR rtl-optimization/57448
17311 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17312 using atomic_load pattern.
17313 (expand_atomic_store): Likewise.
17315 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17318 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17319 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17320 handling. Use replace_call_with_value with NULL instead of
17321 gsi_replace, unlink_stmt_vdef and release_defs.
17323 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17326 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17328 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17331 * rtl.h (emit_library_call_value_1): Declare.
17332 (emit_library_call): Replace declaration with a series of overloads.
17333 Remove the parameter count argument.
17334 (emit_library_call_value): Likewise.
17335 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17336 with an "rtx_mode_t *".
17337 (emit_library_call_value): Delete.
17338 (emit_library_call): Likewise.
17339 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17340 (asan_emit_allocas_unpoison): Likewise.
17341 * builtins.c (expand_builtin_powi): Likewise.
17342 (expand_asan_emit_allocas_unpoison): Likewise.
17343 * cfgexpand.c (expand_main_function): Likewise.
17344 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17345 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17346 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17347 * config/arm/arm.c (arm_trampoline_init): Likewise.
17348 (arm_call_tls_get_addr): Likewise.
17349 (arm_expand_divmod_libfunc): Likewise.
17350 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17351 (smulsi3_highpart): Likewise.
17352 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17353 (c6x_expand_compare): Likewise.
17354 (c6x_expand_movmem): Likewise.
17355 * config/frv/frv.c (frv_trampoline_init): Likewise.
17356 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17357 (ix86_expand_divmod_libfunc): Likewise.
17358 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17359 (ia64_expand_compare): Likewise.
17360 (ia64_profile_hook): Likewise.
17361 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17362 (nonlocal_goto): Likewise.
17363 (restore_stack_nonlocal): Likewise.
17364 * config/m32r/m32r.c (block_move_call): Likewise.
17365 (m32r_trampoline_init): Likewise.
17366 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17367 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17368 (m68k_call_m68k_read_tp): Likewise.
17369 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17370 (microblaze_expand_divide): Likewise.
17371 * config/mips/mips.h (mips_args): Likewise.
17372 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17373 (MIPS_ICACHE_SYNC): Likewise.
17374 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17375 (nios2_trampoline_init): Likewise.
17376 * config/pa/pa.c (hppa_tls_call): Likewise.
17377 (pa_trampoline_init): Likewise.
17378 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17379 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17380 (expand_strn_compare): Likewise.
17381 (rs6000_generate_compare): Likewise.
17382 (rs6000_expand_float128_convert): Likewise.
17383 (output_profile_hook): Likewise.
17384 (rs6000_trampoline_init): Likewise.
17385 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17386 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17387 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17388 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17389 (rs6000_generate_compare): Likewise.
17390 (rs6000_expand_float128_convert): Likewise.
17391 (output_profile_hook): Likewise.
17392 (rs6000_trampoline_init): Likewise.
17393 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17394 * config/sh/sh.c (sh_trampoline_init): Likewise.
17395 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17396 (sparc_emit_float_lib_cmp): Likewise.
17397 (sparc32_initialize_trampoline): Likewise.
17398 (sparc64_initialize_trampoline): Likewise.
17399 (sparc_profile_hook): Likewise.
17400 * config/spu/spu.c (ea_load_store): Likewise.
17401 * config/spu/spu.md (floatunssidf2): Likewise.
17402 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17403 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17404 * config/visium/visium.c (expand_block_move_4): Likewise.
17405 (expand_block_move_2): Likewise.
17406 (expand_block_move_1): Likewise.
17407 (expand_block_set_4): Likewise.
17408 (expand_block_set_2): Likewise.
17409 (expand_block_set_1): Likewise.
17410 (visium_trampoline_init): Likewise.
17411 (visium_profile_hook): Likewise.
17412 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17413 (xtensa_setup_frame_addresses): Likewise.
17414 (xtensa_trampoline_init): Likewise.
17415 * except.c (sjlj_emit_function_enter): Likewise.
17416 (sjlj_emit_function_exit): Likewise.
17417 * explow.c (allocate_dynamic_stack_space): Likewise.
17418 (probe_stack_range): Likewise.
17419 * expr.c (convert_mode_scalar): Likewise.
17420 * optabs.c (expand_binop): Likewise.
17421 (expand_twoval_binop_libfunc): Likewise.
17422 (expand_unop): Likewise.
17423 (prepare_cmp_insn): Likewise.
17424 (prepare_float_lib_cmp): Likewise.
17425 (expand_float): Likewise.
17426 (expand_fix): Likewise.
17427 (expand_fixed_convert): Likewise.
17428 (maybe_emit_sync_lock_test_and_set): Likewise.
17429 (expand_atomic_compare_and_swap): Likewise.
17430 (expand_mem_thread_fence): Likewise.
17431 (expand_atomic_fetch_op): Likewise.
17433 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17435 * doc/generic.texi (OpenACC): Adjust URL.
17436 * doc/invoke.texi (C Dialect Options): Ditto.
17438 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17440 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17441 predicate for operand 1. Add (m,<S>) constraint.
17442 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17443 Prevent memory operand 1 with register operand 2.
17445 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17447 PR rtl-optimization/82024
17448 * combine.c (try_combine): If the combination result is a PARALLEL,
17449 and we only need to retain the SET in there that would be placed
17450 at I2, check that we can place that at I3 instead, before doing so.
17452 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17455 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17457 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17458 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17461 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17462 Jeff Law <law@redhat.com>
17464 * varasm.c (bss_initializer_p): Do not put constants into .bss
17465 (categorize_decl_for_section): Handle bss_initializer_p returning
17466 false when DECL_INITIAL is NULL.
17468 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17471 * config/s390/s390.c (s390_can_inline_p): New function.
17473 2017-09-01 Jeff Law <law@redhat.com>
17475 PR tree-optimization/82052
17476 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17477 Always initialize the returned slot after a hash table miss
17478 when INSERT is true.
17480 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17482 * config/s390/s390.md (mem_signal_fence): Remove.
17483 * doc/md.texi (mem_signal_fence): Remove.
17484 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17486 * target-insns.def (mem_signal_fence): Remove.
17488 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17491 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17494 * asan.c (create_odr_indicator): Strip name encoding from assembler
17495 name before appending it after __odr_asan_.
17497 2017-09-01 Martin Liska <mliska@suse.cz>
17499 PR tree-optimization/82059
17500 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17501 frequency only when an edge is redirected.
17503 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17505 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17506 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17507 (arc_conditional_register_usage): Remove ARC600 lp_count
17509 (arc_file_start): Emit Tag_ARC_CPU_variation.
17510 (arc_can_use_doloop_p): New conditions to use ZOLs.
17511 (hwloop_fail): New function.
17512 (hwloop_optimize): Likewise.
17513 (hwloop_pattern_reg): Likewise.
17514 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17515 (arc_reorg_loops): New function, calls reorg_loops.
17516 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17517 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17519 (arc_loop_hazard): Remove function, functionality moved into
17521 (arc_hazard): Remove arc_loop_hazard call.
17522 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17523 into hwloop_optimize.
17524 (arc_label_align): Remove ZOL handling.
17525 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17526 * config/arc/arc.md (doloop_begin): Remove pattern.
17527 (doloop_begin_i): Likewise.
17528 (doloop_end_i): Likewise.
17529 (doloop_fallback): Likewise.
17530 (doloop_fallback_m): Likewise.
17531 (doloop_end): Reimplement expand.
17532 (arc_lp): New pattern for LP instruction.
17533 (loop_end): New pattern.
17534 (loop_fail): Likewise.
17535 (decrement_and_branch_until_zero): Likewise.
17536 * config/arc/arc.opt (mlpc-width): New option.
17537 * doc/invoke.texi (mlpc-width): Document option.
17539 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17541 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17542 (arc_ccfsm_advance): Fix checking for delay slots.
17543 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17545 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17547 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17549 (movhi_insn): Update store instruction constraint which are saving
17550 6-bit short immediates.
17551 (movsi_insn): Consider also short scaled load operations.
17552 (zero_extendhisi2_i): Use Usd constraint instead of T.
17553 (extendhisi2_i): Add q constraint.
17554 (arc_clzsi2): Add type and length attributes.
17555 (arc_ctzsi2): Likewise.
17556 * config/arc/constraints.md (Usc): Update constraint, the
17557 assembler can parse two relocations for a single instruction.
17559 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17561 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17562 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17564 2017-08-31 Olivier Hainque <hainque@adacore.com>
17566 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17567 match as powerpc-wrs-vxworks*.
17569 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17571 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17572 register constraint for by-element operand.
17573 (aarch64_mls_elt_merge<mode>): Likewise.
17575 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17577 * config/arc/arc.c (arc_can_follow_jump): Check for short
17580 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17582 * config.gcc: Use g.opt for arc.
17583 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17584 functionality moved to ...
17585 (legitimate_scaled_address_p): New function, ...here.
17586 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17587 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17588 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17590 (arc_override_options): Handle G option.
17591 (arc_output_pic_addr_const): Correct function definition.
17592 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17593 (arc_decl_anon_ns_mem_p): Delete.
17594 (arc_in_small_data_p): Overhaul this function to take into
17595 consideration the value given via G option.
17596 (arc_rewrite_small_data_1): Renamed and corrected old
17597 arc_rewrite_small_data function.
17598 (arc_rewrite_small_data): New function.
17599 (small_data_pattern): Don't use pic_offset_table_rtx.
17600 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17601 * config/arc/simdext.md (movmisalignv2hi): Use
17602 prepare_move_operands function.
17604 (movmisalign*): Likewise.
17605 * doc/invoke.texi (ARC options): Document -G option.
17607 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17609 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17611 * config/arc/arc.c (arc_print_operand): Output scalled address for
17612 sdata whenever is possible.
17613 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17614 load/stores are available.
17615 (compact_sda_memory_operand): Check for the alignment required by
17616 code density instructions.
17617 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17619 * config/arc/constraints.md (Usd): Update constraint.
17620 (Us0): New constraint.
17621 (Usc): Update constraint.
17623 2017-08-31 Richard Biener <rguenther@suse.de>
17625 PR middle-end/82054
17626 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17627 function only once.
17629 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17631 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17632 Resize type_signature.
17634 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17635 Alan Hayward <alan.hayward@arm.com>
17636 David Sherwood <david.sherwood@arm.com>
17638 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17639 subregs whose inner modes can be stored in GPRs.
17640 (aarch64_classify_index): Likewise.
17642 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17643 Alan Hayward <alan.hayward@arm.com>
17644 David Sherwood <david.sherwood@arm.com>
17646 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17647 (V_INT_EQUIV): ...this.
17648 (v_cmp_result): Rename to...
17649 (v_int_equiv): ...this.
17650 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17651 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17652 (copysign<mode>3): Likewise.
17653 (aarch64_simd_bsl<mode>_internal): Likewise.
17654 (aarch64_simd_bsl<mode>): Likewise.
17655 (vec_cmp<mode><mode>): Likewise.
17656 (vcond<mode><mode>): Likewise.
17657 (vcond<v_cmp_mixed><mode>): Likewise.
17658 (vcondu<mode><v_cmp_mixed>): Likewise.
17659 (aarch64_cm<optab><mode>): Likewise.
17660 (aarch64_cmtst<mode>): Likewise.
17661 (aarch64_fac<optab><mode>): Likewise.
17662 (vec_perm_const<mode>): Likewise.
17663 (vcond_mask_<mode><v_cmp_result>): Rename to...
17664 (vcond_mask_<mode><v_int_equiv>): ...this.
17665 (vec_cmp<mode><v_cmp_result>): Rename to...
17666 (vec_cmp<mode><v_int_equiv>): ...this.
17668 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17669 Alan Hayward <alan.hayward@arm.com>
17670 David Sherwood <david.sherwood@arm.com>
17672 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17674 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17675 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17676 (UNSPEC_LD4_DREG): New unspecs.
17677 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17678 (aarch64_ld2<mode>_dreg_be): Replace with...
17679 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17681 (aarch64_ld3<mode>_dreg_le)
17682 (aarch64_ld3<mode>_dreg_be): Replace with...
17683 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17685 (aarch64_ld4<mode>_dreg_le)
17686 (aarch64_ld4<mode>_dreg_be): Replace with...
17687 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17690 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17692 PR tree-optimization/81987
17693 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17694 insert an initializer in a location not dominated by the stride
17697 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17699 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17700 on the entire header of the finally block in the fallthru case.
17702 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17704 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17706 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17708 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17709 rs6000_emit_move_from_cr and call renamed function.
17710 (rs6000_emit_prologue): Call renamed functions.
17711 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17712 movesi_from_cr, remove volatile CRs.
17714 2017-08-30 Jon Beniston <jon@beniston.com>
17715 Richard Biener <rguenther@suse.de>
17717 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17718 of VECTOR_MODE_P check.
17719 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17720 element vector types.
17722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17724 * df.h (df_read_modify_subreg_p): Remove in favor of...
17725 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17726 const_rtx instead of an rtx.
17727 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17728 * df-problems.c (df_word_lr_mark_ref): Likewise.
17729 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17730 (mark_pseudo_reg_dead): Likewise.
17731 (mark_ref_dead): Likewise.
17732 * reginfo.c (init_subregs_of_mode): Likewise.
17733 * sched-deps.c (sched_analyze_1): Likewise.
17734 * df-scan.c (df_def_record_1): Likewise.
17735 (df_uses_record): Likewise.
17736 (df_read_modify_subreg_p): Remove in favor of...
17737 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17738 const_rtx instead of an rtx.
17740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17741 Alan Hayward <alan.hayward@arm.com>
17742 David Sherwood <david.sherwood@arm.com>
17744 * rtl.h (partial_subreg_p): New function.
17745 * caller-save.c (save_call_clobbered_regs): Use it.
17746 * calls.c (expand_call): Likewise.
17747 * combine.c (combinable_i3pat): Likewise.
17748 (simplify_set): Likewise.
17749 (make_extraction): Likewise.
17750 (make_compound_operation_int): Likewise.
17751 (gen_lowpart_or_truncate): Likewise.
17752 (force_to_mode): Likewise.
17753 (make_field_assignment): Likewise.
17754 (reg_truncated_to_mode): Likewise.
17755 (record_truncated_value): Likewise.
17756 (move_deaths): Likewise.
17757 * cse.c (record_jump_cond): Likewise.
17758 (cse_insn): Likewise.
17759 * cselib.c (cselib_lookup_1): Likewise.
17760 * expmed.c (extract_bit_field_using_extv): Likewise.
17761 * function.c (assign_parm_setup_reg): Likewise.
17762 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17763 * ira-build.c (create_insn_allocnos): Likewise.
17764 * lra-coalesce.c (merge_pseudos): Likewise.
17765 * lra-constraints.c (match_reload): Likewise.
17766 (simplify_operand_subreg): Likewise.
17767 (curr_insn_transform): Likewise.
17768 * lra-lives.c (process_bb_lives): Likewise.
17769 * lra.c (new_insn_reg): Likewise.
17770 (lra_substitute_pseudo): Likewise.
17771 * regcprop.c (mode_change_ok): Likewise.
17772 (maybe_mode_change): Likewise.
17773 (copyprop_hardreg_forward_1): Likewise.
17774 * reload.c (push_reload): Likewise.
17775 (find_reloads): Likewise.
17776 (find_reloads_subreg_address): Likewise.
17777 * reload1.c (alter_reg): Likewise.
17778 (eliminate_regs_1): Likewise.
17779 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17781 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17783 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17784 back to if statements, including unpack.
17786 2017-08-30 Martin Liska <mliska@suse.cz>
17788 PR inline-asm/82001
17789 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17791 (func_checker::compare_asm_inputs_outputs): ... this function.
17792 (func_checker::compare_gimple_asm): Use the function to compare
17793 also ASM constrains.
17794 * ipa-icf-gimple.h: Rename the function.
17796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17797 Alan Hayward <alan.hayward@arm.com>
17798 David Sherwood <david.sherwood@arm.com>
17800 * coretypes.h (complex_mode): New type.
17801 * gdbhooks.py (build_pretty_printer): Handle it.
17802 * machmode.h (complex_mode): New class.
17803 (complex_mode::includes_p): New function.
17804 (is_complex_int_mode): Likewise.
17805 (is_complex_float_mode): Likewise.
17806 * genmodes.c (get_mode_class): Handle complex mode classes.
17807 * function.c (expand_function_end): Use is_complex_int_mode.
17809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17810 Alan Hayward <alan.hayward@arm.com>
17811 David Sherwood <david.sherwood@arm.com>
17813 * coretypes.h (scalar_mode_pod): New typedef.
17814 * gdbhooks.py (build_pretty_printer): Handle it.
17815 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17816 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17817 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17818 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17819 as_a <scalar_mode>.
17821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17822 Alan Hayward <alan.hayward@arm.com>
17823 David Sherwood <david.sherwood@arm.com>
17825 * machmode.h (mode_for_vector): Take a scalar_mode instead
17827 * stor-layout.c (mode_for_vector): Likewise.
17828 * explow.c (promote_mode): Use as_a <scalar_mode>.
17829 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17832 Alan Hayward <alan.hayward@arm.com>
17833 David Sherwood <david.sherwood@arm.com>
17835 * target.def (preferred_simd_mode): Take a scalar_mode
17836 instead of a machine_mode.
17837 * targhooks.h (default_preferred_simd_mode): Likewise.
17838 * targhooks.c (default_preferred_simd_mode): Likewise.
17839 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17840 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17841 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17842 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17843 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17844 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17845 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17846 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17848 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17849 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17850 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17851 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17852 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17853 * doc/tm.texi: Regenerate.
17854 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17858 Alan Hayward <alan.hayward@arm.com>
17859 David Sherwood <david.sherwood@arm.com>
17861 * target.def (scalar_mode_supported_p): Take a scalar_mode
17862 instead of a machine_mode.
17863 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17864 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17865 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17866 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17867 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17868 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17869 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17870 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17871 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17872 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17873 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17874 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17875 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17876 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17878 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17879 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17880 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17881 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17882 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17884 * doc/tm.texi: Regenerate.
17886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17887 Alan Hayward <alan.hayward@arm.com>
17888 David Sherwood <david.sherwood@arm.com>
17890 * coretypes.h (opt_scalar_mode): New typedef.
17891 * gdbhooks.py (build_pretty_printers): Handle it.
17892 * machmode.h (mode_iterator::get_2xwider): Add overload for
17894 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17896 * expr.c (convert_mode_scalar): Likewise.
17897 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17898 * optabs.c (expand_float): Likewise.
17899 (expand_fix): Likewise.
17900 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17903 Alan Hayward <alan.hayward@arm.com>
17904 David Sherwood <david.sherwood@arm.com>
17906 * optabs.c (expand_float): Explicitly check for scalars before
17907 using a branching expansion.
17908 (expand_fix): Likewise.
17910 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17911 Alan Hayward <alan.hayward@arm.com>
17912 David Sherwood <david.sherwood@arm.com>
17914 * expr.c (convert_mode): Split scalar handling out into...
17915 (convert_mode_scalar): ...this new function. Treat the modes
17918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17919 Alan Hayward <alan.hayward@arm.com>
17920 David Sherwood <david.sherwood@arm.com>
17922 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17924 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17926 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17927 Alan Hayward <alan.hayward@arm.com>
17928 David Sherwood <david.sherwood@arm.com>
17930 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17931 rather than a machine_mode.
17932 (fixed_from_string): Likewise.
17933 (fixed_convert): Likewise.
17934 (fixed_convert_from_int): Likewise.
17935 (fixed_convert_from_real): Likewise.
17936 (real_convert_from_fixed): Likewise.
17937 * fixed-value.c (fixed_from_double_int): Likewise.
17938 (fixed_from_string): Likewise.
17939 (fixed_convert): Likewise.
17940 (fixed_convert_from_int): Likewise.
17941 (fixed_convert_from_real): Likewise.
17942 (real_convert_from_fixed): Likewise.
17943 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17946 Alan Hayward <alan.hayward@arm.com>
17947 David Sherwood <david.sherwood@arm.com>
17949 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17950 of separate mode class checks. Do not allow vector modes here.
17951 (immed_wide_int_const): Use as_a <scalar_mode>.
17952 * explow.c (trunc_int_for_mode): Likewise.
17953 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17954 (wi::shwi): Likewise.
17955 (wi::min_value): Likewise.
17956 (wi::max_value): Likewise.
17957 * dwarf2out.c (loc_descriptor): Likewise.
17958 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17959 for CONST_WIDE_INT.
17961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17962 Alan Hayward <alan.hayward@arm.com>
17963 David Sherwood <david.sherwood@arm.com>
17965 * tree.h (SCALAR_TYPE_MODE): New macro.
17966 * expr.c (expand_expr_addr_expr_1): Use it.
17967 (expand_expr_real_2): Likewise.
17968 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17969 (fold_convert_const_fixed_from_int): Likewise.
17970 (fold_convert_const_fixed_from_real): Likewise.
17971 (native_encode_fixed): Likewise
17972 (native_encode_complex): Likewise
17973 (native_encode_vector): Likewise.
17974 (native_interpret_fixed): Likewise.
17975 (native_interpret_real): Likewise.
17976 (native_interpret_complex): Likewise.
17977 (native_interpret_vector): Likewise.
17978 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17979 (simd_clone_adjust_argument_types): Likewise.
17980 (simd_clone_init_simd_arrays): Likewise.
17981 (simd_clone_adjust): Likewise.
17982 * stor-layout.c (layout_type): Likewise.
17983 * tree.c (build_minus_one_cst): Likewise.
17984 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17985 * tree-inline.c (estimate_move_cost): Likewise.
17986 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17987 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17988 (vectorizable_reduction): Likewise.
17989 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17990 (vect_recog_mixed_size_cond_pattern): Likewise.
17991 (check_bool_pattern): Likewise.
17992 (adjust_bool_pattern): Likewise.
17993 (search_type_for_mask_1): Likewise.
17994 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17995 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17996 (vectorizable_load): Likewise.
17997 (vectorizable_store): Likewise.
17998 * ubsan.c (ubsan_encode_value): Likewise.
17999 * varasm.c (output_constant): Likewise.
18001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18002 Alan Hayward <alan.hayward@arm.com>
18003 David Sherwood <david.sherwood@arm.com>
18005 * coretypes.h (scalar_mode): New class.
18006 * machmode.h (scalar_mode): Likewise.
18007 (scalar_mode::includes_p): New function.
18008 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18009 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18010 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18011 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18012 * expmed.c (store_bit_field_1): Likewise.
18013 (extract_bit_field_1): Likewise.
18014 * expr.c (write_complex_part): Likewise.
18015 (read_complex_part): Likewise.
18016 (emit_move_complex_push): Likewise.
18017 (expand_expr_real_2): Likewise.
18018 * function.c (assign_parm_setup_reg): Likewise.
18019 (assign_parms_unsplit_complex): Likewise.
18020 * optabs.c (expand_binop): Likewise.
18021 * rtlanal.c (subreg_get_info): Likewise.
18022 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18023 * varasm.c (output_constant_pool_2): Likewise.
18025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18026 Alan Hayward <alan.hayward@arm.com>
18027 David Sherwood <david.sherwood@arm.com>
18029 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18031 (expmed_mult_highpart_optab): Likewise.
18032 (expmed_mult_highpart): Likewise.
18034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18035 Alan Hayward <alan.hayward@arm.com>
18036 David Sherwood <david.sherwood@arm.com>
18038 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18039 instead of a machine_mode.
18040 (builtin_memset_read_str): Likewise.
18041 * builtins.c (c_readstr): Likewise.
18042 (builtin_memcpy_read_str): Likewise.
18043 (builtin_strncpy_read_str): Likewise.
18044 (builtin_memset_read_str): Likewise.
18045 (builtin_memset_gen_str): Likewise.
18046 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18047 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18048 instead of a machine_mode.
18049 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18051 (make_extraction): Likewise.
18052 (try_widen_shift_mode): Take and return scalar_int_modes instead
18054 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18055 a scalar_int_mode instead of a machine_mode.
18056 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18057 (avr_addr_space_pointer_mode): Likewise.
18058 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18059 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18060 (msp430_unwind_word_mode): Likewise.
18061 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18062 (spu_addr_space_pointer_mode): Likewise.
18063 (spu_addr_space_address_mode): Likewise.
18064 (spu_libgcc_cmp_return_mode): Likewise.
18065 (spu_libgcc_shift_count_mode): Likewise.
18066 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18067 (rl78_addr_space_pointer_mode): Likewise.
18068 (fl78_unwind_word_mode): Likewise.
18069 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18071 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18072 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18073 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18074 (mips_valid_pointer_mode): Likewise.
18075 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18076 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18077 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18079 (ft32_addr_space_address_mode): Likewise.
18080 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18081 scalar_int_mode instead of a machine_mode.
18082 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18084 (m32c_addr_space_address_mode): Likewise.
18085 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18086 (rs6000_eh_return_filter_mode): Likewise.
18087 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18088 (rs6000_eh_return_filter_mode): Likewise.
18089 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18090 (s390_libgcc_shift_count_mode): Likewise.
18091 (s390_unwind_word_mode): Likewise.
18092 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18094 * target.def (mode_rep_extended): Likewise.
18095 (valid_pointer_mode): Likewise.
18096 (addr_space.valid_pointer_mode): Likewise.
18097 (eh_return_filter_mode): Return a scalar_int_mode rather than
18099 (libgcc_cmp_return_mode): Likewise.
18100 (libgcc_shift_count_mode): Likewise.
18101 (unwind_word_mode): Likewise.
18102 (addr_space.pointer_mode): Likewise.
18103 (addr_space.address_mode): Likewise.
18104 * doc/tm.texi: Regenerate.
18105 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18107 (do_jump): Use scalar_int_mode for local variables.
18108 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18109 rather than a machine_mode.
18110 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18111 (scompare_loc_descriptor_wide): Likewise.
18112 (scompare_loc_descriptor_narrow): Likewise.
18113 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18115 * except.c (sjlj_emit_dispatch_table): Likewise.
18116 (expand_builtin_eh_copy_values): Likewise.
18117 * explow.c (convert_memory_address_addr_space_1): Likewise.
18118 Take a scalar_int_mode rather than a machine_mode.
18119 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18120 than a machine_mode.
18121 (memory_address_addr_space): Use scalar_int_mode for local variables.
18122 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18123 rather than a machine_mode.
18124 * expmed.c (mask_rtx): Likewise.
18125 (init_expmed_one_conv): Likewise.
18126 (expand_mult_highpart_adjust): Likewise.
18127 (extract_high_half): Likewise.
18128 (expmed_mult_highpart_optab): Likewise.
18129 (expmed_mult_highpart): Likewise.
18130 (expand_smod_pow2): Likewise.
18131 (expand_sdiv_pow2): Likewise.
18132 (emit_store_flag_int): Likewise.
18133 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18135 (extract_low_bits): Likewise.
18136 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18138 * expr.c (pieces_addr::adjust): Likewise.
18139 (can_store_by_pieces): Likewise.
18140 (store_by_pieces): Likewise.
18141 (clear_by_pieces_1): Likewise.
18142 (expand_expr_addr_expr_1): Likewise.
18143 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18144 (expand_expr_real_1): Likewise.
18145 (try_casesi): Likewise.
18146 * final.c (shorten_branches): Likewise.
18147 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18148 type of "mode" to machine_mode.
18149 * internal-fn.c (expand_arith_overflow_result_store): Take a
18150 scalar_int_mode rather than a machine_mode.
18151 (expand_mul_overflow): Use scalar_int_mode for local variables.
18152 * loop-doloop.c (doloop_modify): Likewise.
18153 (doloop_optimize): Likewise.
18154 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18155 than a machine_mode.
18156 (expand_doubleword_shift_condmove): Likewise.
18157 (expand_doubleword_shift): Likewise.
18158 (expand_doubleword_clz): Likewise.
18159 (expand_doubleword_popcount): Likewise.
18160 (expand_doubleword_parity): Likewise.
18161 (expand_absneg_bit): Use scalar_int_mode for local variables.
18162 (prepare_float_lib_cmp): Likewise.
18163 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18164 rather than a machine_mode.
18165 (convert_memory_address_addr_space): Likewise.
18166 (get_mode_bounds): Likewise.
18167 (get_address_mode): Return a scalar_int_mode rather than a
18169 * rtlanal.c (get_address_mode): Likewise.
18170 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18171 than a machine_mode.
18172 * targhooks.c (default_mode_rep_extended): Likewise.
18173 (default_valid_pointer_mode): Likewise.
18174 (default_addr_space_valid_pointer_mode): Likewise.
18175 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18176 than a machine_mode.
18177 (default_libgcc_cmp_return_mode): Likewise.
18178 (default_libgcc_shift_count_mode): Likewise.
18179 (default_unwind_word_mode): Likewise.
18180 (default_addr_space_pointer_mode): Likewise.
18181 (default_addr_space_address_mode): Likewise.
18182 * targhooks.h (default_eh_return_filter_mode): Likewise.
18183 (default_libgcc_cmp_return_mode): Likewise.
18184 (default_libgcc_shift_count_mode): Likewise.
18185 (default_unwind_word_mode): Likewise.
18186 (default_addr_space_pointer_mode): Likewise.
18187 (default_addr_space_address_mode): Likewise.
18188 (default_mode_rep_extended): Take a scalar_int_mode rather than
18190 (default_valid_pointer_mode): Likewise.
18191 (default_addr_space_valid_pointer_mode): Likewise.
18192 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18194 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18195 rather than a machine_mode.
18196 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18197 for local variables.
18198 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18199 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18200 than a machine_mode.
18202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18203 Alan Hayward <alan.hayward@arm.com>
18204 David Sherwood <david.sherwood@arm.com>
18206 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18207 the mode argument to scalar_int_mode.
18208 (do_jump_by_parts_zero_rtx): Likewise.
18209 (do_jump_by_parts_equality_rtx): Likewise.
18210 (do_jump_by_parts_greater): Take a mode argument.
18211 (do_jump_by_parts_equality): Likewise.
18212 (do_jump_1): Update calls accordingly.
18214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18215 Alan Hayward <alan.hayward@arm.com>
18216 David Sherwood <david.sherwood@arm.com>
18218 * is-a.h (safe_dyn_cast): New function.
18219 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18220 (jump_table_for_label): Likewise.
18221 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18222 instead of an rtx_insn *.
18223 (shorten_branches): Use dyn_cast instead of LABEL_P and
18224 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18225 rtx_jump_table_data::get_data_mode.
18226 (final_scan_insn): Likewise.
18228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18229 Alan Hayward <alan.hayward@arm.com>
18230 David Sherwood <david.sherwood@arm.com>
18232 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18233 trying to combine a full-register integer set with a subreg
18236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18237 Alan Hayward <alan.hayward@arm.com>
18238 David Sherwood <david.sherwood@arm.com>
18240 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18241 that is always either address_mode or pointer_mode.
18243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18244 Alan Hayward <alan.hayward@arm.com>
18245 David Sherwood <david.sherwood@arm.com>
18247 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18248 when the two are known to be equal.
18250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18251 Alan Hayward <alan.hayward@arm.com>
18252 David Sherwood <david.sherwood@arm.com>
18254 * simplify-rtx.c (simplify_const_unary_operation): Use
18255 is_a <scalar_int_mode> instead of checking for a nonzero
18256 precision. Forcibly convert op_mode to a scalar_int_mode
18257 in that case. More clearly differentiate the operand and
18258 result modes and use the former when deciding what the value
18259 of a count-bits operation should be. Use is_int_mode instead
18260 of checking for a MODE_INT. Remove redundant check for whether
18261 this mode has a zero precision.
18263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18264 Alan Hayward <alan.hayward@arm.com>
18265 David Sherwood <david.sherwood@arm.com>
18267 * optabs.c (widen_leading): Change the type of the mode argument
18268 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18269 (widen_bswap): Likewise.
18270 (expand_parity): Likewise.
18271 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18272 (expand_ffs): Likewise.
18273 (epand_unop): Check for scalar integer modes before calling the
18276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18277 Alan Hayward <alan.hayward@arm.com>
18278 David Sherwood <david.sherwood@arm.com>
18280 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18282 (expand_expr_real_1): Update call accordingly.
18284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18285 Alan Hayward <alan.hayward@arm.com>
18286 David Sherwood <david.sherwood@arm.com>
18288 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18289 value_mode arguments. Use scalar_int_mode internally.
18290 (store_bit_field_1): Rename the new integer mode from imode
18291 to op0_mode and use it instead of GET_MODE (op0). Update calls
18292 to store_split_bit_field, store_bit_field_using_insv and
18293 store_fixed_bit_field.
18294 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18295 Use scalar_int_mode internally. Use a bit count rather than a mode
18296 when calculating the largest bit size for get_best_mode.
18297 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18298 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18299 Remove assertion that OP0 has a scalar integer mode.
18300 (store_split_bit_field): Add op0_mode and value_mode arguments.
18301 Update calls to extract_fixed_bit_field.
18302 (extract_bit_field_using_extv): Add an op0_mode argument.
18303 Use scalar_int_mode internally.
18304 (extract_bit_field_1): Rename the new integer mode from imode to
18305 op0_mode and use it instead of GET_MODE (op0). Update calls to
18306 extract_split_bit_field, extract_bit_field_using_extv and
18307 extract_fixed_bit_field.
18308 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18309 to extract_split_bit_field and extract_fixed_bit_field_1.
18310 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18311 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18312 on the target mode.
18313 (extract_split_bit_field): Add an op0_mode argument. Update call
18314 to extract_fixed_bit_field.
18316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18317 Alan Hayward <alan.hayward@arm.com>
18318 David Sherwood <david.sherwood@arm.com>
18320 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18321 * explow.c (hard_function_value): Likewise.
18322 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18323 convert_to_mode call outside the loop.
18324 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18325 for the mode iterator. Require the mode specified by max_pieces
18327 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18329 (copy_blkmode_to_reg): Likewise.
18330 (set_storage_via_setmem): Likewise.
18331 * optabs.c (prepare_cmp_insn): Likewise.
18332 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18333 * stor-layout.c (finish_bitfield_representative): Likewise.
18335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18336 Alan Hayward <alan.hayward@arm.com>
18337 David Sherwood <david.sherwood@arm.com>
18339 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18340 * expr.c (convert_move): Use them.
18341 (convert_modes): Likewise.
18342 (store_expr_with_bounds): Likewise.
18344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18345 Alan Hayward <alan.hayward@arm.com>
18346 David Sherwood <david.sherwood@arm.com>
18348 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18349 parameter for the mode of "x". Remove the "known_x", "known_mode"
18350 and "known_ret" arguments. Change the type of the mode argument
18351 to scalar_int_mode.
18352 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18353 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18354 (reg_num_sign_bit_copies_for_combine): Likewise.
18355 * rtlanal.c (nonzero_bits1): Likewise.
18356 (num_sign_bit_copies1): Likewise.
18357 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18358 (reg_num_sign_bit_copies_general): Likewise.
18359 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18360 (reg_nonzero_bits_general): Likewise.
18362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18363 Alan Hayward <alan.hayward@arm.com>
18364 David Sherwood <david.sherwood@arm.com>
18366 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18367 than in subroutines. Return 1 for non-integer modes.
18368 (cached_num_sign_bit_copies): Change the type of the mode parameter
18369 to scalar_int_mode.
18370 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18371 classes. Handle CONST_INT_P first and then check whether X also
18372 has a scalar integer mode. Check the same thing for inner registers
18373 of a SUBREG and for values that are being extended or truncated.
18375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18376 Alan Hayward <alan.hayward@arm.com>
18377 David Sherwood <david.sherwood@arm.com>
18379 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18380 in subroutines. Return the mode mask for non-integer modes.
18381 (cached_nonzero_bits): Change the type of the mode parameter
18382 to scalar_int_mode.
18383 (nonzero_bits1): Likewise. Remove early exit for other mode
18384 classes. Handle CONST_INT_P first and then check whether X
18385 also has a scalar integer mode.
18387 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18388 Alan Hayward <alan.hayward@arm.com>
18389 David Sherwood <david.sherwood@arm.com>
18391 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18392 Return a scalar_int_mode and assert that the size is greater than
18394 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18395 (op_by_pieces_d::op_by_pieces_d): Likewise.
18396 (op_by_pieces_d::run): Likewise.
18397 (can_store_by_pieces): Likewise.
18399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18400 Alan Hayward <alan.hayward@arm.com>
18401 David Sherwood <david.sherwood@arm.com>
18403 * combine.c (extract_left_shift): Add a mode argument and update
18405 (make_compound_operation_int): Change the type of the mode parameter
18406 to scalar_int_mode and update the call to extract_left_shift.
18408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18409 Alan Hayward <alan.hayward@arm.com>
18410 David Sherwood <david.sherwood@arm.com>
18412 * combine.c (simplify_and_const_int): Change the type of the mode
18413 parameter to scalar_int_mode.
18414 (simplify_and_const_int_1): Likewise. Update recursive call.
18416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18417 Alan Hayward <alan.hayward@arm.com>
18418 David Sherwood <david.sherwood@arm.com>
18420 * combine.c (simplify_compare_const): Check that the mode is a
18421 scalar_int_mode (rather than VOIDmode) before testing its
18423 (simplify_comparison): Move COMPARISON_P handling out of the
18424 loop and restrict the latter part of the loop to scalar_int_modes.
18425 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18426 and when considering SUBREG_REGs. Use is_int_mode instead of
18427 checking GET_MODE_CLASS against MODE_INT.
18429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18430 Alan Hayward <alan.hayward@arm.com>
18431 David Sherwood <david.sherwood@arm.com>
18433 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18434 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18435 shift_unit_mode and for modes involved in scalar shifts.
18437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18438 Alan Hayward <alan.hayward@arm.com>
18439 David Sherwood <david.sherwood@arm.com>
18441 * combine.c (force_int_to_mode): New function, split out from...
18442 (force_to_mode): ...here. Keep xmode up-to-date and use it
18443 instead of GET_MODE (x).
18445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18446 Alan Hayward <alan.hayward@arm.com>
18447 David Sherwood <david.sherwood@arm.com>
18449 * optabs-query.h (extraction_insn::struct_mode): Change type to
18450 opt_scalar_int_mode and update comment.
18451 (extraction_insn::field_mode): Change type to scalar_int_mode.
18452 (extraction_insn::pos_mode): Likewise.
18453 * combine.c (make_extraction): Update accordingly.
18454 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18455 (get_optab_extraction_insn): Likewise.
18456 * recog.c (simplify_while_replacing): Likewise.
18457 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18458 parameter to opt_scalar_int_mode.
18460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18461 Alan Hayward <alan.hayward@arm.com>
18462 David Sherwood <david.sherwood@arm.com>
18464 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18465 to a scalar_int_mode instead of a machine_mode.
18466 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18467 (get_best_mode): Return a boolean and use a pointer argument to store
18468 the selected mode. Replace the limit mode parameter with a bit limit.
18469 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18470 for the values returned by bit_field_mode_iterator::next_mode.
18471 (store_bit_field): Update call to get_best_mode.
18472 (store_fixed_bit_field): Likewise.
18473 (extract_fixed_bit_field): Likewise.
18474 * expr.c (optimize_bitfield_assignment_op): Likewise.
18475 * fold-const.c (optimize_bit_field_compare): Likewise.
18476 (fold_truth_andor_1): Likewise.
18477 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18478 Update for new type of m_mode.
18479 (get_best_mode): As above.
18481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18482 Alan Hayward <alan.hayward@arm.com>
18483 David Sherwood <david.sherwood@arm.com>
18485 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18486 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18487 (store_bit_field): Check is_a <scalar_int_mode> before calling
18488 strict_volatile_bitfield_p.
18489 (extract_bit_field): Likewise.
18491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18492 Alan Hayward <alan.hayward@arm.com>
18493 David Sherwood <david.sherwood@arm.com>
18495 * target.def (cstore_mode): Return a scalar_int_mode.
18496 * doc/tm.texi: Regenerate.
18497 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18498 * targhooks.h (default_cstore_mode): Likewise.
18499 * targhooks.c (default_cstore_mode): Likewise, using a forced
18501 * expmed.c (emit_cstore): Expect the target of the cstore to be
18504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18505 Alan Hayward <alan.hayward@arm.com>
18506 David Sherwood <david.sherwood@arm.com>
18508 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18510 (niter_desc): Likewise mode.
18511 (iv_analyze): Add a mode parameter.
18513 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18514 and change its type to scalar_int_mode.
18515 * loop-iv.c: Update commentary at head of file.
18516 (iv_constant): Pass the mode paraeter before the rtx it describes
18517 and change its type to scalar_int_mode. Remove VOIDmode handling.
18518 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18519 (iv_extend): Likewise.
18520 (shorten_into_mode): Likewise.
18521 (iv_add): Use scalar_int_mode.
18522 (iv_mult): Likewise.
18523 (iv_shift): Likewise.
18524 (canonicalize_iv_subregs): Likewise.
18525 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18526 it describes and change its mode to scalar_int_mode. Also change
18527 the type of the returned inner_mode to scalar_int_mode.
18528 (get_biv_step): Likewise, turning outer_mode from a pointer
18529 into a direct parameter. Update call to get_biv_step_1.
18530 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18531 iv_constant and get_biv_step.
18532 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18533 and change its type to scalar_int_mode. Don't initialise iv->mode
18534 to VOIDmode and remove later checks for its still being VOIDmode.
18535 Update calls to iv_analyze_op and iv_analyze_expr. Check
18536 is_a <scalar_int_mode> when changing the mode under consideration.
18537 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18538 Update call to iv_analyze_expr.
18539 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18540 inner register is not also a scalar_int_mode. Update call to
18542 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18543 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18544 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18545 separate mode class checks. Update calls to iv_analyze. Remove
18546 fix-up of VOIDmodes after iv_analyze_biv.
18547 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18548 don't have a scalar_int_mode. Update call to biv_p.
18550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18551 Alan Hayward <alan.hayward@arm.com>
18552 David Sherwood <david.sherwood@arm.com>
18554 * cfgexpand.c (convert_debug_memory_address): Use
18555 as_a <scalar_int_mode>.
18556 * combine.c (expand_compound_operation): Likewise.
18557 (make_extraction): Likewise.
18558 (change_zero_ext): Likewise.
18559 (simplify_comparison): Likewise.
18560 * cse.c (cse_insn): Likewise.
18561 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18562 (mem_loc_descriptor): Likewise.
18563 (loc_descriptor): Likewise.
18564 * expmed.c (init_expmed_one_mode): Likewise.
18565 (synth_mult): Likewise.
18566 (emit_store_flag_1): Likewise.
18567 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18568 of a comparison with size.
18569 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18570 (reduce_to_bit_field_precision): Likewise.
18571 * function.c (expand_function_end): Likewise.
18572 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18573 * loop-doloop.c (doloop_modify): Likewise.
18574 * optabs.c (expand_binop): Likewise.
18575 (expand_unop): Likewise.
18576 (expand_copysign_absneg): Likewise.
18577 (prepare_cmp_insn): Likewise.
18578 (maybe_legitimize_operand): Likewise.
18579 * recog.c (const_scalar_int_operand): Likewise.
18580 * rtlanal.c (get_address_mode): Likewise.
18581 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18582 (simplify_cond_clz_ctz): Likewise.
18583 * tree-nested.c (get_nl_goto_field): Likewise.
18584 * tree.c (build_vector_type_for_mode): Likewise.
18585 * var-tracking.c (use_narrower_mode): Likewise.
18587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18588 Alan Hayward <alan.hayward@arm.com>
18589 David Sherwood <david.sherwood@arm.com>
18591 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18592 * builtins.c (expand_builtin_signbit): Use it.
18593 * cfgexpand.c (expand_debug_expr): Likewise.
18594 * dojump.c (do_jump): Likewise.
18595 (do_compare_and_jump): Likewise.
18596 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18597 * expmed.c (make_tree): Likewise.
18598 * expr.c (expand_expr_real_2): Likewise.
18599 (expand_expr_real_1): Likewise.
18600 (try_casesi): Likewise.
18601 * fold-const-call.c (fold_const_call_ss): Likewise.
18602 * fold-const.c (unextend): Likewise.
18603 (extract_muldiv_1): Likewise.
18604 (fold_single_bit_test): Likewise.
18605 (native_encode_int): Likewise.
18606 (native_encode_string): Likewise.
18607 (native_interpret_int): Likewise.
18608 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18609 * internal-fn.c (expand_addsub_overflow): Likewise.
18610 (expand_neg_overflow): Likewise.
18611 (expand_mul_overflow): Likewise.
18612 (expand_arith_overflow): Likewise.
18613 * match.pd: Likewise.
18614 * stor-layout.c (layout_type): Likewise.
18615 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18616 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18617 * tree-ssanames.c (get_range_info): Likewise.
18618 * tree-switch-conversion.c (array_value_type) Likewise.
18619 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18620 (vect_recog_divmod_pattern): Likewise.
18621 (vect_recog_mixed_size_cond_pattern): Likewise.
18622 * tree-vrp.c (extract_range_basic): Likewise.
18623 (simplify_float_conversion_using_ranges): Likewise.
18624 * tree.c (int_fits_type_p): Likewise.
18625 * ubsan.c (instrument_bool_enum_load): Likewise.
18626 * varasm.c (mergeable_string_section): Likewise.
18627 (narrowing_initializer_constant_valid_p): Likewise.
18628 (output_constant): Likewise.
18630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18631 Alan Hayward <alan.hayward@arm.com>
18632 David Sherwood <david.sherwood@arm.com>
18634 * machmode.h (NARROWEST_INT_MODE): New macro.
18635 * expr.c (alignment_for_piecewise_move): Use it instead of
18636 GET_CLASS_NARROWEST_MODE (MODE_INT).
18637 (push_block): Likewise.
18638 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18640 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18643 Alan Hayward <alan.hayward@arm.com>
18644 David Sherwood <david.sherwood@arm.com>
18646 * postreload.c (move2add_valid_value_p): Change the type of the
18647 mode parameter to scalar_int_mode.
18648 (move2add_use_add2_insn): Add a mode parameter and use it instead
18650 (move2add_use_add3_insn): Likewise.
18651 (reload_cse_move2add): Update accordingly.
18653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18654 Alan Hayward <alan.hayward@arm.com>
18655 David Sherwood <david.sherwood@arm.com>
18657 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18659 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18660 * optabs.c (expand_unop): Likewise.
18662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18663 Alan Hayward <alan.hayward@arm.com>
18664 David Sherwood <david.sherwood@arm.com>
18666 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18667 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18668 (popcount_loc_descriptor): Likewise.
18669 (bswap_loc_descriptor): Likewise.
18670 (rotate_loc_descriptor): Likewise.
18671 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18672 calling the functions above.
18674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18675 Alan Hayward <alan.hayward@arm.com>
18676 David Sherwood <david.sherwood@arm.com>
18678 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18680 (try_combine): Likewise.
18681 (simplify_if_then_else): Likewise.
18682 * cse.c (cse_insn): Likewise.
18683 * dwarf2out.c (mem_loc_descriptor): Likewise.
18684 * emit-rtl.c (gen_lowpart_common): Likewise.
18685 * simplify-rtx.c (simplify_truncation): Likewise.
18686 (simplify_binary_operation_1): Likewise.
18687 (simplify_const_relational_operation): Likewise.
18688 (simplify_ternary_operation): Likewise.
18689 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18692 Alan Hayward <alan.hayward@arm.com>
18693 David Sherwood <david.sherwood@arm.com>
18695 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18696 * reload.c (push_reload): Likewise.
18697 (find_reloads): Likewise.
18699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18700 Alan Hayward <alan.hayward@arm.com>
18701 David Sherwood <david.sherwood@arm.com>
18703 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18704 (make_compound_operation_int): Likewise.
18705 (change_zero_ext): Likewise.
18706 * expr.c (convert_move): Likewise.
18707 (convert_modes): Likewise.
18708 * fwprop.c (forward_propagate_subreg): Likewise.
18709 * loop-iv.c (get_biv_step_1): Likewise.
18710 * optabs.c (widen_operand): Likewise.
18711 * postreload.c (move2add_valid_value_p): Likewise.
18712 * recog.c (simplify_while_replacing): Likewise.
18713 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18714 (simplify_binary_operation_1): Likewise. Remove redundant
18715 mode equality check.
18717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18718 Alan Hayward <alan.hayward@arm.com>
18719 David Sherwood <david.sherwood@arm.com>
18721 * combine.c (combine_simplify_rtx): Add checks for
18722 is_a <scalar_int_mode>.
18723 (simplify_if_then_else): Likewise.
18724 (make_field_assignment): Likewise.
18725 (simplify_comparison): Likewise.
18726 * ifcvt.c (noce_try_bitop): Likewise.
18727 * loop-invariant.c (canonicalize_address_mult): Likewise.
18728 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18731 Alan Hayward <alan.hayward@arm.com>
18732 David Sherwood <david.sherwood@arm.com>
18734 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18735 is_a <scalar_int_mode> instead of != BLKmode.
18737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18738 Alan Hayward <alan.hayward@arm.com>
18739 David Sherwood <david.sherwood@arm.com>
18741 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18742 instead of != VOIDmode.
18743 * combine.c (if_then_else_cond): Likewise.
18744 (change_zero_ext): Likewise.
18745 * dwarf2out.c (mem_loc_descriptor): Likewise.
18746 (loc_descriptor): Likewise.
18747 * rtlanal.c (canonicalize_condition): Likewise.
18748 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18751 Alan Hayward <alan.hayward@arm.com>
18752 David Sherwood <david.sherwood@arm.com>
18754 * simplify-rtx.c (simplify_binary_operation_1): Use
18755 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18758 Alan Hayward <alan.hayward@arm.com>
18759 David Sherwood <david.sherwood@arm.com>
18761 * wide-int.h (int_traits<unsigned char>) New class.
18762 (int_traits<unsigned short>) Likewise.
18763 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18764 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18766 * combine.c (set_nonzero_bits_and_sign_copies): Use
18767 is_a <scalar_int_mode>.
18768 (find_split_point): Likewise.
18769 (combine_simplify_rtx): Likewise.
18770 (simplify_logical): Likewise.
18771 (expand_compound_operation): Likewise.
18772 (expand_field_assignment): Likewise.
18773 (make_compound_operation): Likewise.
18774 (extended_count): Likewise.
18775 (change_zero_ext): Likewise.
18776 (simplify_comparison): Likewise.
18777 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18778 (ucompare_loc_descriptor): Likewise.
18779 (minmax_loc_descriptor): Likewise.
18780 (mem_loc_descriptor): Likewise.
18781 (loc_descriptor): Likewise.
18782 * expmed.c (init_expmed_one_mode): Likewise.
18783 * lra-constraints.c (lra_constraint_offset): Likewise.
18784 * optabs.c (prepare_libcall_arg): Likewise.
18785 * postreload.c (move2add_note_store): Likewise.
18786 * reload.c (operands_match_p): Likewise.
18787 * rtl.h (load_extend_op): Likewise.
18788 * rtlhooks.c (gen_lowpart_general): Likewise.
18789 * simplify-rtx.c (simplify_truncation): Likewise.
18790 (simplify_unary_operation_1): Likewise.
18791 (simplify_binary_operation_1): Likewise.
18792 (simplify_const_binary_operation): Likewise.
18793 (simplify_const_relational_operation): Likewise.
18794 (simplify_subreg): Likewise.
18795 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18796 * var-tracking.c (adjust_mems): Likewise.
18797 (prepare_call_arguments): Likewise.
18799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18800 Alan Hayward <alan.hayward@arm.com>
18801 David Sherwood <david.sherwood@arm.com>
18803 * machmode.h (is_int_mode): New fuction.
18804 * combine.c (find_split_point): Use it.
18805 (combine_simplify_rtx): Likewise.
18806 (simplify_if_then_else): Likewise.
18807 (simplify_set): Likewise.
18808 (simplify_shift_const_1): Likewise.
18809 (simplify_comparison): Likewise.
18810 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18811 * cse.c (notreg_cost): Likewise.
18812 (cse_insn): Likewise.
18813 * cselib.c (cselib_lookup_1): Likewise.
18814 * dojump.c (do_jump_1): Likewise.
18815 (do_compare_rtx_and_jump): Likewise.
18816 * dse.c (get_call_args): Likewise.
18817 * dwarf2out.c (rtl_for_decl_init): Likewise.
18818 (native_encode_initializer): Likewise.
18819 * expmed.c (emit_store_flag_1): Likewise.
18820 (emit_store_flag): Likewise.
18821 * expr.c (convert_modes): Likewise.
18822 (store_field): Likewise.
18823 (expand_expr_real_1): Likewise.
18824 * fold-const.c (fold_read_from_constant_string): Likewise.
18825 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18826 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18827 * optabs.c (expand_binop): Likewise.
18828 (expand_unop): Likewise.
18829 (expand_abs_nojump): Likewise.
18830 (expand_one_cmpl_abs_nojump): Likewise.
18831 * simplify-rtx.c (mode_signbit_p): Likewise.
18832 (val_signbit_p): Likewise.
18833 (val_signbit_known_set_p): Likewise.
18834 (val_signbit_known_clear_p): Likewise.
18835 (simplify_relational_operation_1): Likewise.
18836 * tree.c (vector_type_mode): Likewise.
18838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18839 Alan Hayward <alan.hayward@arm.com>
18840 David Sherwood <david.sherwood@arm.com>
18842 * machmode.h (smallest_mode_for_size): Fix formatting.
18843 (smallest_int_mode_for_size): New function.
18844 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18845 instead of smallest_mode_for_size.
18846 * combine.c (make_extraction): Likewise.
18847 * config/arc/arc.c (arc_expand_movmem): Likewise.
18848 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18849 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18850 * config/s390/s390.c (s390_expand_insv): Likewise.
18851 * config/sparc/sparc.c (assign_int_registers): Likewise.
18852 * config/spu/spu.c (spu_function_value): Likewise.
18853 (spu_function_arg): Likewise.
18854 * coverage.c (get_gcov_type): Likewise.
18855 (get_gcov_unsigned_t): Likewise.
18856 * dse.c (find_shift_sequence): Likewise.
18857 * expmed.c (store_bit_field_1): Likewise.
18858 * expr.c (convert_move): Likewise.
18859 (store_field): Likewise.
18860 * internal-fn.c (expand_arith_overflow): Likewise.
18861 * optabs-query.c (get_best_extraction_insn): Likewise.
18862 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18863 * stor-layout.c (layout_type): Likewise.
18864 (initialize_sizetypes): Likewise.
18865 * targhooks.c (default_get_mask_mode): Likewise.
18866 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18869 Alan Hayward <alan.hayward@arm.com>
18870 David Sherwood <david.sherwood@arm.com>
18872 * machmode.h (opt_mode::else_blk): New function.
18873 (int_mode_for_mode): Declare.
18874 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18875 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18877 * cfgexpand.c (expand_debug_expr): Likewise.
18878 * combine.c (gen_lowpart_or_truncate): Likewise.
18879 (gen_lowpart_for_combine): Likewise.
18880 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18881 * config/avr/avr.c (avr_to_int_mode): Likewise.
18882 (avr_out_plus_1): Likewise.
18883 (avr_out_plus): Likewise.
18884 (avr_out_round): Likewise.
18885 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18886 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18887 (s390_expand_vcond): Likewise.
18888 * config/spu/spu.c (spu_split_immediate): Likewise.
18889 (spu_expand_mov): Likewise.
18890 * dse.c (get_stored_val): Likewise.
18891 * expmed.c (store_bit_field_1): Likewise.
18892 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18894 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18895 (extract_low_bits): Likewise.
18896 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18897 handling rather than repeating the check.
18898 (emit_group_store): Likewise.
18899 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18900 * optabs.c (expand_absneg_bit): Likewise.
18901 (expand_copysign_absneg): Likewise.
18902 (expand_copysign_bit): Likewise.
18903 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18904 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18905 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18906 * var-tracking.c (prepare_call_arguments): Likewise.
18907 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18908 int_mode_for_mode instead of mode_for_size.
18909 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18912 Alan Hayward <alan.hayward@arm.com>
18913 David Sherwood <david.sherwood@arm.com>
18915 * machmode.h (int_mode_for_size): New function.
18916 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18917 instead of mode_for_size.
18918 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18920 * combine.c (expand_field_assignment): Use int_mode_for_size
18921 instead of mode_for_size.
18922 (make_extraction): Likewise.
18923 (simplify_shift_const_1): Likewise.
18924 (simplify_comparison): Likewise.
18925 * dojump.c (do_jump): Likewise.
18926 * dwarf2out.c (mem_loc_descriptor): Likewise.
18927 * emit-rtl.c (init_derived_machine_modes): Likewise.
18928 * expmed.c (flip_storage_order): Likewise.
18929 (convert_extracted_bit_field): Likewise.
18930 * expr.c (copy_blkmode_from_reg): Likewise.
18931 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18932 * internal-fn.c (expand_mul_overflow): Likewise.
18933 * lower-subreg.c (simple_move): Likewise.
18934 * optabs-libfuncs.c (init_optabs): Likewise.
18935 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18936 * tree.c (vector_type_mode): Likewise.
18937 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18938 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18939 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18940 * tree-vect-stmts.c (vectorizable_load): Likewise.
18941 (vectorizable_store): Likewise.
18943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18944 Alan Hayward <alan.hayward@arm.com>
18945 David Sherwood <david.sherwood@arm.com>
18947 * coretypes.h (pod_mode): New type.
18948 (scalar_int_mode_pod): New typedef.
18949 * machmode.h (pod_mode): New class.
18950 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18951 * genmodes.c (emit_mode_int_n): Update accordingly.
18952 * lower-subreg.h (target_lower_subreg): Change type to
18953 scalar_int_mode_pod.
18954 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18955 scalar_int_mode_pod.
18957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18958 Alan Hayward <alan.hayward@arm.com>
18959 David Sherwood <david.sherwood@arm.com>
18961 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18962 machine_mode to scalar_int_mode.
18963 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18964 (rs6000_option_override_internal): Remove cast to int.
18965 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18966 machine_mode to scalar_int_mode.
18967 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18968 (rs6000_option_override_internal): Remove cast to int.
18969 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18970 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18972 * config/s390/s390.c (s390_expand_builtin): Likewise.
18973 * coretypes.h (scalar_int_mode): New type.
18974 (opt_scalar_int_mode): New typedef.
18975 * machmode.h (scalar_int_mode): New class.
18976 (scalar_int_mode::includes_p): New function.
18977 (byte_mode): Change type to scalar_int_mode.
18978 (word_mode): Likewise.
18979 (ptr_mode): Likewise.
18980 * emit-rtl.c (byte_mode): Likewise.
18981 (word_mode): Likewise.
18982 (ptr_mode): Likewise.
18983 (init_derived_machine_modes): Update accordingly.
18984 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18985 and MODE_PARTIAL_INT.
18986 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18987 opt_scalar_int_mode.
18989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18990 Alan Hayward <alan.hayward@arm.com>
18991 David Sherwood <david.sherwood@arm.com>
18993 * target.def (libgcc_floating_mode_supported_p): Take a
18995 * doc/tm.texi: Regenerate.
18996 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18998 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18999 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19003 Alan Hayward <alan.hayward@arm.com>
19004 David Sherwood <david.sherwood@arm.com>
19006 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19007 * doc/tm.texi: Regenerate.
19008 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19009 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19010 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19011 * targhooks.h (default_floatn_mode): Likewise.
19012 * targhooks.c (default_floatn_mode): Likewise.
19013 * tree.c (build_common_tree_nodes): Update accordingly.
19015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19016 Alan Hayward <alan.hayward@arm.com>
19017 David Sherwood <david.sherwood@arm.com>
19019 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19020 (mode_iterator::iterate_p): Likewise.
19021 (mode_iterator::get_wider): Likewise.
19022 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19025 Alan Hayward <alan.hayward@arm.com>
19026 David Sherwood <david.sherwood@arm.com>
19028 * coretypes.h (opt_scalar_float_mode): New typedef.
19029 * machmode.h (float_mode_for_size): New function.
19030 * emit-rtl.c (double_mode): Delete.
19031 (init_emit_once): Use float_mode_for_size.
19032 * stor-layout.c (layout_type): Likewise.
19033 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19036 Alan Hayward <alan.hayward@arm.com>
19037 David Sherwood <david.sherwood@arm.com>
19039 * output.h (assemble_real): Take a scalar_float_mode.
19040 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19041 * config/arm/arm.md (consttable_4): Likewise.
19042 (consttable_8): Likewise.
19043 (consttable_16): Likewise.
19044 * config/mips/mips.md (consttable_float): Likewise.
19045 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19046 * varasm.c (assemble_real): Take a scalar_float_mode.
19047 (output_constant_pool_2): Update accordingly.
19048 (output_constant): Likewise.
19050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19051 Alan Hayward <alan.hayward@arm.com>
19052 David Sherwood <david.sherwood@arm.com>
19054 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19055 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19056 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19057 (native_encode_real): Likewise.
19058 (native_interpret_real): Likewise.
19059 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19060 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19063 Alan Hayward <alan.hayward@arm.com>
19064 David Sherwood <david.sherwood@arm.com>
19066 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19067 <scalar_float_mode>. Simplify.
19068 (gen_extend_conv_libfunc): Likewise.
19070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19071 Alan Hayward <alan.hayward@arm.com>
19072 David Sherwood <david.sherwood@arm.com>
19074 * coretypes.h (scalar_float_mode): New type.
19075 * machmode.h (mode_traits::from_int): Use machine_mode if
19076 USE_ENUM_MODES is defined.
19077 (is_a): New function.
19079 (dyn_cast): Likewise.
19080 (scalar_float_mode): New class.
19081 (scalar_float_mode::includes_p): New function.
19082 (is_float_mode): Likewise.
19083 * gdbhooks.py (MachineModePrinter): New class.
19084 (build_pretty_printer): Use it for scalar_float_mode.
19085 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19086 (format_helper::format_helper): Turn into a template.
19087 * genmodes.c (get_mode_class): New function.
19088 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19089 or machine_mode if none.
19090 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19091 as_a <scalar_float_mode>.
19092 * dwarf2out.c (mem_loc_descriptor): Likewise.
19093 (insert_float): Likewise.
19094 (add_const_value_attribute): Likewise.
19095 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19096 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19097 (expand_unop): Update accordingly.
19098 (expand_abs_nojump): Likewise.
19099 (expand_copysign_absneg): Take a scalar_float_mode.
19100 (expand_copysign_bit): Likewise.
19101 (expand_copysign): Update accordingly.
19103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19104 Alan Hayward <alan.hayward@arm.com>
19105 David Sherwood <david.sherwood@arm.com>
19107 * coretypes.h (opt_mode): New class.
19108 * machmode.h (opt_mode): Likewise.
19109 (opt_mode::else_void): New function.
19110 (opt_mode::require): Likewise.
19111 (opt_mode::exists): Likewise.
19112 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19113 (GET_MODE_2XWIDER_MODE): Likewise.
19114 (mode_iterator::get_wider): Update accordingly.
19115 (mode_iterator::get_2xwider): Likewise.
19116 (mode_iterator::get_known_wider): Likewise, turning into a template.
19117 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19118 forcing a wider mode to exist.
19119 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19120 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19121 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19122 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19123 * lower-subreg.c (init_lower_subreg): Likewise.
19124 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19125 on the final iteration.
19126 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19127 a wider mode exists before asking for a move pattern.
19128 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19129 forcing a wider mode to exist.
19130 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19131 returning false if no such mode exists.
19132 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19133 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19134 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19135 Avoid checking for a MODE_INT if we already know the mode is not a
19137 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19138 forcing a wider mode to exist.
19139 (expmed_mult_highpart_optab): Likewise.
19140 (expmed_mult_highpart): Likewise.
19141 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19143 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19144 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19145 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19146 * internal-fn.c (expand_mul_overflow): Update use of
19147 GET_MODE_2XWIDER_MODE.
19148 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19149 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19150 GET_MODE_WIDER_MODE.
19151 (convert_plusminus_to_widen): Likewise.
19152 * tree-switch-conversion.c (array_value_type): Likewise.
19153 * var-tracking.c (emit_note_insn_var_location): Likewise.
19154 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19155 Return false inside rather than outside the loop if no wider mode
19157 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19158 and GET_MODE_2XWIDER_MODE
19159 (can_compare_p): Use else_void.
19160 * gdbhooks.py (OptMachineModePrinter): New class.
19161 (build_pretty_printer): Use it for opt_mode.
19163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19164 Alan Hayward <alan.hayward@arm.com>
19165 David Sherwood <david.sherwood@arm.com>
19167 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19168 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19171 Alan Hayward <alan.hayward@arm.com>
19172 David Sherwood <david.sherwood@arm.com>
19174 * machmode.h (mode_traits): New structure.
19175 (get_narrowest_mode): New function.
19176 (mode_iterator::start): Likewise.
19177 (mode_iterator::iterate_p): Likewise.
19178 (mode_iterator::get_wider): Likewise.
19179 (mode_iterator::get_known_wider): Likewise.
19180 (mode_iterator::get_2xwider): Likewise.
19181 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19182 (FOR_EACH_MODE): Likewise.
19183 (FOR_EACH_MODE_FROM): Likewise.
19184 (FOR_EACH_MODE_UNTIL): Likewise.
19185 (FOR_EACH_WIDER_MODE): Likewise.
19186 (FOR_EACH_2XWIDER_MODE): Likewise.
19187 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19188 * combine.c (simplify_comparison): Likewise
19189 * config/i386/i386.c (type_natural_mode): Likewise.
19190 * cse.c (cse_insn): Likewise.
19191 * dse.c (find_shift_sequence): Likewise.
19192 * emit-rtl.c (init_derived_machine_modes): Likewise.
19193 (init_emit_once): Likewise.
19194 * explow.c (hard_function_value): Likewise.
19195 * expmed.c (extract_fixed_bit_field_1): Likewise.
19196 (extract_bit_field_1): Likewise.
19197 (expand_divmod): Likewise.
19198 (emit_store_flag_1): Likewise.
19199 * expr.c (init_expr_target): Likewise.
19200 (convert_move): Likewise.
19201 (alignment_for_piecewise_move): Likewise.
19202 (widest_int_mode_for_size): Likewise.
19203 (emit_block_move_via_movmem): Likewise.
19204 (copy_blkmode_to_reg): Likewise.
19205 (set_storage_via_setmem): Likewise.
19206 (compress_float_constant): Likewise.
19207 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19208 * optabs-query.c (get_best_extraction_insn): Likewise.
19209 * optabs.c (expand_binop): Likewise.
19210 (expand_twoval_unop): Likewise.
19211 (expand_twoval_binop): Likewise.
19212 (widen_leading): Likewise.
19213 (widen_bswap): Likewise.
19214 (expand_parity): Likewise.
19215 (expand_unop): Likewise.
19216 (prepare_cmp_insn): Likewise.
19217 (prepare_float_lib_cmp): Likewise.
19218 (expand_float): Likewise.
19219 (expand_fix): Likewise.
19220 (expand_sfix_optab): Likewise.
19221 * postreload.c (move2add_use_add2_insn): Likewise.
19222 * reg-stack.c (reg_to_stack): Likewise.
19223 * reginfo.c (choose_hard_reg_mode): Likewise.
19224 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19225 * stor-layout.c (mode_for_size): Likewise.
19226 (smallest_mode_for_size): Likewise.
19227 (mode_for_vector): Likewise.
19228 (finish_bitfield_representative): Likewise.
19229 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19230 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19231 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19232 * var-tracking.c (prepare_call_arguments): Likewise.
19234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19235 Alan Hayward <alan.hayward@arm.com>
19236 David Sherwood <david.sherwood@arm.com>
19238 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19239 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19240 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19241 * machmode.h (mode_size): Move earlier in file.
19242 (mode_precision): Likewise.
19243 (mode_inner): Likewise.
19244 (mode_nunits): Likewise.
19245 (mode_unit_size): Likewise.
19246 (unit_unit_precision): Likewise.
19247 (mode_wider): Likewise.
19248 (mode_2xwider): Likewise.
19249 (machine_mode): New class.
19250 (mode_to_bytes): New function.
19251 (mode_to_bits): Likewise.
19252 (mode_to_precision): Likewise.
19253 (mode_to_inner): Likewise.
19254 (mode_to_unit_size): Likewise.
19255 (mode_to_unit_precision): Likewise.
19256 (mode_to_nunits): Likewise.
19257 (GET_MODE_SIZE): Use mode_to_bytes.
19258 (GET_MODE_BITSIZE): Use mode_to_bits.
19259 (GET_MODE_PRECISION): Use mode_to_precision.
19260 (GET_MODE_INNER): Use mode_to_inner.
19261 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19262 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19263 (GET_MODE_NUNITS): Use mode_to_nunits.
19264 * system.h (ALWAYS_INLINE): New macro.
19265 * config/powerpcspe/powerpcspe-c.c
19266 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19267 int for arg1_mode and arg2_mode.
19269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19270 Alan Hayward <alan.hayward@arm.com>
19271 David Sherwood <david.sherwood@arm.com>
19273 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19274 Prefix mode names with E_ in case statements.
19275 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19276 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19277 (aarch64_split_simd_move): Likewise.
19278 (aarch64_gen_storewb_pair): Likewise.
19279 (aarch64_gen_loadwb_pair): Likewise.
19280 (aarch64_gen_store_pair): Likewise.
19281 (aarch64_gen_load_pair): Likewise.
19282 (aarch64_get_condition_code_1): Likewise.
19283 (aarch64_constant_pool_reload_icode): Likewise.
19284 (get_rsqrte_type): Likewise.
19285 (get_rsqrts_type): Likewise.
19286 (get_recpe_type): Likewise.
19287 (get_recps_type): Likewise.
19288 (aarch64_gimplify_va_arg_expr): Likewise.
19289 (aarch64_simd_container_mode): Likewise.
19290 (aarch64_emit_load_exclusive): Likewise.
19291 (aarch64_emit_store_exclusive): Likewise.
19292 (aarch64_expand_compare_and_swap): Likewise.
19293 (aarch64_gen_atomic_cas): Likewise.
19294 (aarch64_emit_bic): Likewise.
19295 (aarch64_emit_atomic_swap): Likewise.
19296 (aarch64_emit_atomic_load_op): Likewise.
19297 (aarch64_evpc_trn): Likewise.
19298 (aarch64_evpc_uzp): Likewise.
19299 (aarch64_evpc_zip): Likewise.
19300 (aarch64_evpc_ext): Likewise.
19301 (aarch64_evpc_rev): Likewise.
19302 (aarch64_evpc_dup): Likewise.
19303 (aarch64_gen_ccmp_first): Likewise.
19304 (aarch64_gen_ccmp_next): Likewise.
19305 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19306 (alpha_emit_xfloating_libcall): Likewise.
19307 (emit_insxl): Likewise.
19308 (alpha_arg_type): Likewise.
19309 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19310 (arc_preferred_simd_mode): Likewise.
19311 (arc_secondary_reload): Likewise.
19312 (get_arc_condition_code): Likewise.
19313 (arc_print_operand): Likewise.
19314 (arc_legitimate_constant_p): Likewise.
19315 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19316 * config/arc/arc.md (casesi_load): Likewise.
19317 (casesi_compact_jump): Likewise.
19318 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19319 (cc_use_register): Likewise.
19320 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19321 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19322 (arm_init_iwmmxt_builtins): Likewise.
19323 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19324 (neon_expand_vector_init): Likewise.
19325 (arm_attr_length_move_neon): Likewise.
19326 (maybe_get_arm_condition_code): Likewise.
19327 (arm_emit_vector_const): Likewise.
19328 (arm_preferred_simd_mode): Likewise.
19329 (arm_output_iwmmxt_tinsr): Likewise.
19330 (thumb1_output_casesi): Likewise.
19331 (thumb2_output_casesi): Likewise.
19332 (arm_emit_load_exclusive): Likewise.
19333 (arm_emit_store_exclusive): Likewise.
19334 (arm_expand_compare_and_swap): Likewise.
19335 (arm_evpc_neon_vuzp): Likewise.
19336 (arm_evpc_neon_vzip): Likewise.
19337 (arm_evpc_neon_vrev): Likewise.
19338 (arm_evpc_neon_vtrn): Likewise.
19339 (arm_evpc_neon_vext): Likewise.
19340 (arm_validize_comparison): Likewise.
19341 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19342 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19343 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19344 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19345 (c6x_preferred_simd_mode): Likewise.
19346 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19347 (epiphany_rtx_costs): Likewise.
19348 * config/epiphany/predicates.md (proper_comparison_operator):
19350 * config/frv/frv.c (condexec_memory_operand): Likewise.
19351 (frv_emit_move): Likewise.
19352 (output_move_single): Likewise.
19353 (output_condmove_single): Likewise.
19354 (frv_hard_regno_mode_ok): Likewise.
19355 (frv_matching_accg_mode): Likewise.
19356 * config/h8300/h8300.c (split_adds_subs): Likewise.
19357 (h8300_rtx_costs): Likewise.
19358 (h8300_print_operand): Likewise.
19359 (compute_mov_length): Likewise.
19360 (output_logical_op): Likewise.
19361 (compute_logical_op_length): Likewise.
19362 (compute_logical_op_cc): Likewise.
19363 (h8300_shift_needs_scratch_p): Likewise.
19364 (output_a_shift): Likewise.
19365 (compute_a_shift_length): Likewise.
19366 (compute_a_shift_cc): Likewise.
19367 (expand_a_rotate): Likewise.
19368 (output_a_rotate): Likewise.
19369 * config/i386/i386.c (classify_argument): Likewise.
19370 (function_arg_advance_32): Likewise.
19371 (function_arg_32): Likewise.
19372 (function_arg_64): Likewise.
19373 (function_value_64): Likewise.
19374 (ix86_gimplify_va_arg): Likewise.
19375 (ix86_legitimate_constant_p): Likewise.
19376 (put_condition_code): Likewise.
19377 (split_double_mode): Likewise.
19378 (ix86_avx256_split_vector_move_misalign): Likewise.
19379 (ix86_expand_vector_logical_operator): Likewise.
19380 (ix86_split_idivmod): Likewise.
19381 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19382 (ix86_build_const_vector): Likewise.
19383 (ix86_build_signbit_mask): Likewise.
19384 (ix86_match_ccmode): Likewise.
19385 (ix86_cc_modes_compatible): Likewise.
19386 (ix86_expand_branch): Likewise.
19387 (ix86_expand_sse_cmp): Likewise.
19388 (ix86_expand_sse_movcc): Likewise.
19389 (ix86_expand_int_sse_cmp): Likewise.
19390 (ix86_expand_vec_perm_vpermi2): Likewise.
19391 (ix86_expand_vec_perm): Likewise.
19392 (ix86_expand_sse_unpack): Likewise.
19393 (ix86_expand_int_addcc): Likewise.
19394 (ix86_split_to_parts): Likewise.
19395 (ix86_vectorize_builtin_gather): Likewise.
19396 (ix86_vectorize_builtin_scatter): Likewise.
19397 (avx_vpermilp_parallel): Likewise.
19398 (inline_memory_move_cost): Likewise.
19399 (ix86_tieable_integer_mode_p): Likewise.
19400 (x86_maybe_negate_const_int): Likewise.
19401 (ix86_expand_vector_init_duplicate): Likewise.
19402 (ix86_expand_vector_init_one_nonzero): Likewise.
19403 (ix86_expand_vector_init_one_var): Likewise.
19404 (ix86_expand_vector_init_concat): Likewise.
19405 (ix86_expand_vector_init_interleave): Likewise.
19406 (ix86_expand_vector_init_general): Likewise.
19407 (ix86_expand_vector_set): Likewise.
19408 (ix86_expand_vector_extract): Likewise.
19409 (emit_reduc_half): Likewise.
19410 (ix86_emit_i387_round): Likewise.
19411 (ix86_mangle_type): Likewise.
19412 (ix86_expand_round_sse4): Likewise.
19413 (expand_vec_perm_blend): Likewise.
19414 (canonicalize_vector_int_perm): Likewise.
19415 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19416 (expand_vec_perm_1): Likewise.
19417 (expand_vec_perm_interleave3): Likewise.
19418 (expand_vec_perm_even_odd_pack): Likewise.
19419 (expand_vec_perm_even_odd_1): Likewise.
19420 (expand_vec_perm_broadcast_1): Likewise.
19421 (ix86_vectorize_vec_perm_const_ok): Likewise.
19422 (ix86_expand_vecop_qihi): Likewise.
19423 (ix86_expand_mul_widen_hilo): Likewise.
19424 (ix86_expand_sse2_abs): Likewise.
19425 (ix86_expand_pextr): Likewise.
19426 (ix86_expand_pinsr): Likewise.
19427 (ix86_preferred_simd_mode): Likewise.
19428 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19429 * config/i386/sse.md (*andnot<mode>3): Likewise.
19430 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19431 (*<code><mode>3): Likewise.
19432 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19433 (ia64_expand_atomic_op): Likewise.
19434 (ia64_arg_type): Likewise.
19435 (ia64_mode_to_int): Likewise.
19436 (ia64_scalar_mode_supported_p): Likewise.
19437 (ia64_vector_mode_supported_p): Likewise.
19438 (expand_vec_perm_broadcast): Likewise.
19439 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19440 (iq2000_function_arg_advance): Likewise.
19441 (iq2000_function_arg): Likewise.
19442 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19443 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19444 (m68k_libcall_value): Likewise.
19445 (m68k_function_value): Likewise.
19446 (sched_attr_op_type): Likewise.
19447 * config/mcore/mcore.c (mcore_output_move): Likewise.
19448 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19450 (microblaze_function_arg): Likewise.
19451 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19452 (mips_print_operand): Likewise.
19453 (mips_mode_ok_for_mov_fmt_p): Likewise.
19454 (mips_vector_mode_supported_p): Likewise.
19455 (mips_preferred_simd_mode): Likewise.
19456 (mips_expand_vpc_loongson_even_odd): Likewise.
19457 (mips_expand_vec_unpack): Likewise.
19458 (mips_expand_vi_broadcast): Likewise.
19459 (mips_expand_vector_init): Likewise.
19460 (mips_expand_vec_reduc): Likewise.
19461 (mips_expand_msa_cmp): Likewise.
19462 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19463 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19464 (cc_flags_for_mode): Likewise.
19465 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19466 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19467 (nds32_output_casesi_pc_relative): Likewise.
19468 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19469 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19470 (nvptx_gen_unpack): Likewise.
19471 (nvptx_gen_pack): Likewise.
19472 (nvptx_gen_shuffle): Likewise.
19473 (nvptx_gen_wcast): Likewise.
19474 (nvptx_preferred_simd_mode): Likewise.
19475 * config/pa/pa.c (pa_secondary_reload): Likewise.
19476 * config/pa/predicates.md (base14_operand): Likewise.
19477 * config/powerpcspe/powerpcspe-c.c
19478 (altivec_resolve_overloaded_builtin): Likewise.
19479 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19481 (rs6000_preferred_simd_mode): Likewise.
19482 (output_vec_const_move): Likewise.
19483 (rs6000_expand_vector_extract): Likewise.
19484 (rs6000_split_vec_extract_var): Likewise.
19485 (reg_offset_addressing_ok_p): Likewise.
19486 (rs6000_legitimate_offset_address_p): Likewise.
19487 (rs6000_legitimize_address): Likewise.
19488 (rs6000_emit_set_const): Likewise.
19489 (rs6000_const_vec): Likewise.
19490 (rs6000_emit_move): Likewise.
19491 (spe_build_register_parallel): Likewise.
19492 (rs6000_darwin64_record_arg_recurse): Likewise.
19493 (swap_selector_for_mode): Likewise.
19494 (spe_init_builtins): Likewise.
19495 (paired_init_builtins): Likewise.
19496 (altivec_init_builtins): Likewise.
19497 (do_load_for_compare): Likewise.
19498 (rs6000_generate_compare): Likewise.
19499 (rs6000_expand_float128_convert): Likewise.
19500 (emit_load_locked): Likewise.
19501 (emit_store_conditional): Likewise.
19502 (rs6000_output_function_epilogue): Likewise.
19503 (rs6000_handle_altivec_attribute): Likewise.
19504 (rs6000_function_value): Likewise.
19505 (emit_fusion_gpr_load): Likewise.
19506 (emit_fusion_p9_load): Likewise.
19507 (emit_fusion_p9_store): Likewise.
19508 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19509 (fusion_gpr_mem_load): Likewise.
19510 (fusion_addis_mem_combo_load): Likewise.
19511 (fusion_addis_mem_combo_store): Likewise.
19512 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19513 (fusion_gpr_mem_load): Likewise.
19514 (fusion_addis_mem_combo_load): Likewise.
19515 (fusion_addis_mem_combo_store): Likewise.
19516 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19518 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19519 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19520 (rs6000_preferred_simd_mode): Likewise.
19521 (output_vec_const_move): Likewise.
19522 (rs6000_expand_vector_extract): Likewise.
19523 (rs6000_split_vec_extract_var): Likewise.
19524 (reg_offset_addressing_ok_p): Likewise.
19525 (rs6000_legitimate_offset_address_p): Likewise.
19526 (rs6000_legitimize_address): Likewise.
19527 (rs6000_emit_set_const): Likewise.
19528 (rs6000_const_vec): Likewise.
19529 (rs6000_emit_move): Likewise.
19530 (rs6000_darwin64_record_arg_recurse): Likewise.
19531 (swap_selector_for_mode): Likewise.
19532 (paired_init_builtins): Likewise.
19533 (altivec_init_builtins): Likewise.
19534 (rs6000_expand_float128_convert): Likewise.
19535 (emit_load_locked): Likewise.
19536 (emit_store_conditional): Likewise.
19537 (rs6000_output_function_epilogue): Likewise.
19538 (rs6000_handle_altivec_attribute): Likewise.
19539 (rs6000_function_value): Likewise.
19540 (emit_fusion_gpr_load): Likewise.
19541 (emit_fusion_p9_load): Likewise.
19542 (emit_fusion_p9_store): Likewise.
19543 * config/rx/rx.c (rx_gen_move_template): Likewise.
19544 (flags_from_mode): Likewise.
19545 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19546 (s390_slb_comparison): Likewise.
19547 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19548 (s390_vector_mode_supported_p): Likewise.
19549 (s390_cc_modes_compatible): Likewise.
19550 (s390_match_ccmode_set): Likewise.
19551 (s390_canonicalize_comparison): Likewise.
19552 (s390_emit_compare_and_swap): Likewise.
19553 (s390_branch_condition_mask): Likewise.
19554 (s390_rtx_costs): Likewise.
19555 (s390_secondary_reload): Likewise.
19556 (__SECONDARY_RELOAD_CASE): Likewise.
19557 (s390_expand_cs): Likewise.
19558 (s390_preferred_simd_mode): Likewise.
19559 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19560 * config/sh/sh.c (sh_print_operand): Likewise.
19561 (dump_table): Likewise.
19562 (sh_secondary_reload): Likewise.
19563 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19564 * config/sh/sh.md (casesi_worker_1): Likewise.
19565 (casesi_worker_2): Likewise.
19566 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19567 (fcc_comparison_operator): Likewise.
19568 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19569 (emit_soft_tfmode_cvt): Likewise.
19570 (sparc_preferred_simd_mode): Likewise.
19571 (output_cbranch): Likewise.
19572 (sparc_print_operand): Likewise.
19573 (sparc_expand_vec_perm_bmask): Likewise.
19574 (vector_init_bshuffle): Likewise.
19575 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19576 (spu_vector_mode_supported_p): Likewise.
19577 (spu_expand_insv): Likewise.
19578 (spu_emit_branch_or_set): Likewise.
19579 (spu_handle_vector_attribute): Likewise.
19580 (spu_builtin_splats): Likewise.
19581 (spu_builtin_extract): Likewise.
19582 (spu_builtin_promote): Likewise.
19583 (spu_expand_sign_extend): Likewise.
19584 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19585 (tilegx_simd_int): Likewise.
19586 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19587 (tilepro_simd_int): Likewise.
19588 * config/v850/v850.c (const_double_split): Likewise.
19589 (v850_print_operand): Likewise.
19590 (ep_memory_offset): Likewise.
19591 * config/vax/vax.c (vax_rtx_costs): Likewise.
19592 (vax_output_int_move): Likewise.
19593 (vax_output_int_add): Likewise.
19594 (vax_output_int_subtract): Likewise.
19595 * config/visium/predicates.md (visium_branch_operator): Likewise.
19596 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19597 (visium_print_operand_address): Likewise.
19598 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19599 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19600 (xtensa_expand_conditional_branch): Likewise.
19601 (xtensa_copy_incoming_a7): Likewise.
19602 (xtensa_output_literal): Likewise.
19603 * dfp.c (decimal_real_maxval): Likewise.
19604 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19607 Alan Hayward <alan.hayward@arm.com>
19608 David Sherwood <david.sherwood@arm.com>
19610 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19611 (mode_nunits_inline): Likewise.
19612 (mode_inner_inline): Likewise.
19613 (mode_unit_size_inline): Likewise.
19614 (mode_unit_precision_inline): Likewise.
19615 (emit_insn_modes_h): Likewise. Also emit a #define of the
19617 (emit_mode_wider): Add an E_ prefix to mode names.
19618 (emit_mode_complex): Likewise.
19619 (emit_mode_inner): Likewise.
19620 (emit_mode_adjustments): Likewise.
19621 (emit_mode_int_n): Likewise.
19622 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19623 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19624 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19625 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19626 (CRC32_BUILTIN, ENTRY): Likewise.
19627 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19628 (aarch64_pop_regs): Likewise.
19629 (aarch64_process_components): Likewise.
19630 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19631 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19632 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19633 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19634 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19635 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19637 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19638 (pcmpistr): Likewise.
19639 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19640 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19641 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19643 * config/rl78/rl78.c (mduc_regs): Likewise.
19644 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19645 (htm_expand_builtin): Likewise.
19646 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19647 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19648 * config/xtensa/xtensa.c (print_operand): Likewise.
19649 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19650 (NUM_MODE_VECTOR_INT): Likewise.
19651 * genoutput.c (null_operand): Likewise.
19652 (output_operand_data): Likewise.
19653 * genrecog.c (print_parameter_value): Likewise.
19654 * lra.c (debug_operand_data): Likewise.
19656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19658 * dwarf2out.c (scompare_loc_descriptor_wide)
19659 (scompare_loc_descriptor_narrow): New functions, split out from...
19660 (scompare_loc_descriptor): ...here.
19661 * expmed.c (emit_store_flag_int): New function, split out from...
19662 (emit_store_flag): ...here.
19664 2017-08-30 Richard Biener <rguenther@suse.de>
19666 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19667 (dwarf2out_early_finish): Move setting of AT_pubnames from
19668 early debug output to early finish.
19670 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19672 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19673 and -mdata-region to the assembler.
19675 2017-08-30 Richard Biener <rguenther@suse.de>
19677 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19679 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19680 (dwarf2out_early_global_decl): Only generate a DIE for the
19681 abstract origin if it doesn't already exist or is a declaration DIE.
19682 (resolve_addr): Do not add the linkage name twice when
19683 generating a stub DIE for the DW_TAG_GNU_call_site target.
19685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19687 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19688 Use machine_mode rather than int for arg1_mode.
19690 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19693 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19694 that the second argument of the built-in functions to unpack
19695 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19696 switch statement instead a lot of if statements.
19697 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19698 Allow 64-bit values to be in Altivec registers as well as
19699 traditional floating point registers.
19700 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19702 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19704 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19705 MAX_REGS_PER_ADDRESS == 1.
19707 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19709 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19710 * config/i386/i386.c (ix86_option_override_internal): Simplify
19711 setting of opts->x_flag_entry.
19713 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19714 Jakub Jelinek <jakub@redhat.com>
19715 Richard Biener <rguenther@suse.de>
19717 PR tree-optimization/81503
19718 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19719 folded constant fits in the target type; reorder tests for clarity.
19721 2017-08-29 Martin Liska <mliska@suse.cz>
19723 * passes.def: Include pass_lower_switch.
19724 * stmt.c (dump_case_nodes): Remove and move to
19725 tree-switch-conversion.
19726 (case_values_threshold): Likewise.
19727 (expand_switch_as_decision_tree_p): Likewise.
19728 (emit_case_decision_tree): Likewise.
19729 (expand_case): Likewise.
19730 (balance_case_nodes): Likewise.
19731 (node_has_low_bound): Likewise.
19732 (node_has_high_bound): Likewise.
19733 (node_is_bounded): Likewise.
19734 (emit_case_nodes): Likewise.
19735 (struct simple_case_node): New struct.
19736 (add_case_node): Remove.
19737 (emit_case_dispatch_table): Use vector instead of case_list.
19738 (reset_out_edges_aux): Remove.
19739 (compute_cases_per_edge): Likewise.
19740 (expand_case): Build list of simple_case_node.
19741 (expand_sjlj_dispatch_table): Use it.
19742 * tree-switch-conversion.c (struct case_node): Moved from
19743 stmt.c and adjusted.
19744 (emit_case_nodes): Likewise.
19745 (node_has_low_bound): Likewise.
19746 (node_has_high_bound): Likewise.
19747 (node_is_bounded): Likewise.
19748 (case_values_threshold): Likewise.
19749 (reset_out_edges_aux): Likewise.
19750 (compute_cases_per_edge): Likewise.
19751 (add_case_node): Likewise.
19752 (dump_case_nodes): Likewise.
19753 (balance_case_nodes): Likewise.
19754 (expand_switch_as_decision_tree_p): Likewise.
19755 (emit_jump): Likewise.
19756 (emit_case_decision_tree): Likewise.
19757 (try_switch_expansion): Likewise.
19758 (do_jump_if_equal): Likewise.
19759 (emit_cmp_and_jump_insns): Likewise.
19760 (fix_phi_operands_for_edge): New function.
19761 (record_phi_operand_mapping): Likewise.
19762 (class pass_lower_switch): New pass.
19763 (pass_lower_switch::execute): New function.
19764 (make_pass_lower_switch): Likewise.
19765 (conditional_probability):
19766 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19767 * tree-pass.h: Add make_pass_lower_switch.
19769 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19772 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19775 2017-08-29 Richard Biener <rguenther@suse.de>
19777 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19778 we do not add a DW_AT_inline attribute twice.
19779 (gen_subprogram_die): Remove code setting DW_AT_inline on
19780 DECL_ABSTRACT_P nodes.
19782 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19784 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19785 calls to internal functions.
19786 (gimplify_modify_expr): Likewise.
19787 * tree-call-cdce.c (use_internal_fn): Likewise.
19788 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19789 (convert_to_divmod): Set the nothrow flag.
19790 * tree-if-conv.c (predicate_mem_writes): Likewise.
19791 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19792 (vectorizable_call): Likewise.
19793 (vectorizable_store): Likewise.
19794 (vectorizable_load): Likewise.
19795 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19796 (vect_recog_mask_conversion_pattern): Likewise.
19798 2017-08-29 Martin Liska <mliska@suse.cz>
19801 * gcc.c (driver_handle_option): Add new argument.
19802 * opts-common.c (handle_option): Pass
19803 target_option_override_hook.
19804 * opts-global.c (lang_handle_option): Add new option.
19805 (set_default_handlers): Add new argument.
19806 (decode_options): Likewise.
19807 * opts.c (target_handle_option): Likewise.
19808 (common_handle_option): Call target_option_override_hook.
19809 * opts.h (struct cl_option_handler_func): Add hook for
19810 target option override.
19811 (struct cl_option_handlers): Likewise.
19812 (set_default_handlers): Add new argument.
19813 (decode_options): Likewise.
19814 (common_handle_option): Likewise.
19815 (target_handle_option): Likewise.
19816 * toplev.c (toplev::main): Pass targetm.target_option.override
19819 2017-08-29 Richard Biener <rguenther@suse.de>
19820 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19822 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19823 life to the active subtree.
19825 2017-08-28 Jeff Law <law@redhat.com>
19827 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19828 derive_equivalences.
19829 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19830 Code moved into....
19831 (edge_info::derive_equivalences): New private member function
19833 * tree-ssa-dom.c (class edge_info): Changed from a struct
19834 to a class. Add ctor/dtor, methods and data members.
19835 (edge_info::edge_info): Renamed from allocate_edge_info.
19836 Initialize additional members.
19837 (edge_info::~edge_info): New.
19838 (free_dom_edge_info): Delete the edge info.
19839 (record_edge_info): Use new class & associated member functions.
19840 Tighten forms for testing for edge equivalences.
19841 (record_temporary_equivalences): Iterate over the simple
19842 equivalences rather than assuming there's only one per edge.
19843 (cprop_into_successor_phis): Iterate over the simple
19844 equivalences rather than assuming there's only one per edge.
19845 (optimize_stmt): Use operand_equal_p rather than pointer
19846 equality for mini-DSE code.
19848 2017-08-28 Nathan Sidwell <nathan@acm.org>
19850 * gcc.c (execute): Fold SIGPIPE handling into switch
19851 statement. Adjust internal error message.
19853 2017-08-28 Richard Biener <rguenther@suse.de>
19856 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19857 Do nothing for removed DIEs.
19859 2017-08-28 Richard Biener <rguenther@suse.de>
19861 PR tree-optimization/81977
19862 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19865 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19868 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19869 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19870 using targetm.gen_mem_thread_fence.
19872 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19875 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19876 predicate to register_operand. Reorder operands.
19877 (*btr<mode>): Ditto.
19878 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19879 (*btr<mode>_mask): Ditto.
19881 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19883 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19884 * config/rs6000/xmmintrin.h: New file.
19885 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19887 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19890 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19891 parameter and_insn and return it.
19892 (recombine_lvx_pattern): Insert a copy to ensure availability of
19893 the base register of the copied masking operation at the point of
19894 the instruction replacement.
19895 (recombine_stvx_pattern): Likewise.
19897 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19899 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19900 undocumented switches.
19901 (-mpower9-dform-vector): Likewise.
19902 (-mpower9-dform): Likewise.
19903 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19904 comments to delete references to -mpower9-dform* switches.
19905 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19906 Delete reference to -mpower9-dform* switches, test for
19907 -mpower9-vector instead.
19908 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19909 (OTHER_P9_VECTOR_MASKS): Likewise.
19910 (POWERPC_MASKS): Likewise.
19911 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19912 tests against -mpower9-dform* to -mpower9-vector. Delete code
19913 that checked for -mpower9-dform* consistancy with other options.
19914 Add test for -mpower9-misc to enable other power9 switches.
19915 (rs6000_init_hard_regno_mode_ok): Likewise.
19916 (rs6000_option_override_internal): Likewise.
19917 (rs6000_emit_prologue): Likewise.
19918 (rs6000_emit_epilogue): Likewise.
19919 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19920 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19921 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19923 (emit_fusion_p9_store): Likewise.
19924 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19925 resetting these macros if the assembler does not support ISA 3.0
19927 (TARGET_P9_DFORM_VECTOR): Likewise.
19928 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19929 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19931 2017-08-25 Alan Modra <amodra@gmail.com>
19933 PR rtl-optimization/81747
19934 * cse.c (cse_extended_basic_block): Don't attempt to record
19935 equivalences for degenerate conditional branches that branch
19936 to their fall-through.
19938 2017-08-24 Martin Sebor <msebor@redhat.com>
19940 PR middle-end/81908
19941 * gimple-fold.c (size_must_be_zero_p): New function.
19942 (gimple_fold_builtin_memory_op): Call it.
19944 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19946 * config/rs6000/mm_malloc.h: New file.
19948 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19950 PR tree-optimization/81913
19951 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19952 analysis when either IVs in condition can wrap.
19954 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19956 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19957 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19959 2017-08-24 Richard Biener <rguenther@suse.de>
19962 * targhooks.c (default_target_can_inline_p): Properly
19963 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19964 is present and always compare.
19965 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19966 infer -mfpmath=sse from TARGET_SSE_P.
19967 (ix86_can_inline_p): Properly use target_option_default_node when
19968 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19970 2017-08-24 Richard Biener <rguenther@suse.de>
19973 * dwarf2out.c (output_die): Handle flag_generate_offload like
19975 (output_comp_unit): Likewise.
19976 (gen_array_type_die): Likewise.
19977 (dwarf2out_early_finish): Likewise.
19978 (note_variable_value_in_expr): Likewise.
19979 (dwarf2out_finish): Likewise. Adjust assert.
19980 * cgraphunit.c (symbol_table::compile): Move setting of
19981 flag_generate_offload earlier ...
19982 (symbol_table::finalize_compilation_unit): ... here, before
19983 early debug finalization.
19985 2017-08-24 Richard Biener <rguenther@suse.de>
19987 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19988 and ipa-fnsummary.h.
19989 (ix86_can_inline_p): When ix86_fpmath flags do not match
19990 check whether the callee uses FP math at all.
19992 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19994 PR middle-end/81931
19995 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19996 instead of TYPE_PRECISION.
19998 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19999 Alan Hayward <alan.hayward@arm.com>
20000 David Sherwood <david.sherwood@arm.com>
20002 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20004 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20006 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20008 (ix86_frame::stack_realign_allocate): New field.
20009 (struct machine_frame_state): Modify comments.
20010 (machine_frame_state::sp_realigned_fp_end): New field.
20011 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20012 layout calculation.
20013 (sp_valid_at): Add assertion to assure no attempt to access invalid
20014 offset of a realigned stack.
20015 (fp_valid_at): Likewise.
20016 (choose_baseaddr): Modify comments.
20017 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20018 ix86_expand_prologue.
20019 (ix86_expand_prologue): Modify stack realignment and allocation.
20020 (ix86_expand_epilogue): Modify comments.
20021 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20022 avx2_runtime, avx512f, and avx512f_runtime.
20024 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20026 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20027 (mstackrealign): Do not init to -1.
20028 * config/i386/i386.c (ix86_option_override_internal):
20029 Check opts_set, not opts when setting default value of
20030 opts->x_ix86_force_align_arg_pointer.
20032 2017-08-23 Richard Biener <rguenther@suse.de>
20034 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20035 lang_hooks.decl_printable_name.
20036 * print-rtl-function.c (print_rtx_function): Likewise.
20037 * tree-pretty-print.c (dump_function_header): Likewise.
20039 2017-08-23 Richard Biener <rguenther@suse.de>
20042 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20043 -g0 at compile-time.
20045 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20047 PR middle-end/19706
20048 * doc/sourcebuild.texi (Other hardware attributes):
20051 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20053 PR middle-end/19706
20054 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20055 Fix single-use check.
20057 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20059 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20061 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20063 * doc/install.texi: Modify to add more details on running selected
20066 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20068 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20069 is combined with -mabi=ms.
20070 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20073 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20075 PR tree-optimization/81488
20076 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20077 and cached_basis fields.
20078 (MAX_SPREAD): New constant.
20079 (alloc_cand_and_find_basis): Initialize new fields.
20080 (clear_visited): New function.
20081 (create_phi_basis_1): Rename from create_phi_basis, set visited
20082 and cached_basis fields.
20083 (create_phi_basis): New wrapper function.
20084 (phi_add_costs_1): Rename from phi_add_costs, add spread
20085 parameter, set visited field, short-circuit when limits reached.
20086 (phi_add_costs): New wrapper function.
20087 (record_phi_increments_1): Rename from record_phi_increments, set
20089 (record_phi_increments): New wrapper function.
20090 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20091 (phi_incr_cost): New wrapper function.
20092 (all_phi_incrs_profitable_1): Rename from
20093 all_phi_incrs_profitable, set visited field.
20094 (all_phi_incrs_profitable): New wrapper function.
20096 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20097 Alan Hayward <alan.hayward@arm.com>
20098 David Sherwood <david.sherwood@arm.com>
20100 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20101 that takes the outer and inner modes.
20102 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20103 comparison as the canonical test for a paradoxical subreg.
20104 * combine.c (simplify_set): Use paradoxical_subreg_p.
20105 (make_extraction): Likewise.
20106 (force_to_mode): Likewise.
20107 (rtx_equal_for_field_assignment_p): Likewise.
20108 (gen_lowpart_for_combine): Likewise.
20109 (simplify_comparison): Likewise.
20110 * cse.c (equiv_constant): Likewise.
20111 * expmed.c (store_bit_field_1): Likewise.
20112 * final.c (alter_subreg): Likewise.
20113 * fwprop.c (propagate_rtx): Likewise.
20114 (forward_propagate_subreg): Likewise.
20115 * ira-conflicts.c (ira_build_conflicts): Likewise.
20116 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20117 * lra-constraints.c (curr_insn_transform): Likewise.
20118 (split_reg): Likewise.
20119 * lra-eliminations.c (move_plus_up): Likewise.
20120 (lra_eliminate_regs_1): Likewise.
20121 * recog.c (general_operand): Likewise.
20122 * ree.c (combine_reaching_defs): Likewise.
20123 * reload.c (push_reload): Likewise.
20124 (find_reloads): Likewise.
20125 * reload1.c (elimination_effects): Likewise.
20126 (compute_reload_subreg_offset): Likewise.
20127 (choose_reload_regs): Likewise.
20128 * rtlanal.c (subreg_lsb_1): Likewise.
20129 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20130 (simplify_subreg): Likewise.
20131 * var-tracking.c (track_loc_p): Likewise.
20132 * emit-rtl.c (byte_lowpart_offset): Likewise.
20133 (paradoxical_subreg_p): Delete out-of-line definition.
20135 2017-08-22 Jeff Law <law@redhat.com>
20137 PR tree-optimization/81741
20138 PR tree-optimization/71947
20139 * tree-ssa-dom.c: Include tree-inline.h.
20140 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20141 equivalences if one is more expensive to compute than the other.
20142 * tree-ssa-scopedtables.h (class const_or_copies): Make
20143 record_const_or_copy_raw method private.
20144 (class avail_exprs_stack): New method simplify_binary_operation.
20145 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20146 avail_exprs_stack::simplify_binary_operation as needed.
20147 (avail_exprs_stack::simplify_binary_operation): New function.
20149 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20151 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20152 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20153 (DOT_SYMBOLS): Likewise.
20154 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20155 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20156 (RS6000_ABI_NAME): Likewise.
20157 (TARGET_CMODEL): Likewise.
20158 (TOC_SECTION_ASM_OP): Likewise.
20159 (SET_CMODEL): New macro.
20160 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20162 2017-08-22 Richard Biener <rguenther@suse.de>
20164 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20165 to allow for free-lang-data replacements similar to verify_type_variant.
20167 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20168 Alan Hayward <alan.hayward@arm.com>
20169 David Sherwood <david.sherwood@arm.com>
20171 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20172 VOIDmode for the LABEL_REF.
20174 2017-08-22 Richard Biener <rguenther@suse.de>
20176 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20178 2017-08-22 Richard Biener <rguenther@suse.de>
20180 * common.opt (feliminate-dwarf2-dups): Ignore.
20181 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20182 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20183 same_die_p_wrap, compute_section_prefix,
20184 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20185 (comdat_symbol_id, comdat_symbol_number): Likewise.
20186 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20188 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20189 (output_die): Mark unreachable path unreachable.
20190 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20191 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20192 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20193 (dwarf2out_early_finish): Likewise.
20195 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20197 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20199 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20202 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20203 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20204 (avr_attribute_table) <io, io_low, address>: Initialize
20205 .decl_required with true.
20207 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20209 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20210 undocumented debugging options.
20211 (-mvsx-scalar-double): Likewise.
20212 (-mallow-df-permute): Likewise.
20213 (-mvectorize-builtins): Likewise.
20214 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20215 (rs6000_builtin_vectorized_function): Likewise.
20216 (rs6000_builtin_md_vectorized_function): Likewise.
20217 (rs6000_opt_vars): Likewise.
20219 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20222 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20223 (*btrq_imm): Rename from *btrq.
20224 (*btcq_imm): Rename from *btcq.
20225 (btsc): New code attribute.
20226 (*<btsc><mode>): New insn pattern.
20227 (*btr<mode>): Ditto.
20228 (*<btsc><mode>_mask): New insn_and_split pattern.
20229 (*btr<mode>_mask): Ditto.
20231 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20232 Alan Hayward <alan.hayward@arm.com>
20233 David Sherwood <david.sherwood@arm.com>
20235 * function.c (pad_below): Simplify padding calculation.
20237 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20238 Alan Hayward <alan.hayward@arm.com>
20239 David Sherwood <david.sherwood@arm.com>
20241 * target.def (function_prologue): Remove frame size argument.
20242 (function_epilogue): Likewise.
20243 * doc/tm.texi: Regenerate.
20244 * final.c (final_start_function): Update call to function_prologue.
20245 (final_end_function): Update call to function_epilogue.
20246 (default_function_pro_epilogue): Remove frame size argument.
20247 * output.h (default_function_pro_epilogue): Likewise.
20248 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20249 (arm_output_function_prologue): Likewise.
20250 * config/frv/frv.c (frv_function_prologue): Likewise.
20251 (frv_function_epilogue): Likewise.
20252 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20253 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20254 (ia64_output_function_epilogue): Likewise.
20255 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20256 (m32r_output_function_epilogue): Likewise.
20257 * config/microblaze/microblaze.c (microblaze_function_prologue)
20258 (microblaze_function_epilogue): Likewise.
20259 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20260 (mips_output_function_epilogue): Likewise.
20261 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20262 (mmix_target_asm_function_epilogue): Likewise.
20263 * config/msp430/msp430.c (msp430_start_function): Likewise.
20264 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20265 (nds32_asm_function_epilogue): Likewise.
20266 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20267 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20268 (pa_output_function_epilogue): Likewise.
20269 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20270 (rs6000_output_function_epilogue): Likewise.
20271 * config/rl78/rl78.c (rl78_start_function): Likewise.
20272 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20273 (rs6000_output_function_epilogue): Likewise.
20274 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20275 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20276 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20277 (sparc_asm_function_epilogue): Likewise.
20279 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20281 * tree.h (type_has_mode_precision_p): New function.
20282 * convert.c (convert_to_integer_1): Use it.
20283 * expr.c (expand_expr_real_2): Likewise.
20284 (expand_expr_real_1): Likewise.
20285 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20286 * match.pd: Likewise.
20287 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20288 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20289 * tree-tailcall.c (process_assignment): Likewise.
20290 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20291 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20292 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20293 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20294 (vectorizable_assignment): Likewise.
20295 (vectorizable_shift): Likewise.
20296 (vectorizable_operation): Likewise.
20297 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20299 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20301 * match.pd: Add pow (C, x) simplification.
20303 2017-08-21 Richard Biener <rguenther@suse.de>
20305 PR tree-optimization/81900
20306 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20307 for blocks with abnormal predecessors.
20308 (compute_antic): Do not set visited flag prematurely.
20310 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20313 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20315 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20317 * stor-layout.h (vector_type_mode): Move to...
20318 * tree.h (vector_type_mode): ...here.
20319 * stor-layout.c (vector_type_mode): Move to...
20320 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20322 2017-08-21 Richard Biener <rguenther@suse.de>
20324 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20325 register_external_die hooks.
20326 (debug_false_tree_charstarstar_uhwistar): Declare.
20327 (debug_nothing_tree_charstar_uhwi): Likewise.
20328 * debug.c (do_nothing_debug_hooks): Adjust.
20329 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20330 (debug_nothing_tree_charstar_uhwi): Likewise.
20331 * dbxout.c (dbx_debug_hooks): Adjust.
20332 (xcoff_debug_hooks): Likewise.
20333 * sdbout.c (sdb_debug_hooks): Likewise.
20334 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20335 * dwarf2out.c (macinfo_label_base): New global.
20336 (dwarf2out_register_external_die): New function for the
20337 register_external_die hook.
20338 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20339 (dwarf2_debug_hooks): Use them.
20340 (dwarf2_lineno_debug_hooks): Adjust.
20341 (struct die_struct): Add with_offset flag.
20342 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20343 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20344 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20345 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20346 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20347 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20348 defining section names for the early LTO debug variants.
20349 (reset_indirect_string): New helper.
20350 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20351 (print_dw_val): Add support for offsetted symbol references.
20352 (get_ultimate_context): Split out from is_cxx.
20353 (is_cxx): Use get_ultimate_context.
20354 (is_fortran): Add decl overload.
20355 (compute_comp_unit_symbol): Split out worker from
20356 compute_section_prefix.
20357 (compute_section_prefix): Call compute_comp_unit_symbol and
20358 set comdat_type_p here.
20359 (output_die): Skip DIE symbol output for the LTO added one.
20360 Handle DIE symbol references with offset.
20361 (output_comp_unit): Guard section name mangling properly.
20362 For LTO debug sections emit a symbol at the section beginning
20363 which we use to refer to its DIEs.
20364 (add_abstract_origin_attribute): For DIEs registered via
20365 dwarf2out_register_external_die directly refer to the early
20366 DIE rather than indirectly through the shadow one we created.
20367 Remove obsolete call to dwarf2out_abstract_function for
20368 non-function/block origins.
20369 (gen_array_type_die): When generating early LTO debug do
20370 not emit DW_AT_string_length.
20371 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20372 late when in LTO. As suggested place a gcc_unreachable for
20373 the DECL_ABSTRACT_P case.
20374 (gen_subprogram_die): Avoid another specification DIE
20375 for early built declarations/definitions for the late LTO case.
20376 (gen_variable_die): Add type references for late duplicated VLA dies
20378 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20379 we have the abstract instance already.
20380 (process_scope_var): Adjust decl DIE contexts in LTO which
20381 first puts them in limbo.
20382 (gen_decl_die): Do not generate type DIEs late apart from
20383 types for VLAs or for decls we do not yet have a DIE. Do not
20384 call dwarf2out_abstract_function late.
20385 (dwarf2out_early_global_decl): Make sure to create DIEs
20386 for abstract instances of a decl first.
20387 (dwarf2out_late_global_decl): Adjust comment.
20388 (output_macinfo_op): With multiple macro sections use
20389 macinfo_label_base to distinguish labels.
20390 (output_macinfo): Likewise. Update macinfo_label_base.
20391 Pass in the line info label.
20392 (note_variable_value_in_expr): When generating LTO resolve
20393 all variable values here by generating DIEs as needed.
20394 (init_sections_and_labels): Add early LTO debug flag parameter
20395 and generate different sections and names if set. Add generation
20396 counter for the labels so we can have multiple of them.
20397 (reset_dies): Helper to allow DIEs to be output multiple times.
20398 (dwarf2out_finish): When outputting DIEs to the fat part of an
20399 LTO object first reset DIEs.
20400 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20401 (modified_type_die): Check for decl_ultimate_origin being self
20403 (gen_type_die_with_usage): Likewise.
20404 (gen_typedef_die): Allow decl_ultimate_origin being self.
20405 (set_decl_abstract_flags): Remove.
20406 (set_block_abstract_flags): Likewise.
20407 (dwarf2out_abstract_function): Treat the early generated DIEs
20408 as the abstract copy and only add DW_AT_inline and
20409 DW_AT_artificial here and call set_decl_origin_self.
20410 If the DIE has an abstract origin don't do anything.
20411 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20412 if we have none yet (Go fails to build one, PR78628).
20413 (variably_modified_type_p): Prevent endless recursion for Ada
20414 cyclic pointer types.
20415 * lto-streamer-in.c: Include debug.h.
20416 (dref_queue): New global.
20417 (lto_read_tree_1): Stream in DIE references.
20418 (lto_input_tree): Register DIE references.
20419 (input_function): Stream DECL_DEBUG_ARGS.
20420 * lto-streamer-out.c: Include debug.h.
20421 (lto_write_tree_1): Output DIE references.
20422 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20423 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20424 (output_function): Stream DECL_DEBUG_ARGS.
20425 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20426 Stream DECL_ABSTRACT_ORIGIN.
20427 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20428 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20429 DECL_CONTEXT for file-scope decls.
20430 * lto-streamer.h (struct dref_entry): Declare.
20431 (dref_queue): Likewise.
20432 * cfgexpand.c (pass_expand::execute): Do not call the
20433 outlining_inline_function hook here.
20434 * lto-wrapper.c (debug_obj): New global.
20435 (tool_cleanup): Unlink it if required.
20436 (debug_objcopy): New function.
20437 (run_gcc): Handle early debug sections in the IL files by
20438 extracting them to separate files, partially linkin them and
20439 feeding the result back as result to the linker.
20440 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20441 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20442 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20443 sections into a separate segment.
20444 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20446 (darwin_asm_dwarf_section): Likewise.
20447 (darwin_asm_output_dwarf_offset): Likewise.
20448 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20450 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20451 Alan Hayward <alan.hayward@arm.com>
20452 David Sherwood <david.sherwood@arm.com>
20454 * read-md.h (md_reader::record_potential_iterator_use): Replace
20455 pointer argument with an rtx and an index.
20456 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20457 (apply_mode_iterator): Likewise.
20458 (apply_code_iterator): Likewise.
20459 (apply_int_iterator): Likewise.
20460 (apply_subst_iterator): Likewise.
20461 (record_iterator_use): Likewise.
20462 (record_attribute_use): Likewise.
20463 (md_reader::record_potential_iterator_use): Likewise. Update calls
20464 to record_iterator_use and apply_iterator.
20465 (iterator_use): Replace ptr with x and index.
20466 (attribute_use): Likewise.
20467 (apply_attribute_uses): Update calls to apply_iterator.
20468 (apply_iterators): Likewise. Update initialization of iterator_use.
20469 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20470 and record_potential_iterator_use.
20471 (rtx_reader::read_rtx_operand): Likewise.
20473 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20474 Alan Hayward <alan.hayward@arm.com>
20475 David Sherwood <david.sherwood@arm.com>
20477 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20480 2017-08-21 Richard Biener <rguenther@suse.de>
20482 PR middle-end/81884
20483 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20484 at struct end conservatively when comparing common bases.
20486 2017-08-21 Richard Biener <rguenther@suse.de>
20488 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20489 (mem_ref_in_stmt): Remove.
20490 (determine_max_movement): Use ref index to get at the reference.
20491 (invariantness_dom_walker::before_dom_children): Deal with
20492 lim data already initialized.
20493 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20495 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20497 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20498 (TARGET_ISA_ROUND): Ditto.
20499 (TARGET_ROUND): Ditto.
20500 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20501 * config/i386/i386.md: Ditto.
20502 * config/i386/sse.md: Ditto.
20503 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20504 with OPTION_MASK_ISA_SSE4_1.
20506 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20509 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20510 __builtin_ia32_lzcnt_u16.
20512 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20515 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20516 (rs6000_set_current_function): Rewrite function to use it.
20518 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20521 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20522 and TYPE_WARN_IF_NOT_ALIGN.
20523 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20524 (handle_warn_if_not_align): New.
20525 (place_union_field): Call handle_warn_if_not_align.
20526 (place_field): Call handle_warn_if_not_align.
20527 Copy TYPE_WARN_IF_NOT_ALIGN.
20528 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20529 (layout_type): Likewise.
20530 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20532 (tree_decl_common): Add warn_if_not_align.
20533 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20534 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20535 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20536 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20537 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20538 * doc/extend.texi: Document warn_if_not_aligned attribute.
20539 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20541 2017-08-17 Martin Liska <mliska@suse.cz>
20544 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20545 (get_data_dependence): Use it as pointer type.
20546 (distribute_loop): Likewise.
20548 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20550 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20551 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20552 (p8_vmrgow_<mode>_direct): New define_insn.
20553 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20554 handle endianness for vmrgew and vmrgow permute patterns.
20556 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20558 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20559 * config/rs6000/rs6000-cpus.def: Remove comment.
20560 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20561 (POWERPC_MASKS): Likewise.
20562 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20563 use of TARGET_VSX_TIMODE.
20564 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20565 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20566 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20567 (rs6000_option_override_internal): Remove dead code.
20568 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20569 (rs6000_legitimize_reload_address): Likewise.
20570 (rs6000_legitimate_address_p): Likewise.
20571 (rs6000_opt_masks): Delete "vsx-timode".
20572 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20573 from function comment.
20574 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20575 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20576 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20578 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20579 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20580 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20581 (VSX_TI): Likewise.
20583 (define_peephole2): Likewise.
20585 2017-08-17 Martin Sebor <msebor@redhat.com>
20588 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20589 past the end of an array.
20590 (test_pp_format): Add test cases.
20592 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20594 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20595 missing ECF_NOTHROW flags.
20597 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20600 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20601 operands residing in integer registers.
20602 (*vsx_le_perm_load_<mode>): Likewise.
20603 (*vsx_le_perm_store_<mode>): Likewise.
20604 (define_peephole2): Add peepholes to optimize the above.
20606 2017-08-17 Marek Polacek <polacek@redhat.com>
20608 PR middle-end/81814
20609 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20610 to mimic what shorten_compare did. Change the return type to bool.
20611 (fold_cond_expr_with_comparison): Update call to
20612 operand_equal_for_comparison_p.
20613 (fold_ternary_loc): Likewise.
20615 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20617 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20619 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20621 2017-08-17 Richard Biener <rguenther@suse.de>
20623 * tree-ssa-structalias.c (solve_graph): When propagating
20624 to successors update the graphs succ edges and avoid duplicate work.
20626 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20629 * config/i386/i386.c (ix86_option_override_internal): Save target
20630 specific options after ix86_stack_protector_guard_reg was changed.
20632 2017-08-17 Richard Biener <rguenther@suse.de>
20634 PR tree-optimization/81827
20635 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20636 (new_var_info): Initialize it conservatively.
20637 (get_call_vi): Mark register vars.
20638 (new_scalar_tmp_constraint_exp): Likewise.
20639 (handle_rhs_call): Likewise.
20640 (handle_const_call): Likewise.
20641 (create_function_info_for): Likewise.
20642 (solve_constraints): Sort varinfos to separate register from
20643 non-register vars to pack points-to solution bitmaps during
20646 2017-08-17 Marek Polacek <polacek@redhat.com>
20648 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20650 2017-08-17 Richard Biener <rguenther@suse.de>
20652 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20653 to true when overflow is undefined and we saturated the result.
20655 2017-08-17 Alan Modra <amodra@gmail.com>
20658 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20659 Don't use store multiple if only one reg needs saving.
20660 (interesting_frame_related_regno): New function.
20661 (rs6000_frame_related): Don't emit frame info for regs that
20663 (rs6000_emit_epilogue): Likewise.
20665 2017-08-16 Nathan Sidwell <nathan@acm.org>
20667 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20668 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20669 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20670 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20671 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20672 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20673 (verify_type): Adjust for TYPE_BINFO move.
20674 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20675 process TYPE_BINFO directly.
20676 (hash_tree): Likewise.
20677 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20679 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20682 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20684 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20686 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20689 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20690 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20691 Add ix86_binary_operator_ok to insn constraint.
20692 (*iordi_1_bts): Ditto.
20693 (*xordi_1_btc): Ditto.
20694 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20695 Update corresponding peephole2 pattern.
20699 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20701 PR tree-optimization/81832
20702 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20703 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20705 2017-08-16 Marek Polacek <polacek@redhat.com>
20708 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20709 regarding size_zero_node.
20711 2017-08-16 Martin Liska <mliska@suse.cz>
20714 * config.gcc: Respect previously set extra_objs in case
20717 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20719 PR tree-optimization/81835
20720 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20721 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20722 not depend on the phi.
20724 2017-08-16 Alan Modra <amodra@gmail.com>
20726 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20729 2017-08-16 Alan Modra <amodra@gmail.com>
20731 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20732 (save_reg_p): ..into this. Update all callers.
20733 (first_reg_to_save): Simplify.
20735 2017-08-16 Alan Modra <amodra@gmail.com>
20737 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20740 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20744 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20745 constant count if that count is less than 32.
20747 2017-08-15 Nathan Sidwell <nathan@acm.org>
20749 * gcc.c (execute): Emit friendlier message if inferior is killed
20750 by an external cause.
20752 2017-08-15 Richard Biener <rguenther@suse.de>
20754 PR tree-optimization/81790
20755 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20756 CONSTRUCTORs from simplifying and VN.
20758 2017-08-14 Martin Sebor <msebor@redhat.com>
20760 * builtin-attrs.def: Add comments.
20762 2017-08-14 Martin Sebor <msebor@redhat.com>
20765 * doc/extend.texi (attribute nonstring): Document new attribute.
20767 2017-08-14 Martin Sebor <msebor@redhat.com>
20770 * tree-diagnostic.c (default_tree_printer): Handle %G.
20771 * gimple-pretty-print.h (percent_G_format): Declare new function.
20772 * gimple-pretty-print.c (percent_G_format): Define.
20773 * tree-pretty-print.c (percent_K_format): Add argument.
20775 2017-08-14 Martin Sebor <msebor@redhat.com>
20777 PR translation/79998
20778 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20779 Remove a stray space.
20781 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20784 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20785 (*iordi_1_bts): Ditto.
20786 (*xordi_1_btc): Ditto.
20788 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20791 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20792 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20794 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20795 quoted strings, and make more translator-friendly.
20796 (darwin_rs6000_override_options): Likewise.
20797 (rs6000_option_override_internal): Likewise.
20798 (rs6000_return_in_memory): Fix overlong line.
20799 (init_cmulative_args): Use quoted strings, and make more
20800 translator-friendly.
20801 (rs6000_pass_by_reference): Fix overlong line.
20802 (def_builtin): Use quoted strings.
20803 (altivec_expand_predicate_builtin): Use quoted strings, and make
20804 more translator-friendly.
20805 (htm_expand_builtin): Use quoted strings.
20806 (cpu_expand_builtin): Use quoted strings, and make more
20807 translator-friendly.
20808 (altivec_expand_builtin): Likewise.
20809 (paired_expand_predicate_builtin): Likewise.
20810 (rs6000_invalid_builtin): Likewise.
20811 (builtin_function_type): Use quoted strings.
20812 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20813 more translator-friendly.
20814 (rs6000_trampoline_init): Likewise.
20815 (rs6000_handle_altivec_attribute): Likewise.
20816 (rs6000_inner_target_options): Use quoted strings.
20817 (rs6000_disable_incompatible_switches): Likewise.
20818 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20819 strings, and make more translator-friendly.
20820 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20822 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20824 PR tree-optimization/81799
20825 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20826 cond_expr to simple gimple operand.
20828 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20830 PR middle-end/46932
20831 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20833 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20837 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20838 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20839 TARGET_GASISR_PROLOGUES.
20840 * config/avr/avr.c (avr_option_override): Same.
20841 (avr_pass_pre_proep::execute): Same.
20843 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20846 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20847 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20849 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20851 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20852 define_insn_and_split. Split to a memory load from 0 in
20853 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20854 using PTR mode iterator.
20855 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20856 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20857 (*add_tp_<mode>): Redefine as define_insn_and_split.
20858 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20859 address space. Merge with *add_tp_x32 using PTR mode iterator.
20860 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20861 Split to an add with a memory load from 0 in
20862 DEFAULT_TLS_SEG_REG address space.
20864 2017-08-12 Andrew Pinski <apinski@cavium.com>
20866 * config/aarch64/aarch64-option-extensions.def (rdma):
20867 Fix feature string to what Linux prints out in /proc/cpuinfo.
20869 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20872 * dwarf2out.c (modified_type_die): For C typedef types that have
20873 an ultimate origin, process the ultimate origin instead of the
20875 (gen_typedef_die): Assert that input DECLs have no ultimate
20877 (gen_type_die_with_usage): For typedef variants that have an
20878 ultimate origin, just call gen_decl_die on the original DECL.
20879 (process_scope_var): Avoid creating DIEs for local typedefs and
20880 concrete static variables.
20882 2017-08-12 Alan Modra <amodra@gmail.com>
20886 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20887 match gnu-user.h startfile.
20888 (ENDFILE_LINUX_SPEC): Similarly.
20890 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20893 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20895 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20897 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20898 * config/aarch64/aarch64.md (mov<mode>): Change.
20899 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20900 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20901 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20903 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20905 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20906 for storage order barriers.
20908 2017-08-11 Martin Liska <mliska@suse.cz>
20911 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20912 variables of void type.
20914 2017-08-11 Martin Liska <mliska@suse.cz>
20916 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20917 TARGET_SUPPORTS_ALIASES.
20918 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20919 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20920 (optimize_weakref): Likewise.
20921 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20922 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20923 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20925 2017-08-11 Martin Liska <mliska@suse.cz>
20928 * config/i386/i386.c (make_resolver_func): Do complete
20929 refactoring of the function.
20931 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20934 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20935 * config/i386/i386.c (ix86_stack_protect_guard): Use
20936 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20937 * doc/invoke.texi (x86 Options): Document
20938 -mstack-protector-guard-symbol= option.
20940 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20942 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20943 * config/i386/i386.c (ix86_split_stack_guard): New function.
20944 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20945 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20946 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20947 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20948 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20949 (split_stack_space_check): Call ix86_split_stack_guard.
20951 2017-08-10 Martin Sebor <msebor@redhat.com>
20953 * print-tree.c (print_node): Print location using the established
20955 Replace spaces with colons.
20956 (debug_raw, debug): Ditto.
20958 2017-08-10 Martin Sebor <msebor@redhat.com>
20961 * pretty-print.c (pp_format): Correct the handling of %s precision.
20963 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20966 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20968 (ix86_finalize_stack_frame_flags): This. Also clear
20969 frame_pointer_needed if -fno-omit-frame-pointer is used without
20971 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20972 with ix86_finalize_stack_frame_flags.
20973 (ix86_expand_epilogue): Likewise.
20974 (ix86_expand_split_stack_prologue): Likewise.
20975 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20977 2017-08-10 Martin Liska <mliska@suse.cz>
20980 * c-attribs.c (handle_target_attribute):
20981 Report warning for an empty string argument of target attribute.
20983 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20986 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20988 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20989 or DECL_NONLOCAL labels.
20990 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20991 or DECL_NONLOCAL labels here.
20993 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20995 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20996 to indicate when early gimple folding has been disabled.
20997 (rs6000_gimple_fold_builtin): Add debug content.
20998 (rs6000_invalid_builtin): Fix whitespace.
20999 (rs6000_expand_builtin): Fix whitespace.
21000 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21002 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21005 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21006 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21008 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21010 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21011 (qdf24xx): Likewise.
21012 * config/aarch64/aarch64.md: Include falkor.md.
21013 * config/aarch64/falkor.md: New.
21015 2017-08-09 Marek Polacek <polacek@redhat.com>
21018 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21019 * diagnostic.c (emit_diagnostic): Add a comment.
21020 (emit_diagnostic_valist): New function.
21022 2017-08-09 Marek Polacek <polacek@redhat.com>
21025 * input.c (make_location): New overload.
21026 * input.h (make_location): Declare.
21028 2017-08-08 Alan Modra <amodra@gmail.com>
21029 H.J. Lu <hongjiu.lu@intel.com>
21032 * gcc.c (NO_PIE_SPEC): Delete.
21033 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21035 (LINK_PIE_SPEC): ..to here.
21036 (LINK_COMMAND_SPEC): Support -no-pie.
21037 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21038 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21040 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21041 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21042 (ENDFILE_CRTEND_SPEC): Similarly.
21044 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21047 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21048 (mstack-protector-guard-offset=): Ditto.
21049 * config/i386/i386.c (ix86_option_override): Handle
21050 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21052 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21053 ix86_stack_protect_guard_offset variables.
21054 (TARGET_STACK_PROTECT_GUARD): Always define.
21055 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21056 and -mstack-protector-guard-offset= options.
21058 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21060 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21061 boundary case for the last candidate.
21063 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21065 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21066 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21068 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21070 PR middle-end/19706
21071 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21072 * config/aarch64/aarch64-builtins.c
21073 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21074 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21075 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21077 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21078 Andrew Pinski <pinskia@gmail.com>
21080 PR middle-end/19706
21081 * internal-fn.def (XORSIGN): New.
21082 * optabs.def (xorsign_optab): New.
21083 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21084 (convert_expand_mult_copysign): New.
21085 (pass_optimize_widening_mul::execute): Call
21086 convert_expand_mult_copysign.
21088 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21090 PR tree-optimization/81354
21091 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21092 Insert on edges rather than explicitly creating landing pads.
21093 (analyze_candidates_and_replace): Commit edge inserts.
21095 2017-08-08 Richard Biener <rguenther@suse.de>
21097 PR middle-end/81719
21098 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21099 (expand_simple_operations): Also look through ADDR_EXPRs with
21100 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21102 2017-08-08 Richard Biener <rguenther@suse.de>
21104 PR tree-optimization/81723
21105 * tree-vect-slp.c (struct bst_traits): New hash traits.
21106 (bst_fail): New global.
21107 (vect_build_slp_tree_2): New worker, split out from ...
21108 (vect_build_slp_tree): ... this now wrapping it with using
21109 bst_fail set to cache SLP tree build fails. Properly handle
21111 (vect_analyze_slp_instance): Allocate and free bst_fail.
21113 2017-08-08 Martin Liska <mliska@suse.cz>
21116 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21117 LABEL_DECLs that can be from a different function.
21119 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21121 PR tree-optimization/81744
21122 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21123 loop's number of iterations.
21125 2017-08-08 Martin Liska <mliska@suse.cz>
21127 * asan.c: Include header files.
21128 * attribs.c (build_decl_attribute_variant): New function moved
21130 (build_type_attribute_qual_variant): Likewise.
21131 (cmp_attrib_identifiers): Likewise.
21132 (simple_cst_list_equal): Likewise.
21133 (omp_declare_simd_clauses_equal): Likewise.
21134 (attribute_value_equal): Likewise.
21135 (comp_type_attributes): Likewise.
21136 (build_type_attribute_variant): Likewise.
21137 (lookup_ident_attribute): Likewise.
21138 (remove_attribute): Likewise.
21139 (merge_attributes): Likewise.
21140 (merge_type_attributes): Likewise.
21141 (merge_decl_attributes): Likewise.
21142 (merge_dllimport_decl_attributes): Likewise.
21143 (handle_dll_attribute): Likewise.
21144 (attribute_list_equal): Likewise.
21145 (attribute_list_contained): Likewise.
21146 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21147 (lookup_attribute_by_prefix): Likewise.
21148 * bb-reorder.c: Include header files.
21149 * builtins.c: Likewise.
21150 * calls.c: Likewise.
21151 * cfgexpand.c: Likewise.
21152 * cgraph.c: Likewise.
21153 * cgraphunit.c: Likewise.
21154 * convert.c: Likewise.
21155 * dwarf2out.c: Likewise.
21156 * final.c: Likewise.
21157 * fold-const.c: Likewise.
21158 * function.c: Likewise.
21159 * gimple-expr.c: Likewise.
21160 * gimple-fold.c: Likewise.
21161 * gimple-pretty-print.c: Likewise.
21162 * gimple.c: Likewise.
21163 * gimplify.c: Likewise.
21164 * hsa-common.c: Likewise.
21165 * hsa-gen.c: Likewise.
21166 * internal-fn.c: Likewise.
21167 * ipa-chkp.c: Likewise.
21168 * ipa-cp.c: Likewise.
21169 * ipa-devirt.c: Likewise.
21170 * ipa-fnsummary.c: Likewise.
21171 * ipa-inline.c: Likewise.
21172 * ipa-visibility.c: Likewise.
21174 * lto-cgraph.c: Likewise.
21175 * omp-expand.c: Likewise.
21176 * omp-general.c: Likewise.
21177 * omp-low.c: Likewise.
21178 * omp-offload.c: Likewise.
21179 * omp-simd-clone.c: Likewise.
21180 * opts-global.c: Likewise.
21181 * passes.c: Likewise.
21182 * predict.c: Likewise.
21183 * sancov.c: Likewise.
21184 * sanopt.c: Likewise.
21185 * symtab.c: Likewise.
21186 * toplev.c: Likewise.
21187 * trans-mem.c: Likewise.
21188 * tree-chkp.c: Likewise.
21189 * tree-eh.c: Likewise.
21190 * tree-into-ssa.c: Likewise.
21191 * tree-object-size.c: Likewise.
21192 * tree-parloops.c: Likewise.
21193 * tree-profile.c: Likewise.
21194 * tree-ssa-ccp.c: Likewise.
21195 * tree-ssa-live.c: Likewise.
21196 * tree-ssa-loop.c: Likewise.
21197 * tree-ssa-sccvn.c: Likewise.
21198 * tree-ssa-structalias.c: Likewise.
21199 * tree-ssa.c: Likewise.
21200 * tree-streamer-in.c: Likewise.
21201 * tree-vectorizer.c: Likewise.
21202 * tree-vrp.c: Likewise.
21203 * tsan.c: Likewise.
21204 * ubsan.c: Likewise.
21205 * varasm.c: Likewise.
21206 * varpool.c: Likewise.
21207 * tree.c: Remove functions moved to attribs.[ch].
21208 * tree.h: Likewise.
21209 * config/aarch64/aarch64.c: Add attrs.h header file.
21210 * config/alpha/alpha.c: Likewise.
21211 * config/arc/arc.c: Likewise.
21212 * config/arm/arm.c: Likewise.
21213 * config/avr/avr.c: Likewise.
21214 * config/bfin/bfin.c: Likewise.
21215 * config/c6x/c6x.c: Likewise.
21216 * config/cr16/cr16.c: Likewise.
21217 * config/cris/cris.c: Likewise.
21218 * config/darwin.c: Likewise.
21219 * config/epiphany/epiphany.c: Likewise.
21220 * config/fr30/fr30.c: Likewise.
21221 * config/frv/frv.c: Likewise.
21222 * config/ft32/ft32.c: Likewise.
21223 * config/h8300/h8300.c: Likewise.
21224 * config/i386/winnt.c: Likewise.
21225 * config/ia64/ia64.c: Likewise.
21226 * config/iq2000/iq2000.c: Likewise.
21227 * config/lm32/lm32.c: Likewise.
21228 * config/m32c/m32c.c: Likewise.
21229 * config/m32r/m32r.c: Likewise.
21230 * config/m68k/m68k.c: Likewise.
21231 * config/mcore/mcore.c: Likewise.
21232 * config/microblaze/microblaze.c: Likewise.
21233 * config/mips/mips.c: Likewise.
21234 * config/mmix/mmix.c: Likewise.
21235 * config/mn10300/mn10300.c: Likewise.
21236 * config/moxie/moxie.c: Likewise.
21237 * config/msp430/msp430.c: Likewise.
21238 * config/nds32/nds32-isr.c: Likewise.
21239 * config/nds32/nds32.c: Likewise.
21240 * config/nios2/nios2.c: Likewise.
21241 * config/nvptx/nvptx.c: Likewise.
21242 * config/pa/pa.c: Likewise.
21243 * config/pdp11/pdp11.c: Likewise.
21244 * config/powerpcspe/powerpcspe.c: Likewise.
21245 * config/riscv/riscv.c: Likewise.
21246 * config/rl78/rl78.c: Likewise.
21247 * config/rx/rx.c: Likewise.
21248 * config/s390/s390.c: Likewise.
21249 * config/sh/sh.c: Likewise.
21250 * config/sol2.c: Likewise.
21251 * config/sparc/sparc.c: Likewise.
21252 * config/spu/spu.c: Likewise.
21253 * config/stormy16/stormy16.c: Likewise.
21254 * config/tilegx/tilegx.c: Likewise.
21255 * config/tilepro/tilepro.c: Likewise.
21256 * config/v850/v850.c: Likewise.
21257 * config/vax/vax.c: Likewise.
21258 * config/visium/visium.c: Likewise.
21259 * config/xtensa/xtensa.c: Likewise.
21261 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21264 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21265 constraints since the -mupper-regs-* switches have been
21267 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21268 into a vector from a double word element that was extracted from
21269 another vector, and eliminate extra XXPERMDI instructions.
21270 (vsx_concat_<mode>_2): Likewise.
21271 (vsx_concat_<mode>_3): Likewise.
21272 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21273 concat to allow optimizing inserts from previous extracts.
21275 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21277 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21278 memory reference to a SSP offset in TLS address space.
21279 (ix86_print_operand) <case '@'>: Remove.
21280 (ix86_print_operand_punct_valid_p): Remove '@' code.
21281 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21282 UNSPEC_SP_TLS_TEST.
21283 (stack_tls_protect_set_<mode>): Remove.
21284 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21285 (stack_tls_protect_test_<mode>): Remove.
21286 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21288 2017-08-07 Olivier Hainque <hainque@adacore.com>
21291 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21293 2017-08-07 Douglas Rupp <rupp@adacore.com>
21295 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21296 variable was referenced as multidir in command.
21298 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21301 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21304 2017-08-07 Martin Liska <mliska@suse.cz>
21306 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21307 * config/rl78/rl78.c: Add include of attribs.h.
21308 * config/sh/sh.c: Likewise.
21309 * config/v850/v850.c: Likewise.
21311 2017-08-07 Tom de Vries <tom@codesourcery.com>
21313 PR middle-end/78266
21314 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21316 2017-08-07 Martin Liska <mliska@suse.cz>
21318 * config/mips/mips.c: Include attribs.h.
21320 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21323 * doc/invoke.texi: Document change in behvaior for -Ofast for
21326 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21328 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21330 (aarch64_pop_regs): Likewise.
21331 (aarch64_gen_load_pair): Likewise.
21332 (aarch64_save_callee_saves): Likewise.
21333 (aarch64_restore_callee_saves): Likewise.
21335 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21337 * config/i386/i386.c: Revert the last change.
21339 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21342 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21344 (ix86_finalize_stack_frame_flags): This. Also clear
21345 frame_pointer_needed if -fno-omit-frame-pointer is used without
21347 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21348 with ix86_finalize_stack_frame_flags.
21349 (ix86_expand_epilogue): Likewise.
21350 (ix86_expand_split_stack_prologue): Likewise.
21352 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21355 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21356 to P_AES for Westmere.
21358 2017-08-07 Jonathan Yong <10walls@gmail.com>
21360 * config/i386/mingw.opt (fset-stack-executable): Removed.
21361 * config/i386/cygming.opt (fset-stack-executable): Moved
21363 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21365 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21367 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21369 2017-08-07 Marek Polacek <polacek@redhat.com>
21371 PR middle-end/81737
21372 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21374 2017-08-07 Martin Liska <mliska@suse.cz>
21376 * attribs.h (canonicalize_attr_name): New function.
21377 (cmp_attribs): Move from c-format.c and adjusted.
21378 (is_attribute_p): Moved from tree.h.
21379 * tree-inline.c: Add new includes.
21380 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21381 (private_is_attribute_p): Remove.
21382 (private_lookup_attribute): Likewise.
21383 (private_lookup_attribute_by_prefix): Simplify.
21384 (remove_attribute): Use is_attribute_p.
21385 * tree.h: Remove removed declarations.
21387 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21389 PR middle-end/81698
21390 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21391 instead of computing it in the function. Formatting fix.
21392 (expand_case): Don't rely on default_edge being the first edge,
21393 clear it if removing it, pass default_edge to
21394 emit_case_dispatch_table.
21395 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21398 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21400 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21401 insn in the function, emit NOP after the insn.
21403 2017-08-06 Tom de Vries <tom@codesourcery.com>
21405 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21408 2017-08-06 Tom de Vries <tom@codesourcery.com>
21410 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21413 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21415 PR tree-optimization/57371
21416 * match.pd: New pattern.
21418 2017-08-04 Marek Polacek <polacek@redhat.com>
21420 PR middle-end/81695
21421 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21422 perform the computation in offset_int.
21424 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21426 PR tree-optimization/81136
21427 * tree-vectorizer.h: Include tree-hash-traits.h.
21428 (vec_base_alignments): New typedef.
21429 (vec_info): Add a base_alignments field.
21430 (vect_record_base_alignments): Declare.
21431 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21433 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21434 (create_data_ref): Add an is_conditional_in_stmt argument.
21435 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21436 the is_conditional_in_stmt field.
21437 (data_ref_loc): Add an is_conditional_in_stmt field.
21438 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21439 (find_data_references_in_stmt): Update call to create_data_ref.
21440 (graphite_find_data_references_in_stmt): Likewise.
21441 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21442 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21443 (vect_record_base_alignment): New function.
21444 (vect_record_base_alignments): Likewise.
21445 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21446 for nested statements even if we fail to compute a misalignment.
21447 Use pooled base alignments for unconditional references.
21448 (vect_find_same_alignment_drs): Compare base addresses instead
21450 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21451 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21453 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21455 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21456 Add an explicit name for the enum. Use auto_vec for slp_instances
21457 and grouped_stores.
21458 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21460 (_bb_vec_info): Add a constructor and destructor.
21461 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21462 (destroy_loop_vec_info): Delete.
21463 (vect_destroy_datarefs): Likewise.
21464 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21465 (vec_info::vec_info): New function.
21466 (vec_info::~vec_info): Likewise.
21467 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21468 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21469 destroy_loop_vec_info.
21470 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21471 (_loop_vec_info::_loop_vec_info): ...this.
21472 (destroy_loop_vec_info): Replace with...
21473 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21474 the stmt_vec_infos. Leave handling of vec_info information to its
21475 destructor. Remove explicit vector releases.
21476 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21477 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21478 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21479 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21480 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21481 (destroy_bb_vec_info): Replace with...
21482 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21483 information to its destructor.
21484 (vect_slp_analyze_bb_1): Use new and delete instead of
21485 new_bb_vec_info and destroy_bb_vec_info.
21486 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21489 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21491 * tree-data-ref.h (subscript): Add access_fn field.
21492 (data_dependence_relation): Add could_be_independent_p.
21493 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21494 (same_access_functions): Move to tree-data-ref.c.
21495 * tree-data-ref.c (ref_contains_union_access_p): New function.
21496 (access_fn_component_p): Likewise.
21497 (access_fn_components_comparable_p): Likewise.
21498 (dr_analyze_indices): Add a reference to access_fn_component_p.
21499 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21501 (constant_access_functions): Likewise.
21502 (add_other_self_distances): Likewise.
21503 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21504 (initialize_data_dependence_relation): Use XCNEW and remove
21505 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21506 of access functions that have the same type. Allow the
21507 subsequence to end with different bases in some circumstances.
21508 Record the chosen access functions in SUB_ACCESS_FN.
21509 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21510 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21511 (subscript_dependence_tester_1): Likewise dra and drb.
21512 (build_classic_dist_vector): Update calls accordingly.
21513 (subscript_dependence_tester): Likewise.
21514 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21515 DDR_COULD_BE_INDEPENDENT_P.
21516 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21517 comp_alias_ddrs instead of may_alias_ddrs.
21518 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21520 (vect_analyze_data_ref_dependence): Use it if
21521 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21522 distance vectors if that fails.
21523 (dependence_distance_ge_vf): New function.
21524 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21525 LOOP_VINFO_MAY_ALIAS_DDRS.
21527 2017-08-04 Richard Biener <rguenther@suse.de>
21529 PR middle-end/81705
21530 * fold-const.c (fold_binary_loc): Properly restrict
21531 minus_var0 && minus_var1 case when associating undefined overflow
21534 2017-08-04 Tom de Vries <tom@codesourcery.com>
21536 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21538 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21540 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21541 Don't start diagnostic messages with a capital letter.
21542 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21544 (rs6000_invalid_builtin): Likewise.
21545 (rs6000_trampoline_init): Likewise.
21547 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21550 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21551 after setting changeable df flags.
21553 2017-08-03 Richard Biener <rguenther@suse.de>
21555 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21556 up if the use is in USE - X.
21558 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21560 * toplev.c (dumpfile.h): New include.
21561 (internal_error_reentered): New static function. Use it...
21562 (internal_error_function): ...here to handle reentered internal_error.
21564 2017-08-03 Richard Biener <rguenther@suse.de>
21566 PR middle-end/81148
21567 * fold-const.c (split_tree): Add minus_var and minus_con
21568 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21569 here but always use minus_*.
21570 (associate_trees): Assert we never associate with MINUS_EXPR
21571 and NULL first operand. Do not recurse for PLUS_EXPR operands
21572 when associating as MINUS_EXPR either.
21573 (fold_binary_loc): Track minus_var and minus_con.
21575 2017-08-03 Tom de Vries <tom@codesourcery.com>
21578 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21579 ACCEL_COMPILER, apply finish_options on
21580 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21582 2017-08-03 Tom de Vries <tom@codesourcery.com>
21585 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21586 function_entry_patch_area_size > 0.
21588 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21591 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21592 instead of 10??LU, perform unit multiplication in wide_int,
21593 don't change alloc_object_size_limit if the limit is larger
21596 PR tree-optimization/81655
21597 PR tree-optimization/81588
21598 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21599 the case when ranges[i].low and high are 1 for unsigned type with
21602 PR middle-end/81052
21603 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21604 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21606 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21608 * tree-vrp.h: Add include guard.
21610 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21613 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21614 (ud2): New insn pattern.
21615 * config/i386/i386.c (ix86_expand_epilogue):
21616 For naked functions, generate ud2 instead of trap insn.
21618 2017-08-02 Marek Polacek <polacek@redhat.com>
21621 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21623 2017-08-02 Tom de Vries <tom@codesourcery.com>
21624 Cesar Philippidis <cesar@codesourcery.com>
21626 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21627 Add missing edge probabilities.
21629 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21631 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21632 Correct endianness.
21634 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21636 PR middle-end/79499
21637 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21638 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21639 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21640 prologue_seq sequences - if any.
21642 2017-08-02 Richard Biener <rguenther@suse.de>
21644 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21645 via vectors if supported, integer extracts via punning if supported
21646 or otherwise vector extracts.
21648 2017-08-02 Richard Biener <rguenther@suse.de>
21650 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21652 (bitmap_insert_into_set): ... this.
21654 2017-08-02 Richard Biener <rguenther@suse.de>
21656 PR tree-optimization/81633
21658 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21660 PR tree-optimization/71752
21661 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21663 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21665 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21666 (machine_function::call_ms2sysv_pad_out): Remove field.
21667 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21668 (ix86_compute_frame_layout): Likewise.
21670 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21673 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21674 attribute with interrupt attribute.
21676 2017-08-01 Andrew Pinski <apinski@cavium.com>
21678 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21679 BIT_INSERT_EXPR's operand 1
21680 to see if the types precision matches.
21682 2017-08-01 Martin Liska <mliska@suse.cz>
21684 PR middle-end/70140
21685 * builtins.c (expand_builtin_memcpy_args): Remove.
21686 (expand_builtin_memcpy): Call newly added function
21687 expand_builtin_memory_copy_args.
21688 (expand_builtin_memcpy_with_bounds): Likewise.
21689 (expand_builtin_mempcpy): Remove last argument.
21690 (expand_builtin_mempcpy_with_bounds): Likewise.
21691 (expand_builtin_memory_copy_args): New function created from
21692 expand_builtin_mempcpy_args with small modifications.
21693 (expand_builtin_mempcpy_args): Remove.
21694 (expand_builtin_stpcpy): Remove unused argument.
21695 (expand_builtin): Likewise.
21696 (expand_builtin_with_bounds): Likewise.
21698 2017-08-01 Martin Liska <mliska@suse.cz>
21701 Make mempcpy more optimal (PR middle-end/70140).
21703 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21706 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21707 __builtin_vec_cmpne verify both arguments are compatible vectors
21708 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21709 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21710 move computation of aligned to after checking the argument types.
21714 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21715 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21718 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21719 Doug Rupp <rupp@adacore.com>
21720 Olivier Hainque <hainque@adacore.com>
21722 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21723 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21725 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21726 for TARGET_OS_CPP_BUILTIN.
21727 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21728 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21730 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21731 passing required abi options to the assembler for EABI configurations.
21732 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21733 of .text.hot and .text.unlikely sections for kernel modules when
21734 using ARM style exceptions.
21735 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21736 options. Add EXTRA_CC1_SPEC.
21737 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21739 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21741 (ARM_TARGET2_DWARF_FORMAT): Define.
21742 * config/arm/t-vxworks: Adjust multilib control to removal of the
21743 Diab command line options.
21745 2017-08-01 Martin Liska <mliska@suse.cz>
21747 PR gcov-profile/81561
21748 * gcov.c (unblock): Make unblocking safe as we need to preserve
21749 index correspondence of blocks and block_lists.
21751 2017-08-01 Richard Biener <rguenther@suse.de>
21753 PR tree-optimization/81181
21754 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21755 (compute_antic): ... end of iteration here.
21757 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21759 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21760 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21761 (ftree-slp-vectorize): Likewise.
21762 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21763 can no longer be set independent of flag_tree_loop_vectorize.
21764 * omp-general.c (emp_max_vf): Likewise.
21765 * opts.c (enable_fdo_optimizations): Remove references to
21766 flag_tree_vectorize, these are now implicit.
21767 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21768 and leave it for the options machinery.
21770 2017-08-01 Martin Liska <mliska@suse.cz>
21772 PR middle-end/70140
21773 * builtins.c (expand_builtin_memcpy_args): Remove.
21774 (expand_builtin_memcpy): Call newly added function
21775 expand_builtin_memory_copy_args.
21776 (expand_builtin_memcpy_with_bounds): Likewise.
21777 (expand_builtin_mempcpy): Remove last argument.
21778 (expand_builtin_mempcpy_with_bounds): Likewise.
21779 (expand_builtin_memory_copy_args): New function created from
21780 expand_builtin_mempcpy_args with small modifications.
21781 (expand_builtin_mempcpy_args): Remove.
21782 (expand_builtin_stpcpy): Remove unused argument.
21783 (expand_builtin): Likewise.
21784 (expand_builtin_with_bounds): Likewise.
21786 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21789 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21790 print "ds:" only for immediates in generic address space.
21792 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21795 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21796 (ix86_function_ok_for_sibcall): Return false for naked functions.
21798 2017-08-01 Richard Biener <rguenther@suse.de>
21800 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21801 (compute_antic): Seed worklist with exit block predecessors.
21802 * cfganal.c (dfs_find_deadend): For a cycle return the source
21803 of the edge closing it.
21805 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21807 * config/aarch64/aarch64.c
21808 (aarch64_can_const_movi_rtx_p): Move 0 check.
21810 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21812 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21813 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21815 * match.pd: Ditto in address comparison pattern.
21817 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21819 PR tree-optimization/81627
21820 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21821 closed ssa form for store-store chain.
21823 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21825 PR tree-optimization/81620
21826 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21827 for store-store chain.
21829 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21831 PR tree-optimization/81588
21832 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21833 ranges[i].in_p, invert comparison code ccode. For >/>=,
21834 swap rhs1 and rhs2 and comparison code unconditionally,
21835 for </<= don't do that. Don't swap rhs1/rhs2 again if
21836 ranges[i].in_p, instead invert comparison code ccode if
21837 opcode or oe->rank is BIT_IOR_EXPR.
21840 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21841 a direct optab to conversion optab.
21842 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21843 with GET_MODE_INNER as last argument instead of optab_handler.
21844 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21845 vector extraction if possible and optab is available.
21846 * expr.c (store_constructor): Use convert_optab_handler instead
21847 of optab_handler. Use vector initialization from smaller
21848 vectors if possible and optab is available.
21849 * tree-vect-stmts.c (vectorizable_load): Likewise.
21850 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21851 now have two modes.
21852 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21853 of vec_init from half-sized vectors with the same element mode.
21854 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21855 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21856 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21857 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21858 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21859 after mode in gen_vec_extract* calls.
21860 (vec_extract<mode>): Renamed to ...
21861 (vec_extract<mode><ssescalarmodelower>): ... this.
21862 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21863 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21864 element mode after mode in gen_vec_init* calls.
21865 (VEC_INIT_HALF_MODE): New mode iterator.
21866 (vec_init<mode>): Renamed to ...
21867 (vec_init<mode><ssescalarmodelower>): ... this.
21868 (vec_init<mode><ssehalfvecmodelower>): New expander.
21869 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21870 (vec_extractv2sfsf): ... this.
21871 (vec_initv2sf): Renamed to ...
21872 (vec_initv2sfsf): ... this.
21873 (vec_extractv2si): Renamed to ...
21874 (vec_extractv2sisi): ... this.
21875 (vec_initv2si): Renamed to ...
21876 (vec_initv2sisi): ... this.
21877 (vec_extractv4hi): Renamed to ...
21878 (vec_extractv4hihi): ... this.
21879 (vec_initv4hi): Renamed to ...
21880 (vec_initv4hihi): ... this.
21881 (vec_extractv8qi): Renamed to ...
21882 (vec_extractv8qiqi): ... this.
21883 (vec_initv8qi): Renamed to ...
21884 (vec_initv8qiqi): ... this.
21885 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21886 (vec_init<mode>): Renamed to ...
21887 (vec_init<mode><VEC_base_l>): ... this.
21888 (vec_extract<mode>): Renamed to ...
21889 (vec_extract<mode><VEC_base_l>): ... this.
21890 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21891 (vec_initv2sfsf): ... this.
21892 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21893 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21894 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21895 element mode after mode in gen_vec_init* calls.
21896 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21897 (vec_init<mode><Vel>): ... this.
21898 (vec_extract<mode>): Renamed to ...
21899 (vec_extract<mode><Vel>): ... this.
21900 * config/aarch64/iterators.md (Vel): New mode attribute.
21901 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21902 Add element mode after mode in gen_vec_extract* calls.
21903 * config/s390/vector.md (non_vec_l): New mode attribute.
21904 (vec_extract<mode>): Renamed to ...
21905 (vec_extract<mode><non_vec_l>): ... this.
21906 (vec_init<mode>): Renamed to ...
21907 (vec_init<mode><non_vec_l>): ... this.
21908 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21909 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21911 * config/arm/iterators.md (V_elem_l): New mode attribute.
21912 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21913 (vec_extract<mode><V_elem_l>): ... this.
21914 (vec_extractv2di): Renamed to ...
21915 (vec_extractv2didi): ... this.
21916 (vec_init<mode>): Renamed to ...
21917 (vec_init<mode><V_elem_l>): ... this.
21918 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21919 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21920 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21921 Add element mode after gen_vec_extract* calls.
21922 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21923 (vec_init<mode><unitmode>): ... this.
21924 (vec_extract<mode>): Renamed to ...
21925 (vec_extract<mode><unitmode>): ... this.
21926 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21927 (vec_init<mode><unitmode>): ... this.
21928 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21929 (vec_initv2sfsf): ... this.
21930 (vec_extractv2sf): Renamed to ...
21931 (vec_extractv2sfsf): ... this.
21932 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21933 Add element mode after gen_vec_extract* calls.
21934 * config/mips/mips.md (unitmode): New mode iterator.
21935 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21936 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21937 * config/spu/spu.md (inner_l): New mode attribute.
21938 (vec_init<mode>): Renamed to ...
21939 (vec_init<mode><inner_l>): ... this.
21940 (vec_extract<mode>): Renamed to ...
21941 (vec_extract<mode><inner_l>): ... this.
21942 * config/sparc/sparc.md (veltmode): New mode iterator.
21943 (vec_init<VMALL:mode>): Renamed to ...
21944 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21945 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21946 (vec_initv2sisi): ... this.
21947 (vec_initv2sf): Renamed to ...
21948 (vec_initv2sfsf): ... this.
21949 (vec_extractv2sf): Renamed to ...
21950 (vec_extractv2sfsf): ... this.
21951 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21952 (vec_init<mode>): Renamed to ...
21953 (vec_init<mode><VEC_base_l>): ... this.
21954 (vec_extract<mode>): Renamed to ...
21955 (vec_extract<mode><VEC_base_l>): ... this.
21956 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21957 (vec_initv2sfsf): ... this.
21958 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21959 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21960 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21961 gen_vec_init* calls.
21963 2017-08-01 Richard Biener <rguenther@suse.de>
21965 PR tree-optimization/81297
21966 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21967 TREE_OVERFLOW from INTEGER_CSTs.
21969 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21971 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21973 2017-07-31 Carl Love <cel@us.ibm.com>
21975 * config/rs6000/rs6000-c: Add support for built-in functions
21976 vector signed char vec_xl_be (signed long long, signed char *);
21977 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21978 vector signed int vec_xl_be (signed long long, signed int *);
21979 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21980 vector signed long long vec_xl_be (signed long long, signed long long *);
21981 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21982 vector signed short vec_xl_be (signed long long, signed short *);
21983 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21984 vector double vec_xl_be (signed long long, double *);
21985 vector float vec_xl_be (signed long long, float *);
21986 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21987 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21988 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21990 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21991 (altivec_expand_builtin): Add switch statement to call
21992 altivec_expand_xl_be for each builtin.
21993 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21994 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21995 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21996 __builtin_vsx_le_be_v16qi.
21997 * doc/extend.texi: Update the built-in documentation file for the
21998 new built-in functions.
22000 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22003 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22005 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22007 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22009 * config.gcc: Add z14.
22010 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22011 CPU model numbers for z13s and z14.
22012 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22014 * config/s390/s390-opts.h (enum processor_type): Rename
22015 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22016 * config/s390/s390.c (processor_table): Add field for CPU name to
22017 be passed to Binutils.
22018 (s390_asm_output_machine_for_arch): Use the new field in
22019 processor_table for Binutils.
22020 (s390_expand_builtin): Replace arch12 with z14.
22021 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22022 (s390_get_sched_attrmask): Likewise.
22023 (s390_get_unit_mask): Likewise.
22024 * config/s390/s390.opt: Add z14 to processor_type enum.
22026 2017-07-31 Martin Jambor <mjambor@suse.cz>
22029 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22030 regardless of optimization level.
22032 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22033 Martin Liska <mliska@suse.cz>
22035 * predict.def: Remove old comment and adjust probability.
22036 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22037 PREDICT statements.
22039 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22042 * config/i386/i386.c (ix86_function_naked): New function.
22043 (ix86_can_use_return_insn_p): Return false for naked functions.
22044 (ix86_expand_prologue): Skip prologue for naked functions.
22045 (ix86_expand_epilogue): Skip epilogue for naked functions
22046 and emit trap instruction.
22047 (ix86_warn_func_return): New function.
22048 (ix86_attribute_table): Add "naked" attribute specification.
22049 (TARGET_WARN_FUNC_RETURN): Define.
22050 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22052 2017-07-31 Martin Liska <mliska@suse.cz>
22054 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22055 (dump_gimple_bb_header): Always dump BB info.
22056 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22058 2017-07-31 Martin Liska <mliska@suse.cz>
22061 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22062 also with current_function_decl non-null equality.
22064 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22067 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22068 change type to the element type, instead add eltype variable and
22069 use it where we are interested in the element type.
22071 PR tree-optimization/81603
22072 * ipa-polymorphic-call.c
22073 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22074 offset arithmetic in offset_int, bail out if the resulting bit offset
22075 doesn't fit into shwi.
22077 2017-07-31 Martin Liska <mliska@suse.cz>
22079 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22080 (gimplify_save_expr): Fix comment.
22082 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22085 * config/i386/i386.c (ix86_function_arg): Update arguments for
22087 (ix86_compute_frame_layout): Set the initial stack offset to
22088 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22089 INCOMING_FRAME_SP_OFFSET.
22090 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22091 stack before exception handler returns.
22092 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22093 the 'ERROR_CODE' for exception handler.
22095 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22097 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22098 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22099 (ASM_OUTPUT_REG_POP): Ditto.
22100 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22101 instead of asm_fprintf to output pure string.
22103 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22105 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22106 to imported_module_or_decl hook.
22107 (debug_nothing_tree_tree_tree_bool): Remove.
22108 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22109 * debug.c (do_nothing_debug_hooks): Use
22110 debug_nothing_tree_tree_tree_bool_bool instead of
22111 debug_nothing_tree_tree_tree_bool.
22112 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22113 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22114 * sdbout.c (sdb_debug_hooks): Likewise.
22115 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22116 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22118 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22119 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22120 attribute, don't add anything.
22122 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22124 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22125 (fold_build2_stat_loc): Likewise.
22126 (fold_build3_stat_loc): Likewise.
22127 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22128 (fold_build1_loc): Remove macro.
22129 (fold_build2_loc): Likewise.
22130 (fold_build3_loc): Likewise.
22132 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22134 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22135 (gimple_build_debug_bind_source_stat): Likewise.
22136 * gimple.h (gimple_build_debug_bind): Remove macro.
22137 (gimple_build_debug_bind_source): Likewise.
22139 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22141 * bitmap.c (bitmap_alloc): Adjust.
22142 (bitmap_gc_alloc): Likewise.
22143 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22145 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22147 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22148 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22149 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22150 (bitmap_gc_alloc_stat): Likewise.
22151 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22153 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22155 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22156 * rtl.h (shallow_copy_rtx): Remove macro.
22158 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22160 * emit-rtl.c (gen_raw_REG): Adjust.
22161 * gengenrtl.c (gendef): Likewise.
22162 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22163 * rtl.h (rtx_alloc): Remove macro.
22165 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22167 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22168 (build_tree_list_stat): Likewise.
22169 * tree.h (build_tree_list): Remove macro.
22170 (build_tree_list_vec): Likewise.
22172 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22174 * tree.c (make_vector_stat): Remove _stat from name.
22175 (build_vector_stat): Likewise.
22176 * tree.h (make_vector_stat): Remove macro.
22177 (build_vector_stat): Likewise.
22179 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22181 * tree.h (build_var_debug_value): Remove prototype.
22183 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22185 * tree.c (tree_cons_stat): Remove _stat from name.
22186 * tree.h (tree_cons): Remove macro.
22188 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22190 * tree.c (build_vl_exp_stat): Remove _stat from name.
22191 * tree.h (build_vl_exp): Remove macro.
22193 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22195 * tree.c (build_decl_stat): Remove _stat from name.
22196 * tree.h (build_decl): Remove macro.
22198 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22200 * gimple.c (gimple_build_with_ops_stat): Adjust.
22201 (gimple_alloc_stat): Remove _stat from name.
22202 * gimple.h (gimple_alloc): Remove macro.
22204 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22206 * tree.c (make_tree_vec_stat): Remove _stat from name.
22207 (grow_tree_vec_stat): Likewise.
22208 * tree.h (make_tree_vec_stat): Adjust prototype.
22209 (grow_tree_vec_stat): Likewise.
22210 (make_tree_vec): Remove macro.
22211 (grow_tree_vec): Likewise.
22213 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22215 * fold-const.c (fold_build1_stat_loc): Adjust.
22216 (fold_build2_stat_loc): Likewise.
22217 (fold_build3_stat_loc): Likewise.
22218 * tree.c (build0_stat): Remove _stat from name.
22219 (build1_stat): Likewise.
22220 (build2_stat): Likewise.
22221 (build3_stat): Likewise.
22222 (build4_stat): Likewise.
22223 (build5_stat): Likewise.
22224 * tree.h (build1_loc): Remove macro, and rename _stat function
22226 (build2_loc): Likewise.
22227 (build3_loc): Likewise.
22228 (build4_loc): Likewise.
22229 (build5_loc): Likewise.
22231 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22233 * tree.c (make_int_cst_stat): Remove _stat from name.
22234 * tree.h (make_int_cst_stat): Adjust prototype.
22235 (make_int_cst): Remove macro.
22237 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22239 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22240 * tree.h (make_tree_binfo_stat): Adjust prototype.
22241 (make_tree_binfo): Remove.
22243 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22245 * tree.c (copy_node_stat): Rename to copy_node.
22246 (build_distinct_type_copy): Adjust.
22247 * tree.h (copy_node_stat): Adjust prototype.
22248 (copy_node): Remove macro.
22250 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22252 * tree.c (make_node_stat): rename to make_node.
22253 (build_tree_list_stat): Adjust.
22254 (build0_stat): Likewise.
22255 (build2_stat): Likewise.
22256 (build3_stat): Likewise.
22257 (build4_stat): Likewise.
22258 (build5_stat): Likewise.
22259 (build_decl_stat): Likewise.
22260 * tree.h (make_node_stat): Adjust prototype.
22261 (make_node): remove macro.
22263 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22265 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22266 (PPC_FEATURE2_SCV): Likewise.
22267 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22269 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22271 * config/aarch64/aarch64.c
22272 (aarch64_internal_mov_immediate): Add new special pattern.
22273 * config/aarch64/aarch64.md (*movdi_aarch64):
22274 Add reg/32bit const mov case.
22276 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22277 Richard Sandiford <richard.sandiford@linaro.org>
22279 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22280 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22281 Add integer and movi cases.
22282 (movi-split-hf-df-sf split, fp16): New.
22283 (enabled): Added TARGET_FP_F16INST.
22284 * config/aarch64/iterators.md (GPF_HF): New.
22285 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22287 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22289 * config/aarch64/aarch64.c
22290 (aarch64_simd_container_mode): Add prototype.
22291 (aarch64_expand_mov_immediate): Add HI support.
22292 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22293 (aarch64_can_const_movi_rtx_p): New.
22294 (aarch64_preferred_reload_class):
22295 Remove restrictions of using FP registers for certain SIMD operations.
22296 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22297 (aarch64_valid_floating_const): Add integer move validation.
22298 (aarch64_simd_imm_scalar_p): Remove.
22299 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22300 (aarch64_legitimate_constant_p): Expand list of supported cases.
22301 * config/aarch64/aarch64-protos.h
22302 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22303 (aarch64_reinterpret_float_as_int): New.
22304 (aarch64_simd_imm_scalar_p): Remove.
22305 * config/aarch64/constraints.md (Uvi): New.
22306 (Dd): Split into Ds and new Dd.
22307 * config/aarch64/aarch64.md (*movsi_aarch64):
22309 (*movdi_aarch64): Add SIMD mov case.
22311 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22313 * tree-predcom.c: (struct chain): Handle store-store chain in which
22314 stores for elimination only store loop invariant values.
22315 (execute_pred_commoning_chain): Ditto.
22316 (prepare_initializers_chain_store_elim): Ditto.
22317 (prepare_finalizers): Ditto.
22318 (is_inv_store_elimination_chain): New function.
22319 (initialize_root_vars_store_elim_1): New function.
22321 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22323 * tree-predcom.c: Revise general description of the pass.
22324 (enum chain_type): New enum type for store elimination.
22325 (struct chain): New field supporting store elimination.
22326 (struct component): Ditto.
22327 (dump_chain): Dump store-stores chain.
22328 (release_chain): Release resources.
22329 (split_data_refs_to_components): Compute and create component
22330 contains only stores for elimination.
22331 (get_chain_last_ref_at): New function.
22332 (make_invariant_chain): Initialization.
22333 (make_rooted_chain): Specify chain type in parameter and record it.
22334 (add_looparound_copies): Skip for store-stores chain.
22335 (determine_roots_comp): Compute type of chain and pass it to
22337 (initialize_root_vars_store_elim_2): New function.
22338 (finalize_eliminated_stores): New function.
22339 (remove_stmt): Handle store for elimination.
22340 (execute_pred_commoning_chain): Execute predictive commoning on
22341 store-store chains.
22342 (determine_unroll_factor): Skip unroll for store-stores chain.
22343 (prepare_initializers_chain_store_elim): New function.
22344 (prepare_initializers_chain): Hanlde store-store chain.
22345 (prepare_finalizers_chain, prepare_finalizers): New function.
22346 (tree_predictive_commoning_loop): Return integer value indicating
22347 if loop is unrolled or lcssa form is corrupted.
22348 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22350 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22352 * tree-predcom.c (initialize_root): Delete.
22353 (execute_pred_commoning_chain): Initialize root vars and replace
22354 reference of non-combined chain directly, rather than call above
22357 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22359 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22360 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22362 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22364 * tree-predcom.c (struct chain): New field init_seq.
22365 (release_chain): Release init_seq.
22366 (prepare_initializers_chain): Record intialization stmts in above
22368 (insert_init_seqs): New function.
22369 (tree_predictive_commoning_loop): Call insert_init_seqs.
22371 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22373 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22375 2017-07-28 Richard Biener <rguenther@suse.de>
22377 * match.pd: Remove superfluous :c.
22378 * genmatch.c (simplify::id): Add member.
22379 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22381 (current_id): New global.
22382 (dt_node::parent): Move from ...
22383 (dt_operand::parent): ... here. Add for_id member.
22384 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22385 (decision_tree::find_node): Relax order requirement when
22386 merging DT_TRUE nodes to ones inbetween the current simplify
22387 and the one we try to merge with. Add diagnostic whenever
22388 we need to enforce pattern order by not merging.
22389 (decision_tree::insert): Set current_id.
22390 (decision_tree::print_node): Dump parent node and for_id.
22391 (parser::last_id): Add member.
22392 (parser::push_simplify): Assign unique id.
22393 (parser::parser): Initialize last_id.
22395 2017-07-28 Martin Liska <mliska@suse.cz>
22398 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22399 gimple_build_debug_bind.
22401 2017-07-28 Richard Biener <rguenther@suse.de>
22403 PR tree-optimization/81502
22404 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22406 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22408 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22409 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22410 for BIT_FIELD_REF args.
22411 * fold-const.c (make_bit_field_ref): Likewise.
22412 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22414 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22417 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22418 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22419 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22420 Or it into SANITIZER_UNDEFINED.
22421 * ubsan.c: Include gimple-fold.h and varasm.h.
22422 (ubsan_expand_ptr_ifn): New function.
22423 (instrument_pointer_overflow): New function.
22424 (maybe_instrument_pointer_overflow): New function.
22425 (instrument_object_size): Formatting fix.
22426 (pass_ubsan::execute): Call instrument_pointer_overflow
22427 and maybe_instrument_pointer_overflow.
22428 * internal-fn.c (expand_UBSAN_PTR): New function.
22429 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22430 * sanitizer.def (__ubsan_handle_pointer_overflow,
22431 __ubsan_handle_pointer_overflow_abort): New builtins.
22432 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22433 * internal-fn.def (UBSAN_PTR): New internal function.
22434 * opts.c (sanitizer_opts): Add pointer-overflow.
22435 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22436 * fold-const.c (build_range_check): Compute pointer range check in
22437 integral type if pointer arithmetics would be needed. Formatting
22440 2017-07-28 Martin Liska <mliska@suse.cz>
22443 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22444 parameters that are of a variable-length.
22446 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22448 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22450 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22451 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22452 (CRT_CALL_STATIC_FUNCTION): Likewise.
22453 (ASM_DEFAULT_SPEC): New define.
22454 (ASM_SPEC32): Likewise.
22455 (ASM_SPEC64): Likewise.
22456 (ASM_SPEC_COMMON): Likewise.
22457 (ASM_SPEC): Likewise.
22458 (INVALID_64BIT): Likewise.
22459 (LINK_OS_DEFAULT_SPEC): Likewise.
22460 (LINK_OS_SPEC32): Likewise.
22461 (LINK_OS_SPEC64): Likewise.
22462 (POWERPC_LINUX): Likewise.
22463 (PTRDIFF_TYPE): Likewise.
22464 (RESTORE_FP_PREFIX): Likewise.
22465 (RESTORE_FP_SUFFIX): Likewise.
22466 (SAVE_FP_PREFIX): Likewise.
22467 (SAVE_FP_SUFFIX): Likewise.
22468 (SIZE_TYPE): Likewise.
22469 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22470 (TARGET_64BIT): Likewise.
22471 (TARGET_64BIT): Likewise.
22472 (TARGET_AIX): Likewise.
22473 (WCHAR_TYPE_SIZE): Likewise.
22474 (WCHAR_TYPE): Undefine.
22475 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22476 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22477 (CPP_OS_RTEMS_SPEC): Delete.
22478 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22479 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22481 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22483 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22485 PR tree-optimization/81578
22486 * tree-parloops.c (build_new_reduction): Bail out if
22487 reduction_code isn't one of the standard OpenMP reductions.
22488 Move the details printing after that decision.
22490 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22492 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22493 related to reload_in_progress.
22494 (splat_input_operand): Likewise.
22495 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22497 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22499 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22500 (TARGET_INSTANTIATE_DECLS): Likewise.
22501 (legitimate_indexed_address_p): Delete reload_in_progress code.
22502 (rs6000_debug_legitimate_address_p): Likewise.
22503 (rs6000_eliminate_indexed_memrefs): Likewise.
22504 (rs6000_emit_le_vsx_store): Likewise.
22505 (rs6000_emit_move_si_sf_subreg): Likewise.
22506 (rs6000_emit_move): Likewise.
22507 (register_to_reg_type): Likewise.
22508 (rs6000_pre_atomic_barrier): Likewise.
22509 (rs6000_machopic_legitimize_pic_address): Likewise.
22510 (rs6000_allocate_stack_temp): Likewise.
22511 (rs6000_address_for_fpconvert): Likewise.
22512 (rs6000_address_for_altivec): Likewise.
22513 (rs6000_secondary_memory_needed_rtx): Delete function.
22514 (rs6000_check_sdmode): Likewise.
22515 (rs6000_alloc_sdmode_stack_slot): Likewise.
22516 (rs6000_instantiate_decls): Likewise.
22517 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22518 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22519 Delete reload_in_progress.
22520 (*vec_reload_and_plus_<mptrsize>): Likewise.
22521 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22522 (vsx_div_v2di): Likewise.
22523 (vsx_udiv_v2di): Likewise.
22525 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22527 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22528 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22529 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22530 (rs6000_debug_reg_global): Delete print of LRA status.
22531 (rs6000_option_override_internal): Delete dead LRA related code.
22532 (rs6000_lra_p): Delete function.
22533 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22535 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22537 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22538 * config/riscv/rtems.h: New file.
22540 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22541 Sudakshina Das <sudi.das@arm.com>
22543 * config/aarch64/aarch64.md
22544 (define_split for and<mode>3nr_compare): Move
22545 non aarch64_logical_operand to a register.
22546 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22547 register immediate operand to a register.
22548 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22550 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22552 PR middle-end/81564
22553 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22555 2017-07-27 Richard Biener <rguenther@suse.de>
22557 PR tree-optimization/81573
22558 PR tree-optimization/81494
22559 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22560 multi defuse cycle case.
22562 2017-07-27 Richard Biener <rguenther@suse.de>
22564 PR tree-optimization/81571
22565 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22568 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22570 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22571 earlier and only if MASK_FPU is set. Adjust formatting.
22573 2017-07-27 Martin Liska <mliska@suse.cz>
22575 * opt-functions.awk: Add validation of value of Init.
22576 * optc-gen.awk: Pass new argument.
22578 2017-07-27 Martin Liska <mliska@suse.cz>
22580 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22581 Fix wrong condition.
22583 2017-07-27 Martin Liska <mliska@suse.cz>
22585 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22586 BBs and edges seen by autoFDO.
22588 2017-07-27 Richard Biener <rguenther@suse.de>
22590 PR tree-optimization/81502
22591 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22592 with incompatible but same sized type.
22593 (execute_update_addresses_taken): Likewise.
22595 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22597 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22598 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22600 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22603 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22604 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22605 Change s_operand to memory_operand.
22607 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22609 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22610 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22611 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22612 Emit instructions rather than returning an expression. Handle TFmode
22613 and KFmode by casting to TImode.
22614 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22615 (rs6000_emit_le_vsx_store): Likewise.
22616 * config/rs6000/vsx.md (VSX_TI): New iterator.
22617 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22618 (*vsx_le_undo_permute_<mode>): Likewise.
22619 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22620 emit the split sequence.
22621 (*vsx_le_perm_store_<mode>): Likewise.
22623 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22625 PR tree-optimization/81555
22626 PR tree-optimization/81556
22627 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22628 if true, force CHANGED for the recursive invocation.
22629 (reassociate_bb): Remember original length of ops array, pass
22630 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22632 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22633 attributes for noipa attribute. For naked attribute use
22634 lookup_attribute first before lookup_attribute_spec.
22635 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22637 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22638 for functions with noipa attribute.
22639 (cgraph_externally_visible_p): Return true for functions with noipa
22641 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22642 for functions with noipa attribute.
22643 * doc/extend.texi: Document noipa function attribute.
22644 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22645 also for functions with noipa attribute.
22646 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22648 2017-07-26 Andrew Pinski <apinski@cavium.com>
22650 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22651 vec_unalign_load_cost and vec_unalign_store_cost.
22653 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22655 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22656 -mvsx-small-integer option.
22657 (ISA_3_0_MASKS_IEEE): Likewise.
22658 (OTHER_VSX_VECTOR_MASKS): Likewise.
22659 (POWERPC_MASKS): Likewise.
22660 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22661 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22662 code, only testing for DImode being allowed in non-VSX floating
22664 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22665 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22667 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22668 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22669 TARGET_P8_VECTOR test.
22670 (rs6000_secondary_reload_simple_move): Likewise.
22671 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22672 since TARGET_P9_VECTOR was already tested.
22673 (rs6000_opt_masks): Remove -mvsx-small-integer.
22674 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22675 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22677 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22678 test for TARGET_VEXTRACTUB was used, and that uses
22680 (p9 extract splitter): Likewise.
22681 (vsx_extract_<mode>_di_p9): Likewise.
22682 (vsx_extract_<mode>_store_p9): Likewise.
22683 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22684 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22685 the elimination of TARGET_VSX_SMALL_INTEGER.
22686 (vsx_extract_<mode>_p8): Likewise.
22687 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22688 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22689 (vsx_set_<mode>_p9): Likewise.
22690 (vsx_set_v4sf_p9): Likewise.
22691 (vsx_set_v4sf_p9_zero): Likewise.
22692 (vsx_insert_extract_v4sf_p9): Likewise.
22693 (vsx_insert_extract_v4sf_p9_2): Likewise.
22694 * config/rs6000/rs6000.md (sign extend splitter): Change
22695 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22696 (floatsi<mode>2_lfiwax_mem): Likewise.
22697 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22698 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22699 since a test for TARGET_P9_VECTOR was used.
22700 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22701 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22702 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22703 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22704 TARGET_P8_VECTOR test.
22705 (fix_trunc<mode>si2_stfiwx): Likewise.
22706 (fix_trunc<mode>si2_internal): Likewise.
22707 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22708 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22710 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22711 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22712 TARGET_P8_VECTOR test.
22713 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22714 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22715 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22717 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22718 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22719 since a test for TARGET_P9_VECTOR was used.
22720 (splitter for loading small constants): Likewise.
22722 2017-07-26 Andrew Pinski <apinski@cavium.com>
22724 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22727 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22730 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22731 (fp_valid_at): Likewise.
22733 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22735 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22736 (qdf24xx_addrcost_table): Likewise.
22737 (cortexa57_tunings): Update to use generic_branch_cost.
22738 (cortexa72_tunings): Likewise.
22739 (cortexa73_tunings): Likewise.
22740 (qdf24xx_tunings): Likewise.
22742 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22744 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22745 (thunderx2t99_branch_cost): Likewise.
22746 (cortexa35_tunings): Update to use generic_branch_cost.
22747 (cortexa53_tunings): Likewise.
22748 (cortexa57_tunings): Likewise.
22749 (cortexa72_tunings): Likewise.
22750 (cortexa73_tunings): Likewise.
22751 (thunderx2t99_tunings): Likewise.
22753 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22755 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22756 (sparc_option_override): Honour MASK_FSMULD.
22757 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22758 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22759 * config/sparc/sparc.opt (mfsmuld): New option.
22760 * doc/invoke.texi (mfsmuld): Document option.
22762 2017-07-26 Marek Polacek <polacek@redhat.com>
22764 PR middle-end/70992
22765 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22767 2017-07-26 Richard Biener <rguenther@suse.de>
22769 * gimple-match-head.c (do_valueize): Return OP if valueize
22771 (get_def): New helper to get at the def stmt of a SSA name
22772 if valueize allows.
22773 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22774 do_valueize to get at the def stmt.
22775 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22777 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22779 PR middle-end/46932
22780 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22782 2017-07-26 Martin Liska <mliska@suse.cz>
22785 * function.c (expand_function_start): Make expansion of
22786 nonlocal_goto_save_area after parm_birth_insn.
22788 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22790 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22791 from all CPU target flags enable members.
22793 2017-07-26 Richard Biener <rguenther@suse.de>
22795 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22796 (decision_tree::gen): Make 'type' const.
22797 (write_predicate): Likewise.
22799 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22801 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22802 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22803 (rs6000_option_override_internal): Likewise.
22804 (rs6000_expand_vector_set): Likewise.
22805 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22806 (TARGET_UPPER_REGS_SF): Likewise.
22807 (TARGET_UPPER_REGS_DI): Likewise.
22808 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22809 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22810 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22811 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22812 (Splitters for DI constants in Altivec registers): Likewise.
22813 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22814 (vsx_set_v4sf_p9): Likewise.
22815 (vsx_set_v4sf_p9_zero): Likewise.
22816 (vsx_insert_extract_v4sf_p9): Likewise.
22817 (vsx_insert_extract_v4sf_p9_2): Likewise.
22819 2017-07-25 Carl Love <cel@us.ibm.com>
22821 * doc/extend.texi: Update the built-in documentation file for the
22822 existing built-in functions
22823 vector signed char vec_cnttz (vector signed char);
22824 vector unsigned char vec_cnttz (vector unsigned char);
22825 vector signed short vec_cnttz (vector signed short);
22826 vector unsigned short vec_cnttz (vector unsigned short);
22827 vector signed int vec_cnttz (vector signed int);
22828 vector unsigned int vec_cnttz (vector unsigned int);
22829 vector signed long long vec_cnttz (vector signed long long);
22830 vector unsigned long long vec_cnttz (vector unsigned long long);
22832 2017-07-25 Andrew Pinski <apinski@cavium.com>
22834 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22835 accesses where the use is for the first operand of a BIT_INSERT.
22837 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22840 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22841 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22843 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22845 * config/i386/gstabs.h: Delete.
22846 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22848 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22850 * config/i386/i386.c (ix86_decompose_address): Do not check for
22851 register RTX when looking at index_reg or base_reg.
22852 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22854 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22856 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22857 to update EH info here.
22859 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22861 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22863 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22865 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22867 2017-07-25 Torsten Duwe <duwe@suse.de>
22869 * common.opt: Introduce -fpatchable-function-entry
22870 command line option, and its variables function_entry_patch_area_size
22871 and function_entry_patch_area_start.
22872 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22873 including a two-value parser.
22874 * target.def (print_patchable_function_entry): New target hook.
22875 * targhooks.h (default_print_patchable_function_entry): New function.
22876 * targhooks.c (default_print_patchable_function_entry): Likewise.
22877 * toplev.c (process_options): Switch off IPA-RA if
22878 patchable function entries are being generated.
22879 * varasm.c (assemble_start_function): Look at the
22880 patchable-function-entry command line switch and current
22881 function attributes and maybe generate NOP instructions by
22882 calling the print_patchable_function_entry hook.
22883 * doc/extend.texi: Document patchable_function_entry attribute.
22884 * doc/invoke.texi: Document -fpatchable_function_entry
22885 command line option.
22886 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22888 * doc/tm.texi: Re-generate.
22890 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22893 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22894 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22896 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22898 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22900 2017-07-25 Richard Biener <rguenther@suse.de>
22902 PR tree-optimization/81455
22903 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22904 not walk in cycles when looking for guards.
22906 2017-07-25 Richard Biener <rguenther@suse.de>
22908 PR tree-optimization/81529
22909 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22910 when optimizing backedge uses.
22912 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22914 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22916 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22917 to dl_section_ref. On AIX, append an expression to subtract
22918 the size of the section length to dl_section_ref.
22920 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22922 * configure.ac: If any of the config.* scripts fail, exit 1.
22923 * configure: Regenerate.
22925 2017-07-25 Richard Biener <rguenther@suse.de>
22927 PR middle-end/81546
22928 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22929 of immediate uses, be more verbose on errors.
22931 2017-07-25 Richard Biener <rguenther@suse.de>
22933 PR tree-optimization/81510
22934 * tree-vect-loop.c (vect_is_simple_reduction): When the
22935 reduction stmt is not inside the loop bail out.
22937 2017-07-25 Richard Biener <rguenther@suse.de>
22939 PR tree-optimization/81303
22940 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22941 profitability check against LOOP_VINFO_NITERSM1.
22943 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22945 * domwalk.c (cmp_bb_postorder): Simplify.
22946 (sort_bbs_postorder): New function. Use it...
22947 (dom_walker::walk): ...here to optimize common cases.
22949 2017-07-25 Martin Liska <mliska@suse.cz>
22952 * ipa-visibility.c (function_and_variable_visibility): Make the
22953 redirection just on target that supports aliasing.
22954 Fix GNU coding style.
22956 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22959 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22961 * config/arm/rtems.h: Likewise.
22962 * config/bfin/rtems.h: Likewise.
22963 * config/i386/rtemself.h: Likewise.
22964 * config/lm32/rtems.h: Likewise.
22965 * config/m32c/rtems.h: Likewise.
22966 * config/m68k/rtemself.h: Likewise.
22967 * config/microblaze/rtems.h: Likewise.
22968 * config/mips/rtems.h: Likewise.
22969 * config/moxie/rtems.h: Likewise.
22970 * config/nios2/rtems.h: Likewise.
22971 * config/powerpcspe/rtems.h: Likewise.
22972 * config/rs6000/rtems.h: Likewise.
22973 * config/rtems.h: Likewise.
22974 * config/sh/rtems.h: Likewise.
22975 * config/sh/rtemself.h: Likewise.
22976 * config/sparc/rtemself.h: Likewise.
22978 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22981 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22982 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22983 * tree-ssa-structalias.c (alias_get_name): Same.
22985 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22988 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22989 instructions if no du chain is found.
22991 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22993 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22995 2017-07-25 Richard Biener <rguenther@suse.de>
22997 PR middle-end/81505
22998 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23001 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23003 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23004 upper-regs options.
23005 (ISA_2_7_MASKS_SERVER): Likewise.
23006 (ISA_3_0_MASKS_IEEE): Likewise.
23007 (OTHER_P8_VECTOR_MASKS): Likewise.
23008 (OTHER_VSX_VECTOR_MASKS): Likewise.
23009 (POWERPC_MASKS): Likewise.
23010 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23011 duplicate list of options.
23012 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23013 explicit -mupper-regs options.
23014 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23015 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23016 alias for -mupper-regs-df.
23017 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23018 (rs6000_init_hard_regno_mode_ok): Likewise.
23019 (rs6000_option_override_internal): Likewise.
23020 (rs6000_opt_masks): Likewise.
23021 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23022 options in terms of whether -mvsx or -mpower8-vector was used.
23023 (TARGET_UPPER_REGS_DI): Likewise.
23024 (TARGET_UPPER_REGS_SF): Likewise.
23025 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23026 -mupper-regs-* options.
23028 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23030 * passes.c (emergency_dump_function): Print some empty lines and a
23031 header before the RTL dump.
23033 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23035 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23037 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23040 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23041 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23043 2017-07-24 Carl Love <cel@us.ibm.com>
23045 * config/rs6000/rs6000-c.c: Add support for built-in functions
23046 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23047 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23048 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23049 vec_extract_fp_from_shortl): Add defines for the two builtins.
23050 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23051 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23053 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23054 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23055 * doc/extend.texi: Update the built-in documentation file for the
23056 new built-in function.
23058 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23061 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23063 * doc/generic.texi: Likewise.
23064 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23065 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23067 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23069 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23070 (aarch64_mls_elt_merge<mode>): Likewise.
23072 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23074 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23075 having __cxa_atexit.
23077 2017-07-23 Michael Collison <michael.collison@arm.com>
23079 * config/arm/arm.c (arm_option_override): Deprecate
23080 use of -mstructure-size-boundary.
23081 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23082 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23084 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23087 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23088 Reduce cost estimate for direct moves.
23090 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23093 * config/i386/i386.c (ix86_option_override_internal): Disable
23094 BMI, BMI2 and TBM instructions for -m16.
23096 2017-07-21 Carl Love <cel@us.ibm.com>
23098 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23099 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23100 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23101 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23102 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23103 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23104 VMULOSW): New enum "unspec" values.
23105 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23106 altivec_vmulosw): New patterns.
23107 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23108 VMULOSW): Add definitions.
23110 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23112 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23113 (qdf24xx): Likewise.
23114 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23115 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23116 (AARCH64_FL_V8_1): Renumber.
23117 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23118 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23119 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23120 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23121 rdma to feature modifiers list.
23123 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23125 PR middle-end/56727
23126 * ipa-visibility (function_and_variable_visibility): Convert
23127 recursive PLT call to direct call if appropriate.
23129 2017-07-21 Andrew Pinski <apinski@cavium.com>
23131 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23132 operand 1 to see if the types precision matches.
23133 * fold-const.c (operand_equal_p): Likewise.
23135 2017-07-21 Richard Biener <rguenther@suse.de>
23137 PR tree-optimization/81303
23138 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23139 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23140 (vect_peeling_hash_get_lowest_cost): Adjust.
23141 (vect_enhance_data_refs_alignment): Likewise. Use
23142 vect_get_peeling_costs_all_drs to compute the penalty for no
23143 peeling to match up costs.
23145 2017-07-21 Richard Biener <rguenther@suse.de>
23147 PR tree-optimization/81500
23148 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23149 we didn't identify a reduction path.
23151 2017-07-21 Tom de Vries <tom@codesourcery.com>
23152 Cesar Philippidis <cesar@codesourcery.com>
23154 PR gcov-profile/81442
23155 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23158 2017-07-21 Tom de Vries <tom@codesourcery.com>
23161 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23163 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23164 nvptx_override_options_after_change.
23166 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23168 * dwarf2out.c (output_file_names): Avoid double testing for
23169 dwarf_version >= 5.
23171 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23173 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23175 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23177 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23179 (try_crossjump_to_edge): Do not punt on partitioned functions.
23181 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23183 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23184 Put all BBs reachable only via paths crossing cold region to cold
23186 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23188 2016-07-21 Richard Biener <rguenther@suse.de>
23190 PR tree-optimization/81303
23191 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23192 into account prologue and epilogue iterations when raising
23193 min_profitable_iters to sth at least covering one vector iteration.
23195 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23197 * config/arm/arm.c (arm_test_cpu_arch_dat):
23200 2017-07-20 Nathan Sidwell <nathan@acm.org>
23202 Remove TYPE_METHODS.
23203 * tree.h (TYPE_METHODS): Delete.
23204 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23205 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23206 (dbxout_type_methods): Scan TYPE_FIELDS.
23207 (dbxout_type): Don't check TYPE_METHODS here.
23208 * function.c (use_register_for_decl): Always ignore register for
23209 class types when not optimizing.
23210 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23211 * tree.c (free_lang_data_in_type): Stitch out member functions and
23212 templates from TYPE_FIELDS.
23213 (build_distinct_type_copy, verify_type_variant,
23214 verify_type): Member fns are on TYPE_FIELDS.
23215 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23216 * tree-pretty-print.c (dump_generic_node): Likewise.
23218 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23221 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23222 V2TImode and V4TImode.
23223 (ix86_expand_vector_extract): Likewise.
23224 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23225 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23226 (ssescalarmode): Handle V4TImode and V2TImode.
23227 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23228 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23229 (VEXTRACTI128_MODE): New mode iterator.
23230 (splitter for *vec_extractv?ti first element): New.
23231 (VEC_INIT_MODE): New mode iterator.
23232 (vec_init<mode>): Consolidate 3 expanders into one using
23233 VEC_INIT_MODE mode iterator.
23235 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23237 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23238 non_spilled_static_chain_regno_p.
23240 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23242 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23244 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23246 * bb-reorder.c (connect_traces): Allow copying of blocks within
23249 2017-07-20 Richard Biener <rguenther@suse.de>
23251 * gimple.h (gimple_phi_result): Add gphi * overload.
23252 (gimple_phi_result_ptr): Likewise.
23253 (gimple_phi_arg): Likewise. Adjust index assert to only
23254 allow actual argument accesses rather than all slots available
23256 (gimple_phi_arg_def): Add gphi * overload.
23257 * tree-phinodes.c (make_phi_node): Initialize only actual
23259 (resize_phi_node): Clear memory not covered by old node,
23260 do not initialize excess argument slots.
23261 (reserve_phi_args_for_new_edge): Initialize new argument slot
23264 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23266 PR tree-optimization/81388
23268 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23270 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23271 by removing computation of may_be_zero.
23273 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23274 Tom de Vries <tom@codesourcery.com>
23276 PR middle-end/81030
23277 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23278 when gimple level profile disagrees with what RTL expander did.
23280 2017-07-20 Richard Biener <rguenther@suse.de>
23282 PR tree-optimization/61171
23283 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23284 (vect_analyze_stmt): Add slp instance parameter.
23285 (vectorizable_reduction): Likewise.
23286 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23287 (vect_is_simple_reduction): Deal with chains not detected
23288 as SLP reduction chain, specifically not properly associated
23289 chains containing a mix of plus/minus.
23290 (get_reduction_op): Remove.
23291 (get_initial_defs_for_reduction): Simplify, pass in whether
23292 this is a reduction chain, pass in the SLP node for the PHIs.
23293 (vect_create_epilog_for_reduction): Get the SLP instance as
23295 (vectorizable_reduction): Get the SLP instance as arg.
23296 During analysis remember the SLP node with the PHIs in the
23297 instance. Simplify getting at the vectorized reduction PHIs.
23298 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23299 through SLP instance.
23300 (vect_slp_analyze_operations): Likewise.
23301 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23302 (vect_transform_stmt): Likewise.
23304 2017-07-20 Tom de Vries <tom@codesourcery.com>
23306 PR tree-optimization/81489
23307 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23308 read of phi arg location to before loop that modifies phi.
23310 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23312 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23315 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23317 PR middle-end/81331
23318 * except.c (execute): Fix ordering issue.
23320 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23322 PR rtl-optimization/81423
23323 * combine.c (make_compound_operation_int): Don't try to optimize
23324 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23326 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23328 PR rtl-optimization/81423
23329 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23330 with a constant that is -1 in the truncated to mode.
23332 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23334 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23335 (determine_unlikely_bbs): ... here.
23336 * predict.h (propagate_unlikely_bbs_forward): Declare.
23337 * cfgexpand.c (pass_expand::execute): Use it.
23338 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23340 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23341 propagate_unlikely_bbs_forward.
23343 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23345 PR middle-end/81331
23346 * except.c (maybe_add_nop_after_section_switch): New function.
23349 2017-07-19 Tom de Vries <tom@codesourcery.com>
23351 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23353 2017-07-19 Tom de Vries <tom@codesourcery.com>
23355 * gimple.h (gimple_phi_arg): Make assert more strict.
23357 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23359 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23360 * config/rs6000/mmintrin.h: New file.
23361 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23363 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23365 PR tree-optimization/81346
23366 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23368 2017-07-19 Tom de Vries <tom@codesourcery.com>
23370 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23372 2017-07-19 Tom de Vries <tom@codesourcery.com>
23374 * config/nvptx/nvptx-modes.def: Add V2DImode.
23375 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23376 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23377 (nvptx_output_mov_insn): Handle lack of mov.b128.
23378 (nvptx_print_operand): Handle 'H' and 'L' codes.
23379 (nvptx_vector_mode_supported): Allow V2DImode.
23380 (nvptx_preferred_simd_mode): New function.
23381 (nvptx_data_alignment): New function.
23382 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23383 nvptx_preferred_simd_mode.
23384 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23386 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23388 2017-07-19 Tom de Vries <tom@codesourcery.com>
23390 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23391 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23392 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23393 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23394 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23395 (mov<VECIM>_insn): New define_insn.
23396 (define_expand "mov<VECIM>): New define_expand.
23398 2017-07-19 Tom de Vries <tom@codesourcery.com>
23400 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23402 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23404 PR tree-optimization/81346
23405 * fold-const.h (fold_div_compare, range_check_type): Declare.
23406 * fold-const.c (range_check_type): New function.
23407 (build_range_check): Use range_check_type.
23408 (fold_div_compare): No longer static, rewritten into
23409 a match.pd helper function.
23410 (fold_comparison): Don't call fold_div_compare here.
23411 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23412 as helper function.
23414 2017-07-19 Nathan Sidwell <nathan@acm.org>
23416 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23417 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23418 * tree.c (find_decls_types_r, verify_type): Use
23419 TYPE_{MIN,MAX}_VALUE_RAW.
23420 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23421 (hash_tree): Likewise.
23422 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23424 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23427 2017-07-18 Tom de Vries <tom@codesourcery.com>
23429 PR middle-end/81464
23430 * omp-expand.c (expand_omp_for_static_chunk): Handle
23431 equal-argument loop exit phi.
23433 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23436 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23437 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23438 operand 2 predicate.
23439 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23440 operand 2 predicate.
23441 (ror,rol -> rorx splitters): Use const_int_operand as
23442 operand 2 predicate.
23444 2017-06-18 Richard Biener <rguenther@suse.de>
23446 PR tree-optimization/81410
23447 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23448 the gap in the ! slp_perm SLP case after each group.
23450 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23452 PR middle-end/81463
23453 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23456 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23458 PR middle-end/81462
23459 * predict.c (set_even_probabilities): Cleanup; do not affect
23460 probabilities that are already known.
23461 (combine_predictions_for_bb): Call even when count is set.
23463 2017-07-18 Nathan Sidwell <nathan@acm.org>
23465 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23468 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23471 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23472 optimization for loop niter analysis.
23474 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23477 * config/avr/avr.c (avr_optimize_casesi): Don't use
23478 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23480 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23482 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23483 body_cost_vec from _vect_peel_extended_info.
23484 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23485 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23488 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23490 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23492 2017-07-18 Richard Biener <rguenther@suse.de>
23494 PR tree-optimization/80620
23495 PR tree-optimization/81403
23496 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23497 info when re-using a VN table entry.
23499 2017-07-18 Richard Biener <rguenther@suse.de>
23501 PR tree-optimization/81418
23502 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23503 vectype_in. Verify that with lane-reducing reduction operations
23504 we have a single def-use cycle.
23506 2017-07-17 Carl Love <cel@us.ibm.com>
23508 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23510 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23511 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23512 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23513 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23514 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23515 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23516 VMULOSW): New enum "unspec" values.
23517 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23518 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23519 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23520 altivec_vmulosw): New patterns.
23521 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23522 VMULOSW): Add definitions.
23524 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23526 * config/alpha/alpha.c: Include predict.h.
23528 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23530 * tree-vrp.c (compare_assert_loc): Fix comparison function
23531 to return predictable results.
23533 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23535 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23537 (subdi3): Likewise.
23538 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23539 * doc/invoke.texi (mexpand-adddi): Update text.
23541 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23543 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23544 that also clobbers the CC register. The old expand code is moved
23546 (*arc_clzsi2): ... here.
23547 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23548 the CC register. The old expand code is moved to ...
23549 (arc_ctzsi2): ... here.
23551 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23553 * config/arc/arc.opt (mindexed-loads): Use initial value
23554 TARGET_INDEXED_LOADS_DEFAULT.
23555 (mauto-modify-reg): Use initial value
23556 TARGET_AUTO_MODIFY_REG_DEFAULT.
23557 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23558 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23559 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23560 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23562 2017-07-17 Martin Liska <mliska@suse.cz>
23565 * opts.c (finish_options): Do not allow -fgnu-tm
23566 w/ -fsanitize={kernel-,}address. Say sorry.
23568 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23571 * tree-loop-distribution.c (classify_partition): Only assert on
23572 numer of iterations.
23573 (merge_dep_scc_partitions): Delete prameter. Update function call.
23574 (distribute_loop): Remove code handling loop with unknown niters.
23575 (pass_loop_distribution::execute): Skip loop with unknown niters.
23577 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23580 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23581 function sort_partitions_by_post_order.
23583 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23585 PR tree-optimization/81374
23586 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23587 the max index of basic blocks, rather than number of basic blocks.
23589 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23591 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23593 (arc_legitimate_pic_operand_p): Likewise.
23594 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23596 (arc_needs_pcl_p): Likewise.
23597 (arc_legitimate_pc_offset_p): Likewise.
23598 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23599 function is also used in constrains.md.
23600 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23601 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23602 PLUS. Only return true/false in known cases, otherwise assert.
23603 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23604 is already called in arc_legitimate_constant_p.
23605 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23607 (LEGITIMATE_PIC_OPERAND_P): Use
23608 arc_raw_symbolic_reference_mentioned_p function.
23609 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23614 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23615 Andrew Burgess <andrew.burgess@embecosm.com>
23617 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23618 (arc_return_address_register): New function.
23619 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23620 (arc_handle_fndecl_attribute): Add naked attribute.
23621 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23622 (TARGET_WARN_FUNC_RETURN): Likewise.
23623 (arc_allocate_stack_slots_for_args): New function.
23624 (arc_warn_func_return): Likewise.
23625 (machine_function): Change type fn_type.
23626 (arc_compute_function_type): Consider new naked function type,
23627 change function return type.
23628 (arc_must_save_register): Adapt to handle new
23629 arc_compute_function_type's return type.
23630 (arc_expand_prologue): Likewise.
23631 (arc_expand_epilogue): Likewise.
23632 (arc_return_address_regs): Delete.
23633 (arc_return_address_register): New function.
23634 (arc_epilogue_uses): Use above function.
23635 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23636 (arc_function_type): Change encoding, add naked type.
23637 (ARC_INTERRUPT_P): Change to handle the new encoding.
23638 (ARC_FAST_INTERRUPT_P): Likewise.
23639 (ARC_NORMAL_P): Define.
23640 (ARC_NAKED_P): Likewise.
23641 (arc_compute_function_type): Delete prototype.
23642 * config/arc/arc.md (in_ret_delay_slot): Use
23643 arc_return_address_register function.
23644 (simple_return): Likewise.
23645 (p_return_i): Likewise.
23647 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23649 PR tree-optimization/81428
23650 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23651 can't be built for those types.
23653 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23655 Remove stuff dead since r239246.
23657 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23658 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23659 (avr_inform_devices): Remove dead stuff.
23661 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23663 * config/arm/arm_neon.h: Fix softp typo.
23665 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23667 PR tree-optimization/81365
23668 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23669 aggregate moves onto bb predecessor edges, make sure there are no
23670 loads that could alias the lhs in between the start of bb and the
23673 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23676 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23677 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23678 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23680 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23682 PR tree-optimization/81396
23683 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23684 (init_symbolic_number): Initialize it to 1.
23685 (perform_symbolic_merge): Add n_ops from both operands into the new
23687 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23688 without base_addr as useless if they need more than one operation.
23689 (bswap_replace): Handle !bswap case for NULL base_addr.
23691 2017-07-17 Tom de Vries <tom@codesourcery.com>
23694 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23697 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23699 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23700 conditional builtin define __FIX_LEON3FT_B2BST.
23702 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23704 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23705 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23708 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23710 PR rtl-optimization/81424
23711 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23712 to remove potential trapping from operands if -fnon-call-exceptions.
23714 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23716 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23717 profile_proability for scalling.
23718 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23720 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23722 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23724 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23726 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23727 fixpoint arithmetics.
23729 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23731 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23732 fixpoint arithmetics.
23734 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23736 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23737 fixpoint arithmetics.
23739 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23741 * profile-count.h (profile_probability::from_reg_br_prob_note,
23742 profile_probability::to_reg_br_prob_note): New functions.
23743 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23744 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23745 * predict.c (probability_reliable_p): Update.
23746 (edge_probability_reliable_p): Update.
23747 (br_prob_note_reliable_p): Update.
23748 (invert_br_probabilities): Update.
23749 (add_reg_br_prob_note): New function.
23750 (combine_predictions_for_insn): Update.
23751 * asan.c (asan_clear_shadow): Update.
23752 * cfgbuild.c (compute_outgoing_frequencies): Update.
23753 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23754 (update_br_prob_note): Update.
23755 (rtl_verify_edges): Update.
23756 (purge_dead_edges): Update.
23757 (fixup_reorder_chain): Update.
23758 * emit-rtl.c (try_split): Update.
23759 * ifcvt.c (cond_exec_process_insns): Update.
23760 (cond_exec_process_if_block): Update.
23761 (dead_or_predicable): Update.
23762 * internal-fn.c (expand_addsub_overflow): Update.
23763 (expand_neg_overflow): Update.
23764 (expand_mul_overflow): Update.
23765 * loop-doloop.c (doloop_modify): Update.
23766 * loop-unroll.c (compare_and_jump_seq): Update.
23767 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23768 * predict.h: Update.
23769 * reorg.c (mostly_true_jump): Update.
23771 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23772 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23773 * config/arc/arc.c: (emit_unlikely_jump): Update.
23774 * config/arm/arm.c: (emit_unlikely_jump): Update.
23775 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23776 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23777 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23778 (ix86_print_operand): Update.
23779 (ix86_split_fp_branch): Update.
23780 (predict_jump): Update.
23781 * config/ia64/ia64.c (ia64_print_operand): Update.
23782 * config/mmix/mmix.c (mmix_print_operand): Update.
23783 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23784 (rs6000_expand_split_stack_prologue): Update.
23785 * config/rs6000/rs6000.c: Update.
23786 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23787 (s390_expand_vec_movstr): Update.
23788 (s390_expand_cs_tdsi): Update.
23789 (s390_expand_split_stack_prologue): Update.
23790 * config/sh/sh.c (sh_print_operand): Update.
23791 (expand_cbranchsi4): Update.
23792 (expand_cbranchdi4): Update.
23793 * config/sparc/sparc.c (output_v9branch): Update.
23794 * config/spu/spu.c (get_branch_target): Update.
23795 (ea_load_store_inline): Update.
23796 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23797 * config/tilepro/tilepro.c: Update.
23799 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23801 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23802 (gimplify_save_expr): Likewise.
23804 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23806 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23808 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23810 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23812 * timevar.def (TV_IPA_FNSUMMARY): Define.
23814 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23816 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23817 to back store errata sensitive sequence from being generated.
23818 (sqrtdf2_fix): Likewise.
23820 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23822 * tree-ssa-threadupdate.c (compute_path_counts,
23823 update_joiner_offpath_counts): Use profile_probability.
23825 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23828 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23830 * config/arm/arm-c.c (arm_cpu_builtins): Define
23831 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23833 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23835 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23836 array entries to represent __ieee128 versions of the
23837 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23838 scalar_extract_sig, and scalar_insert_exp built-in functions.
23839 (altivec_resolve_overloaded_builtin): Add special case handling
23840 for the __builtin_scalar_insert_exp function, as represented by
23841 the P9V_BUILTIN_VEC_VSIEDP constant.
23842 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23843 exponent support for __ieee128 argument.
23844 (VSESQP): Add scalar extract signature support for __ieee128
23846 (VSTDCNQP): Add scalar test negative support for __ieee128
23848 (VSIEQP): Add scalar insert exponent support for __int128 argument
23849 with __ieee128 result.
23850 (VSIEQPF): Add scalar insert exponent support for __ieee128
23851 argument with __ieee128 result.
23852 (VSTDCQP): Add scalar test data class support for __ieee128
23854 (VSTDCNQP): Add overload support for scalar test negative with
23855 __ieee128 argument.
23856 (VSTDCQP): Add overload support for scalar test data class
23857 __ieee128 argument.
23858 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23859 UNSPEC_VSX_SXSIGDP.
23860 (UNSPEC_VSX_SIEXPQP): New constant.
23861 (xsxexpqp): New insn for VSX scalar extract exponent quad
23863 (xsxsigqp): New insn for VSX scalar extract significand quad
23865 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23866 precision with floating point argument.
23867 (xststdcqp): New expand for VSX scalar test data class quad
23869 (xststdcnegqp): New expand for VSX scalar test negative quad
23871 (xststdcqp): New insn to match expansions for VSX scalar test data
23872 class quad precision and VSX scalar test negative quad precision.
23873 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23874 special case operand checking to enforce that second operand of
23875 VSX scalar test data class with quad precision argument is a 7-bit
23877 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23878 prototypes and descriptions of __ieee128 versions of
23879 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23880 scalar_test_data_class, and scalar_test_neg built-in functions.
23882 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23884 PR tree-optimization/81162
23885 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23886 replace a negate with an add.
23888 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23890 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23892 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23894 * config/arm/arm-c.c (arm_cpu_builtins): Define
23895 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23897 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23899 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23900 (armv8-r): Set ARM Cortex-R52 as default CPU.
23901 * config/arm/arm-tables.opt: Regenerate.
23902 * config/arm/arm-tune.md: Regenerate.
23903 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23905 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23906 extension for -mcpu=cortex-r52.
23908 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23910 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23911 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23912 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23913 (fp-armv8): Define it as FP_ARMv8 only.
23914 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23915 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23917 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23918 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23919 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23920 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23921 than TARGET_FPU_ARMV8.
23922 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23923 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23924 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23926 * config/arm/neon.md (neon_vrint): Likewise.
23927 (neon_vcvt): Likewise.
23928 (neon_<fmaxmin_op><mode>): Likewise.
23929 (<fmaxmin><mode>3): Likewise.
23930 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23931 * config/arm/predicates.md (arm_cond_move_operator): Check against
23932 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23934 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23936 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23937 to top of function.
23939 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23941 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23942 loop in comment with memset.
23944 2017-07-14 Martin Liska <mliska@suse.cz>
23946 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23947 * dwarf2out.c (is_java): Remove the function.
23948 (output_pubname): Remove usage of the function.
23949 (lower_bound_default): Remove usage of DW_LANG_Java.
23950 (gen_compile_unit_die): Likewise.
23951 * gcc.c: Remove compiler defaults for .java and .zip files.
23952 * gimple-expr.c (remove_suffix): Change as there's no longer
23953 extension than 4-letter one.
23954 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23955 (gimplify_save_expr): Likewise.
23956 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23957 as it's possible even for other languages than Java.
23958 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23959 * lto-opts.c (lto_write_options): Remove reference to Java.
23960 * opts.c (strip_off_ending): Update file extension handling.
23961 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23962 * tree-eh.c (lower_resx): Likewise.
23963 * tree.c (free_lang_data_in_type): Remove dead code.
23964 (find_decls_types_r): Likewise.
23965 (build_common_builtin_nodes): Remove Java from a comment.
23966 (verify_type): Remove dead code.
23967 * varasm.c (assemble_external): Remove Java from a comment.
23969 2017-07-14 Martin Liska <mliska@suse.cz>
23971 * opts.c (finish_options): Add quotes.
23972 (common_handle_option): Likewise.
23974 2017-07-14 Martin Liska <mliska@suse.cz>
23976 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23977 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23978 Remove N_SO_PASCAL.
23979 * dwarf2out.c (lower_bound_default): Do not handle
23981 (gen_compile_unit_die): Likewise.
23982 * gcc.c: Remove default extension binding for GNU Pascal.
23983 * stmt.c: Remove Pascal language from a comment.
23984 * xcoffout.c: Likewise.
23986 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23989 * diagnostic-show-locus.c (fixit_cmp): New function.
23990 (layout::layout): Sort m_fixit_hints.
23991 (column_range::column_range): Assert that the values are valid.
23992 (struct char_span): New struct.
23993 (correction::overwrite): New method.
23994 (struct source_line): New struct.
23995 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23996 calls in terms of classes source_line and char_span, and
23997 correction::overwrite.
23998 (selftest::test_overlapped_fixit_printing_2): New function.
23999 (selftest::diagnostic_show_locus_c_tests): Call it.
24001 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24003 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24004 early if there is no lhs.
24006 2017-07-13 Martin Liska <mliska@suse.cz>
24008 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24009 (gen_reference_type_die): Likewise.
24010 * stor-layout.c: Remove Pascal-related comment.
24012 2017-07-13 Martin Liska <mliska@suse.cz>
24014 * opts.c (finish_options): Add quotes to error messages.
24015 (parse_sanitizer_options): Likewise.
24017 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24019 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24021 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24023 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24025 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24027 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24029 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24031 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24034 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24035 provides the hardware capability bits, define the macro
24036 __BUILTIN_CPU_SUPPORTS__.
24037 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24038 if GLIBC does not provide the hardware capability bits. Add a
24039 gcc_unreachable call if the built-in cpu function is neither
24040 __builtin_cpu_is nor __builtin_cpu_supports.
24041 (rs6000_get_function_versions_dispatcher): Change the warning
24042 that an old GLIBC is used which does not export the capability
24043 bits to be an error.
24044 * doc/extend.texi (target_clones attribute): Document the
24045 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24046 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24047 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24048 the macros defined by GCC if the newer GLIBC is available.
24050 2017-07-12 Jeff Law <law@redhat.com>
24052 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24053 remaining includes slightly.
24054 * config/riscv/riscv-builtins.c: Include profile-count.h.
24056 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24059 * config/avr/avr.c (avr_set_current_function): In diagnostic
24060 messages: Quote keywords and (parts of) identifiers.
24061 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24064 2017-07-12 Carl Love <cel@us.ibm.com>
24066 * config/rs6000/rs6000-c.c: Add support for built-in functions
24067 vector bool char vec_revb (vector bool char);
24068 vector bool short vec_revb (vector short char);
24069 vector bool int vec_revb (vector bool int);
24070 vector bool long long vec_revb (vector bool long long);
24071 * doc/extend.texi: Update the built-in documentation file for the
24072 new built-in functions.
24074 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24076 * config/s390/s390.md: Remove movcc splitter.
24078 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24080 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24081 load/store on condition.
24083 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24086 * config/avr/avr.c (avr_encode_section_info)
24087 [progmem && !TREE_READONLY]: Error if progmem object needs
24090 2017-07-11 Michael Collison <michael.collison@arm.com>
24092 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24095 2017-07-11 Carl Love <cel@us.ibm.com>
24097 * config/rs6000/rs6000-c.c: Add support for builtins
24098 vector unsigned int vec_parity_lsbb (vector signed int);
24099 vector unsigned int vec_parity_lsbb (vector unsigned int);
24100 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24101 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24102 vector unsigned long long vec_parity_lsbb (vector signed long long);
24103 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24104 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24105 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24106 * doc/extend.texi: Update the built-in documentation file for the
24107 new built-in functions.
24109 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24111 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24112 (layout::m_primary_loc): New field.
24113 (layout::layout): Initialize new field. Move location filtering
24114 logic from here to...
24115 (layout::maybe_add_location_range): ...this new method. Add
24116 support for filtering to just the lines already specified by other
24118 (layout::will_show_line_p): New method.
24119 (gcc_rich_location::add_location_if_nearby): New method.
24120 (selftest::test_add_location_if_nearby): New test function.
24121 (selftest::diagnostic_show_locus_c_tests): Call it.
24122 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24125 2017-07-11 Tom de Vries <tom@codesourcery.com>
24127 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24128 (bb_first_real_insn): New function.
24129 (nvptx_single): Add extra initialization of broadcasted condition
24132 2017-07-11 Nathan Sidwell <nathan@acm.org>
24134 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24136 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24138 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24139 Binutils doc as TEXI will mess them up.
24140 * doc/invoke.texi (AVR Options): Same here.
24142 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24144 * config/sparc/sparc.opt (mfix-ut700): New option.
24145 (mfix-gr712rc): Likewise.
24146 (sparc_fix_b2bst): New variable.
24147 * doc/invoke.texi (SPARC options): Document them.
24148 (ARM options): Fix warnings.
24149 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24150 instructions to prevent sequences that can trigger the store-store
24151 errata for certain LEON3FT processors.
24152 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24153 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24154 * config/sparc/sparc.md (fix_b2bst): New attribute.
24155 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24157 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24160 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24162 (*rsqrtsf2_sse): Ditto.
24164 (div<mode>3): Macroize insn from divdf3 and divsf3
24165 using MODEF mode iterator.
24167 2017-07-10 Martin Sebor <msebor@redhat.com>
24169 PR tree-optimization/80397
24170 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24171 instead of testing for equality to INTEGER_TYPE.
24173 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24175 * config.gcc: Remove uclibc from arc target spec.
24177 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24179 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24181 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24184 * lto-wrapper.c (remove_option): New function.
24185 (merge_and_complain): Merge PIC/PIE options more realistically.
24187 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24189 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24193 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24194 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24195 * config.in: Regenerate.
24196 * configure: Regenerate.
24197 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24198 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24199 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24200 (TARGET_GASISR_PROLOGUES): ...target mask.
24201 * common/config/avr/avr-common.c
24202 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24203 Set -mgas-isr-prologues.
24204 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24205 INSERT_PASS_BEFORE for it.
24206 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24207 * config/avr/avr.c (avr_option_override)
24208 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24209 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24210 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24211 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24212 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24214 (make_avr_pass_pre_proep): New function.
24215 (emit_push_sfr) <treg>: Add argument to function and use it
24216 instead of TMP_REG.
24217 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24218 and set machine->gasisr.yes.
24219 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24220 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24221 __gcc_isr.n_pushed to .L__stack_usage.
24222 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24223 (avr_asm_final_postscan_insn): ...this new static function.
24224 * config/avr/avr.h (machine_function)
24225 <is_no_gccisr, use_L__stack_usage>: New fields.
24226 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24227 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24228 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24229 (gasisr, *gasisr): New expander and insn.
24230 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24231 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24232 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24234 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24236 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24239 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24241 Move jump-tables out of .text again.
24244 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24245 (ASM_OUTPUT_ADDR_VEC): New function.
24246 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24247 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24248 INSN_ADDRESSes as asm comment.
24249 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24250 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24251 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24252 * config/avr/avr.md (*tablejump): Adjust comment.
24253 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24254 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24256 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24257 (avr_output_addr_vec): New proto.
24258 (avr_log_t) <insn_addresses>: New field.
24260 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24263 * config/i386/i386.c (ix86_function_arg_advance): Set
24264 outgoing_args_on_stack to true if there are outgoing arguments
24266 (ix86_function_arg): Likewise.
24267 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24268 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24269 * config/i386/i386.h (machine_function): Add
24270 outgoing_args_on_stack.
24272 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24274 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24275 supporting pthreds.
24276 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24278 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24280 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24281 (REAL_H): Remove $(MACHMODE_H).
24282 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24284 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24285 $(MACHMODE_H) and double-int.h.
24286 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24288 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24291 2017-07-07 Andrew Pinski <apinski@cavium.com>
24293 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24294 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24296 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24298 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24299 Add warning if GCC was not configured to link against a GLIBC that
24300 exports the hardware capability bits.
24301 (make_resolver_func): Make resolver function private and not a
24302 COMDAT function. Create the name with clone_function_name instead
24303 of make_unique_name.
24306 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24307 correct operand in doing the split.
24309 2017-07-07 Carl Love <cel@us.ibm.com>
24311 * config/rs6000/rs6000-c: Add support for built-in function
24312 vector unsigned short vec_pack_to_short_fp32 (vector float,
24314 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24315 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24316 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24317 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24318 (convert_4f32_8i16): Add define_expand.
24319 * doc/extend.texi: Update the built-in documentation file for the
24320 new built-in function.
24322 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24324 * config/sparc/m8.md: New file.
24325 * config/sparc/sparc.md: Include m8.md.
24327 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24329 * config/sparc/sparc.opt: New option -mvis4b.
24330 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24331 (sparc_option_override): Handle VIS4B.
24332 (enum sparc_builtins): Define
24333 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24334 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24335 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24336 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24337 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24338 (check_constant_argument): New function.
24339 (sparc_vis_init_builtins): Define builtins
24340 __builtin_vis_dictunpack{8,16,32},
24341 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24342 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24343 __builtin_vis_fpcmpde{8,16,32}shl and
24344 __builtin_vis_fpcmpur{8,16,32}shl.
24345 (sparc_expand_builtin): Check that the constant operands to
24346 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24347 constant and in range.
24348 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24350 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24351 (SPARC_IMM5_P): Likewise.
24352 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24353 (enabled): Handle vis4b.
24354 (UNSPEC_DICTUNPACK): New unspec.
24355 (UNSPEC_FPCMPSHL): Likewise.
24356 (UNSPEC_FPUCMPSHL): Likewise.
24357 (UNSPEC_FPCMPDESHL): Likewise.
24358 (UNSPEC_FPCMPURSHL): Likewise.
24359 (cpu_feature): New CPU feature `vis4b'.
24360 (dictunpack{8,16,32}): New insns.
24361 (FPCSMODE): New mode iterator.
24362 (fpcscond): New code iterator.
24363 (fpcsucond): Likewise.
24364 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24365 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24366 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24367 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24368 * config/sparc/constraints.md: Define constraints `q' for unsigned
24369 2-bit integer constants and `t' for unsigned 5-bit integer
24371 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24373 (imm5_operand_dictunpack16): Likewise.
24374 (imm5_operand_dictunpack32): Likewise.
24375 (imm2_operand): Likewise.
24376 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24377 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24378 ditunpack* and fpcmp*shl builtins.
24380 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24382 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24383 * config.in: Add HAVE_AS_SPARC6 define.
24384 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24386 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24388 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24389 (CPP_CPU_SPEC): Handle m8.
24390 (ASM_CPU_SPEC): Likewise.
24391 * config/sparc/sparc-opts.h (enum processor_type): Add
24393 * config/sparc/sparc.c (m8_costs): New struct.
24394 (sparc_option_override): Handle TARGET_CPU_m8.
24395 (sparc32_initialize_trampoline): Likewise.
24396 (sparc64_initialize_trampoline): Likewise.
24397 (sparc_issue_rate): Likewise.
24398 (sparc_register_move_cost): Likewise.
24399 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24400 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24401 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24402 (CPP_CPU_SPEC): Handle M8.
24403 (ASM_CPU_SPEC): Likewise.
24404 (AS_M8_FLAG): Define.
24405 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24406 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24407 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24409 * configure: Regenerate.
24410 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24413 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24415 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24417 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24418 ("*movdi_insn_sp32"): Do not set v3pipe.
24419 ("*movsi_insn"): Likewise.
24420 ("*movdi_insn_sp64"): Likewise.
24421 ("*movsf_insn"): Likewise.
24422 ("*movdf_insn_sp32"): Likewise.
24423 ("*movdf_insn_sp64"): Likewise.
24424 ("*zero_extendsidi2_insn_sp64"): Likewise.
24425 ("*sign_extendsidi2_insn"): Likewise.
24426 ("*mov<VM32:mode>_insn"): Likewise.
24427 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24428 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24429 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24430 ("<vlop:code><VL:mode>3"): Likewise.
24431 ("*not_<vlop:code><VL:mode>3"): Likewise.
24432 ("*nand<VL:mode>_vis"): Likewise.
24433 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24434 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24435 ("one_cmpl<VL:mode>2"): Likewise.
24436 ("faligndata<VM64:mode>_vis"): Likewise.
24437 ("alignaddrsi_vis"): Likewise.
24438 ("alignaddrdi_vis"): Likweise.
24439 ("alignaddrlsi_vis"): Likewise.
24440 ("alignaddrldi_vis"): Likewise.
24441 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24442 ("bmaskdi_vis"): Likewise.
24443 ("bmasksi_vis"): Likewise.
24444 ("bshuffle<VM64:mode>_vis"): Likewise.
24445 ("cmask8<P:mode>_vis"): Likewise.
24446 ("cmask16<P:mode>_vis"): Likewise.
24447 ("cmask32<P:mode>_vis"): Likewise.
24448 ("pdistn<P:mode>_vis"): Likewise.
24449 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24451 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24453 * config/sparc/sparc.md ("subtype"): New insn attribute.
24454 ("*wrgsr_sp64"): Set insn subtype.
24455 ("*rdgsr_sp64"): Likewise.
24456 ("alignaddrsi_vis"): Likewise.
24457 ("alignaddrdi_vis"): Likewise.
24458 ("alignaddrlsi_vis"): Likewise.
24459 ("alignaddrldi_vis"): Likewise.
24460 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24461 ("fexpand_vis"): Likewise.
24462 ("fpmerge_vis"): Likewise.
24463 ("faligndata<VM64:mode>_vis"): Likewise.
24464 ("bshuffle<VM64:mode>_vis"): Likewise.
24465 ("cmask8<P:mode>_vis"): Likewise.
24466 ("cmask16<P:mode>_vis"): Likewise.
24467 ("cmask32<P:mode>_vis"): Likewise.
24468 ("fchksm16_vis"): Likewise.
24469 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24470 ("fmean16_vis"): Likewise.
24471 ("fp<plusminus_insn>64_vis"): Likewise.
24472 ("<plusminus_insn>v8qi3"): Likewise.
24473 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24474 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24475 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24476 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24477 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24478 ("*movqi_insn"): Likewise.
24479 ("*movhi_insn"): Likewise.
24480 ("*movsi_insn"): Likewise.
24481 ("movsi_pic_gotdata_op"): Likewise.
24482 ("*movdi_insn_sp32"): Likewise.
24483 ("*movdi_insn_sp64"): Likewise.
24484 ("movdi_pic_gotdata_op"): Likewise.
24485 ("*movsf_insn"): Likewise.
24486 ("*movdf_insn_sp32"): Likewise.
24487 ("*movdf_insn_sp64"): Likewise.
24488 ("*zero_extendhisi2_insn"): Likewise.
24489 ("*zero_extendqihi2_insn"): Likewise.
24490 ("*zero_extendqisi2_insn"): Likewise.
24491 ("*zero_extendqidi2_insn"): Likewise.
24492 ("*zero_extendhidi2_insn"): Likewise.
24493 ("*zero_extendsidi2_insn_sp64"): Likewise.
24494 ("ldfsr"): Likewise.
24495 ("prefetch_64"): Likewise.
24496 ("prefetch_32"): Likewise.
24497 ("tie_ld32"): Likewise.
24498 ("tie_ld64"): Likewise.
24499 ("*tldo_ldub_sp32"): Likewise.
24500 ("*tldo_ldub1_sp32"): Likewise.
24501 ("*tldo_ldub2_sp32"): Likewise.
24502 ("*tldo_ldub_sp64"): Likewise.
24503 ("*tldo_ldub1_sp64"): Likewise.
24504 ("*tldo_ldub2_sp64"): Likewise.
24505 ("*tldo_ldub3_sp64"): Likewise.
24506 ("*tldo_lduh_sp32"): Likewise.
24507 ("*tldo_lduh1_sp32"): Likewise.
24508 ("*tldo_lduh_sp64"): Likewise.
24509 ("*tldo_lduh1_sp64"): Likewise.
24510 ("*tldo_lduh2_sp64"): Likewise.
24511 ("*tldo_lduw_sp32"): Likewise.
24512 ("*tldo_lduw_sp64"): Likewise.
24513 ("*tldo_lduw1_sp64"): Likewise.
24514 ("*tldo_ldx_sp64"): Likewise.
24515 ("*mov<VM32:mode>_insn"): Likewise.
24516 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24517 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24519 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24521 * config/sparc/sparc.md ("type"): New insn type viscmp.
24522 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24524 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24525 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24526 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24527 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24529 ("n7_vis_logical_11cycle"): Likewise.
24530 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24531 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24532 * config/sparc/niagara.md ("niag_vis"): Likewise.
24533 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24534 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24536 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24538 * config/sparc/sparc.md: New instruction type `bmask'.
24539 (bmaskdi_vis): Use the `bmask' type.
24540 (bmasksi_vis): Likewise.
24541 * config/sparc/ultra3.md (us3_array): Likewise.
24542 * config/sparc/niagara7.md (n7_array): Likewise.
24543 * config/sparc/niagara4.md (n4_array): Likewise.
24544 * config/sparc/niagara2.md (niag2_vis): Likewise.
24545 (niag3_vis): Likewise.
24546 * config/sparc/niagara.md (niag_vis): Likewise.
24548 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24550 * ipa-comdats.c: Remove optimize check from gate.
24551 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24552 for functions not optimized.
24553 (ipa_fn_summary_read): Skip optimize check.
24554 (ipa_fn_summary_write): Likewise.
24555 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24557 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24559 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24560 (check_callers): Give up on caller which is not optimized.
24561 (inline_small_functions): Likewise.
24562 (ipa_inline): Do not give up when not optimizing.
24563 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24564 away unoptimizes cdtors.
24565 (whole_program_function_and_variable_visibility): Do
24566 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24567 * ipa.c (process_references): Do not check optimize.
24568 (symbol_table::remove_unreachable_nodes): Update optimize check.
24569 (set_writeonly_bit): Update optimize check.
24570 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24571 (pass_ipa_single_use::gate): Remove.
24573 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24575 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24576 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24577 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24578 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24579 permute_load, permute_store, adjust_extract, adjust_splat,
24580 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24581 replace_swap_with_copy, dump_swap_insn_table,
24582 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24583 recombine_lvx_pattern, recombine_stvx_pattern,
24584 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24585 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24586 to file rs6000-p8swap.c.
24587 * config/rs6000/rs6000-p8swap.c: New file.
24588 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24589 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24590 and rs6000*-*-* targets.
24592 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24594 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24596 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24598 * lto-wrapper.c (merge_and_complain): Do not merge
24599 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24600 fsigned_zeros, ftrapping_math, fwrapv.
24601 (append_compiler_options): Do not track these options.
24602 (append_linker_options): Likewie
24604 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24606 * cgraphunit.c (cgraph_node::finalize_function): When
24607 !flag_toplevel_reorde set no_reorder flag.
24608 (varpool_node::finalize_decl): Likewise.
24609 (symbol_table::compile): Drop no toplevel reorder path.
24611 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24613 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24614 edges; zero probability is not better than uninitialized.
24616 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24618 * asan.h (asan_sanitize_allocas_p): Declare.
24619 * asan.c (asan_sanitize_allocas_p): New function.
24620 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24621 (handle_builtin_alloca): Likewise.
24622 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24623 if !asan_sanitize_allocas_p.
24624 * params.def (asan-instrument-allocas): Add new option.
24625 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24626 * opts.c (common_handle_option): Disable allocas sanitization for
24629 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24631 * asan.c: Include gimple-fold.h.
24632 (get_last_alloca_addr): New function.
24633 (handle_builtin_stackrestore): Likewise.
24634 (handle_builtin_alloca): Likewise.
24635 (asan_emit_allocas_unpoison): Likewise.
24636 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24637 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24638 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24639 (instrument_builtin_call): Pass gimple iterator to
24640 get_mem_refs_of_builtin_call.
24641 (last_alloca_addr): New global.
24642 * asan.h (asan_emit_allocas_unpoison): Declare.
24643 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24644 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24645 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24646 if function calls alloca.
24647 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24648 * gimple-fold.h (replace_call_with_value): Declare.
24649 * internal-fn.c: Include asan.h.
24650 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24651 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24653 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24655 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24656 (C_SELFTEST_FLAGS): New.
24657 (CPP_SELFTEST_FLAGS): New.
24658 (SELFTEST_DEPS): New, from deps of s-selftest.
24659 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24660 (CPP_SELFTEST_DEPS): New.
24661 (selftest): Add dependency on s-selftest-c++.
24662 (s-selftest): Rename to...
24663 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24664 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24665 than SELFTEST_FLAGS.
24666 (selftest-gdb): Rename to...
24667 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24669 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24670 (selftest-valgrind): Rename to...
24671 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24673 (selftest-valgrind): Reintroduce as an alias for
24674 selftest-c-valgrind.
24675 (s-selftest-c++): New.
24676 (selftest-c++-gdb): New.
24677 (selftest-c++-valgrind): New.
24679 2017-07-06 Olivier Hainque <hainque@adacore.com>
24681 * gcc.c (process_command): When deciding if undefined variables
24682 should be ignored when processing specs, accept "gcc -v" as well.
24684 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24686 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24687 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24689 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24691 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24692 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24693 * config/arm/arm-tables.opt: Regenerate.
24694 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24696 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24698 2017-07-06 Carl Love <cel@us.ibm.com>
24700 * ChangeLog: Clean up from mid air collision
24702 2017-07-06 Carl Love <cel@us.ibm.com>
24704 * config/rs6000/rs6000-c.c: Add support for built-in functions
24705 vector signed int vec_subc (vector signed int, vector signed int);
24706 vector signed __int128 vec_subc (vector signed __int128,
24707 vector signed __int128);
24708 vector unsigned __int128 vec_subc (vector unsigned __int128,
24709 vector unsigned __int128);
24710 vector signed int vec_sube (vector signed int, vector signed int,
24711 vector signed int);
24712 vector unsigned int vec_sube (vector unsigned int,
24713 vector unsigned int,
24714 vector unsigned int);
24715 vector signed __int128 vec_sube (vector signed __int128,
24716 vector signed __int128,
24717 vector signed__int128);
24718 vector unsigned __int128 vec_sube (vector unsigned __int128,
24719 vector unsigned __int128,
24720 vector unsigned __int128);
24721 vector signed int vec_subec (vector signed int, vector signed int,
24722 vector signed int);
24723 vector unsigned int vec_subec (vector unsigned int,
24724 vector unsigned int,
24725 vector unsigned int);
24726 vector signed __int128 vec_subec (vector signed __int128,
24727 vector signed __int128,
24728 vector signed__int128);
24729 vector unsigned __int128 vec_subec (vector unsigned __int128,
24730 vector unsigned __int128,
24731 vector unsigned __int128);
24732 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24733 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24734 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24735 BU_ALTIVEC_OVERLOAD_X definitions.
24736 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24737 * doc/extend.texi: Update the built-in documentation file for the new
24738 built-in functions.
24740 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24743 * diagnostic-show-locus.c (layout::layout): Use start and finish
24744 spelling location for the start and finish of each range.
24745 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24746 Add unused aspect param.
24747 * input.c (expand_location_1): Add "aspect" param, and use it
24748 to access the correct part of the location.
24749 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24751 (expand_location_to_spelling_point): Likewise.
24752 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24753 param, and pass it to expand_location_1.
24755 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24757 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24758 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24759 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24760 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24761 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24762 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24763 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24764 _mm_maskz_getmant_ss): New intrinsics.
24765 (__builtin_ia32_getexpss128_mask): Changed to ...
24766 __builtin_ia32_getexpss128_round ... this.
24767 (__builtin_ia32_getexpsd128_mask): Changed to ...
24768 __builtin_ia32_getexpsd128_round ... this.
24769 * config/i386/i386-builtin-types.def
24770 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24771 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24772 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24773 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24774 __builtin_ia32_getmantss_mask_round): New builtins.
24775 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24776 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24777 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24778 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24779 * config/i386/sse.md
24780 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24781 avx512f_sgetexp<mode><mask_scalar_name>
24782 <round_saeonly_scalar_name> ... this.
24783 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24784 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24785 vgetexp<ssescalarmodesuffix>
24786 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24787 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24788 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24789 avx512f_vgetmant<mode><mask_scalar_name>
24790 <round_saeonly_scalar_name> ... this.
24791 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24792 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24793 vgetmant<ssescalarmodesuffix>
24794 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24795 %0<mask_scalar_operand4>, %1, %2
24796 <round_saeonly_scalar_mask_op4>, %3} ... this.
24797 * config/i386/subst.md (mask_scalar_operand4,
24798 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24799 round_saeonly_scalar_nimm_predicate): New subst attributes.
24801 2017-07-06 Julia Koval <julia.koval@intel.com>
24803 * config/i386/i386.c (ix86_erase_embedded_rounding):
24804 Remove code for old rounding pattern.
24806 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24808 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24810 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24812 * doc/sourcebuild.texi (Test Directives, Variants of
24813 dg-require-support): Add documentation for dg-require-stack-check.
24815 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24817 * config/i386/subst.md (mask_scalar, round_scalar,
24818 round_saeonly_scalar): New meta-templates.
24819 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24820 round_scalar_mask_operand3, round_scalar_mask_op3,
24821 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24822 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24823 round_saeonly_scalar_constraint,
24824 round_saeonly_scalar_prefix): New subst attribute.
24825 * config/i386/sse.md
24826 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24827 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24828 <round_scalar_name> ... this.
24829 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24830 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24831 <round_scalar_name> ... this.
24832 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24833 <sse>_vm<code><mode>3<mask_scalar_name>
24834 <round_saeonly_scalar_name> ... this.
24835 (v<plusminus_mnemonic><ssescalarmodesuffix>
24836 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24837 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24838 v<plusminus_mnemonic><ssescalarmodesuffix>
24839 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24840 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24841 (v<multdiv_mnemonic><ssescalarmodesuffix>
24842 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24843 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24844 v<multdiv_mnemonic><ssescalarmodesuffix>
24845 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24846 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24847 (v<maxmin_float><ssescalarmodesuffix>
24848 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24849 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24850 v<maxmin_float><ssescalarmodesuffix>
24851 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24852 %0<mask_scalar_operand3>, %1, %<iptr>2
24853 <round_saeonly_scalar_mask_op3>} ... this.
24855 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24857 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24858 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24860 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24861 Alan Hayward <alan.hayward@arm.com>
24862 David Sherwood <david.sherwood@arm.com>
24864 * combine.c (simplify_if_then_else): Remove "enum" before
24866 * compare-elim.c (can_eliminate_compare): Likewise.
24867 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24869 (aarch64_lookup_simd_builtin_type): Likewise.
24870 (aarch64_simd_builtin_type): Likewise.
24871 (aarch64_init_simd_builtin_types): Likewise.
24872 (aarch64_simd_expand_args): Likewise.
24873 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24875 (aarch64_reverse_mask): Likewise.
24876 (aarch64_simd_emit_reg_reg_move): Likewise.
24877 (aarch64_gen_adjusted_ldpstp): Likewise.
24878 (aarch64_ccmp_mode_to_code): Likewise.
24879 (aarch64_operands_ok_for_ldpstp): Likewise.
24880 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24881 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24883 (aarch64_min_divisions_for_recip_mul): Likewise.
24884 (aarch64_reassociation_width): Likewise.
24885 (aarch64_get_condition_code_1): Likewise.
24886 (aarch64_simd_emit_reg_reg_move): Likewise.
24887 (aarch64_simd_attr_length_rglist): Likewise.
24888 (aarch64_reverse_mask): Likewise.
24889 (aarch64_operands_ok_for_ldpstp): Likewise.
24890 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24891 (aarch64_gen_adjusted_ldpstp): Likewise.
24892 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24894 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24895 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24896 (arm_lookup_simd_builtin_type): Likewise.
24897 (arm_simd_builtin_type): Likewise.
24898 (arm_init_simd_builtin_types): Likewise.
24899 (arm_expand_builtin_args): Likewise.
24900 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24901 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24902 (ft32_setup_incoming_varargs): Likewise.
24903 (ft32_function_arg): Likewise.
24904 (ft32_function_arg_advance): Likewise.
24905 (ft32_pass_by_reference): Likewise.
24906 (ft32_arg_partial_bytes): Likewise.
24907 (ft32_valid_pointer_mode): Likewise.
24908 (ft32_addr_space_pointer_mode): Likewise.
24909 (ft32_addr_space_legitimate_address_p): Likewise.
24910 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24912 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24913 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24914 (iamcu_alignment): Likewise.
24915 (canonicalize_vector_int_perm): Likewise.
24916 (ix86_noce_conversion_profitable_p): Likewise.
24917 (ix86_mpx_bound_mode): Likewise.
24918 (ix86_operands_ok_for_move_multiple): Likewise.
24919 * config/microblaze/microblaze-protos.h
24920 (microblaze_expand_conditional_branch_reg): Likewise.
24921 * config/microblaze/microblaze.c
24922 (microblaze_expand_conditional_branch_reg): Likewise.
24923 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24925 (rs6000_reassociation_width): Likewise.
24926 (rs6000_invalid_binary_op): Likewise.
24927 (fusion_p9_p): Likewise.
24928 (emit_fusion_p9_load): Likewise.
24929 (emit_fusion_p9_store): Likewise.
24930 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24932 (riscv_hard_regno_mode_ok_p): Likewise.
24933 (riscv_address_insns): Likewise.
24934 (riscv_split_symbol): Likewise.
24935 (riscv_legitimize_move): Likewise.
24936 (riscv_function_value): Likewise.
24937 (riscv_hard_regno_nregs): Likewise.
24938 (riscv_expand_builtin): Likewise.
24939 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24940 (riscv_build_integer): Likewise.
24941 (riscv_split_integer): Likewise.
24942 (riscv_legitimate_constant_p): Likewise.
24943 (riscv_cannot_force_const_mem): Likewise.
24944 (riscv_regno_mode_ok_for_base_p): Likewise.
24945 (riscv_valid_base_register_p): Likewise.
24946 (riscv_valid_offset_p): Likewise.
24947 (riscv_valid_lo_sum_p): Likewise.
24948 (riscv_classify_address): Likewise.
24949 (riscv_legitimate_address_p): Likewise.
24950 (riscv_address_insns): Likewise.
24951 (riscv_load_store_insns): Likewise.
24952 (riscv_force_binary): Likewise.
24953 (riscv_split_symbol): Likewise.
24954 (riscv_force_address): Likewise.
24955 (riscv_legitimize_address): Likewise.
24956 (riscv_move_integer): Likewise.
24957 (riscv_legitimize_const_move): Likewise.
24958 (riscv_legitimize_move): Likewise.
24959 (riscv_address_cost): Likewise.
24960 (riscv_subword): Likewise.
24961 (riscv_output_move): Likewise.
24962 (riscv_canonicalize_int_order_test): Likewise.
24963 (riscv_emit_int_order_test): Likewise.
24964 (riscv_function_arg_boundary): Likewise.
24965 (riscv_pass_mode_in_fpr_p): Likewise.
24966 (riscv_pass_fpr_single): Likewise.
24967 (riscv_pass_fpr_pair): Likewise.
24968 (riscv_get_arg_info): Likewise.
24969 (riscv_function_arg): Likewise.
24970 (riscv_function_arg_advance): Likewise.
24971 (riscv_arg_partial_bytes): Likewise.
24972 (riscv_function_value): Likewise.
24973 (riscv_pass_by_reference): Likewise.
24974 (riscv_setup_incoming_varargs): Likewise.
24975 (riscv_print_operand): Likewise.
24976 (riscv_elf_select_rtx_section): Likewise.
24977 (riscv_save_restore_reg): Likewise.
24978 (riscv_for_each_saved_reg): Likewise.
24979 (riscv_register_move_cost): Likewise.
24980 (riscv_hard_regno_mode_ok_p): Likewise.
24981 (riscv_hard_regno_nregs): Likewise.
24982 (riscv_class_max_nregs): Likewise.
24983 (riscv_memory_move_cost): Likewise.
24984 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24985 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24986 (rl78_addr_space_address_mode): Likewise.
24987 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24989 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24990 (rs6000_reassociation_width): Likewise.
24991 (rs6000_invalid_binary_op): Likewise.
24992 (fusion_p9_p): Likewise.
24993 (emit_fusion_p9_load): Likewise.
24994 (emit_fusion_p9_store): Likewise.
24995 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24996 (ok_for_simple_move_operands): Likewise.
24997 (ok_for_simple_move_strict_operands): Likewise.
24998 (ok_for_simple_arith_logic_operands): Likewise.
24999 (visium_legitimize_reload_address): Likewise.
25000 (visium_select_cc_mode): Likewise.
25001 (output_cbranch): Likewise.
25002 (visium_split_double_move): Likewise.
25003 (visium_expand_copysign): Likewise.
25004 (visium_expand_int_cstore): Likewise.
25005 (visium_expand_fp_cstore): Likewise.
25006 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25007 (visium_function_arg): Likewise.
25008 (visium_function_arg_advance): Likewise.
25009 (visium_libcall_value): Likewise.
25010 (visium_setup_incoming_varargs): Likewise.
25011 (visium_legitimate_constant_p): Likewise.
25012 (visium_legitimate_address_p): Likewise.
25013 (visium_legitimize_address): Likewise.
25014 (visium_secondary_reload): Likewise.
25015 (visium_register_move_cost): Likewise.
25016 (visium_memory_move_cost): Likewise.
25017 (prepare_move_operands): Likewise.
25018 (ok_for_simple_move_operands): Likewise.
25019 (ok_for_simple_move_strict_operands): Likewise.
25020 (ok_for_simple_arith_logic_operands): Likewise.
25021 (visium_function_value_1): Likewise.
25022 (rtx_ok_for_offset_p): Likewise.
25023 (visium_legitimize_reload_address): Likewise.
25024 (visium_split_double_move): Likewise.
25025 (visium_expand_copysign): Likewise.
25026 (visium_expand_int_cstore): Likewise.
25027 (visium_expand_fp_cstore): Likewise.
25028 (visium_split_cstore): Likewise.
25029 (visium_select_cc_mode): Likewise.
25030 (visium_split_cbranch): Likewise.
25031 (output_cbranch): Likewise.
25032 (visium_print_operand_address): Likewise.
25033 * expmed.c (flip_storage_order): Likewise.
25034 * expmed.h (emit_cstore): Likewise.
25035 (flip_storage_order): Likewise.
25036 * genrecog.c (validate_pattern): Likewise.
25037 * hsa-gen.c (gen_hsa_addr): Likewise.
25038 * internal-fn.c (expand_arith_overflow): Likewise.
25039 * ira-color.c (allocno_copy_cost_saving): Likewise.
25040 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25041 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25042 (process_invariant_for_inheritance): Likewise.
25043 * lra-eliminations.c (move_plus_up): Likewise.
25044 * omp-low.c (lower_oacc_reductions): Likewise.
25045 * simplify-rtx.c (simplify_subreg): Likewise.
25046 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25047 (TARGET_CHKP_BOUND_MODE): Likewise..
25048 * targhooks.c (default_chkp_bound_mode): Likewise.
25049 (default_setup_incoming_vararg_bounds): Likewise.
25050 * targhooks.h (default_chkp_bound_mode): Likewise.
25051 (default_setup_incoming_vararg_bounds): Likewise.
25052 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25053 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25054 (have_whole_vector_shift): Likewise.
25055 * tree-vect-stmts.c (vectorizable_load): Likewise.
25056 * doc/tm.texi: Regenerate.
25058 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25060 Graceful degrade if Binutils PR21472 is not available.
25063 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25064 .rodata in flash test fails.
25065 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25066 * confgure: Regenerate.
25067 * config.in: Regenerate.
25068 * config/avr/avr.c (avr_asm_named_section)
25069 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25070 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25071 (avr_asm_init_sections): Same.
25073 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25075 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25076 (fma<VH:mode>4_intrinsic): Likewise.
25077 (*fmsub<VCVTF:mode>4): Likewise.
25078 (*fmsub<VH:mode>4_intrinsic): Likewise.
25080 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25083 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25084 Don't depend on "optimize > 0".
25085 (out_movhi_r_mr, out_movqi_mr_r): Same.
25086 (out_movhi_mr_r, out_movqi_r_mr): Same.
25087 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25088 io_address_operand on "optimize > 0".
25090 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25092 * tree-loop-distribution.c: Add general explanantion on the pass.
25093 (generate_loops_for_partition): Mark distributed loop.
25094 (pg_add_dependence_edges): New parameter. Handle alias data
25095 dependence specially and record it in the parameter if asked.
25096 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25097 (init_partition_graph_vertices, add_partition_graph_edge): New.
25098 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25099 (free_partition_graph_vdata, build_partition_graph): New.
25100 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25101 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25102 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25103 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25104 (version_for_distribution_p, finalize_partitions): New.
25105 (distribute_loop): Handle alias data dependence specially. Factor
25106 out loop fusion code as functions and call these functions.
25108 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25110 * tree-loop-distribution.c (classify_partition): New parameter and
25111 better handle reduction statement.
25112 (rdg_build_partitions): Revise comment.
25113 (distribute_loop): Compute statements in all partitions and pass it
25114 to classify_partition.
25116 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25118 * tree-loop-distribution.c (enum partition_type): New.
25119 (struct partition): New field type.
25120 (partition_merge_into): Add parameter. Update partition type.
25121 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25122 (build_rdg_partition_for_vertex): Compute partition type.
25123 (rdg_build_partitions): Dump partition type.
25124 (distribute_loop): Update calls to partition_merge_into.
25126 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25128 * tree-loop-distribution.c (struct ddr_hasher): New.
25129 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25131 (classify_partition): Call get_data_dependence.
25132 (pg_add_dependence_edges): Ditto.
25133 (distribute_loop): Release data dependence hash table.
25135 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25137 * tree-loop-distribution.c (ref_base_address): Delete.
25138 (similar_memory_accesses): Rename ...
25139 (share_memory_accesses): ... to this. Check if partitions access
25140 the same memory reference.
25141 (distribute_loop): Call share_memory_accesses.
25143 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25145 * tree-loop-distribution.c (struct partition): New field recording
25146 its data reference.
25147 (partition_alloc, partition_free): Init and release data refs.
25148 (partition_merge_into): Merge data refs.
25149 (build_rdg_partition_for_vertex): Collect data refs for partition.
25150 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25152 (distribute_loop): Remve data refs from vertice data of partition
25155 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25157 * tree-loop-distribution.c (params.h): Include header file.
25158 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25159 (datarefs_vec): New global var.
25160 (create_rdg_vertices): Use datarefs_vec directly.
25161 (free_rdg): Don't free data references.
25162 (build_rdg): Update use. Don't free data references.
25163 (distribute_loop): Compute global variable for data references.
25164 Bail out if there are too many data references.
25166 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25168 * tree-loop-distribution.c (loop_nest): New global var.
25169 (build_rdg): Use loop directly, rather than loop nest.
25170 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25172 (distribute_loop): Compute global variable loop nest. Update use.
25174 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25176 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25177 (partition_merge_into): New parameter. Dump reason for fusion.
25178 (distribute_loop): Update use of partition_merge_into.
25180 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25182 * tree-loop-distribution.c (bb_top_order_index): New.
25183 (bb_top_order_index_size, bb_top_order_cmp): New.
25184 (stmts_from_loop): Use topological order.
25185 (pass_loop_distribution::execute): Compute and release topological
25186 order for basic blocks.
25188 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25190 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25193 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25195 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25196 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25197 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25198 * internal-fn.def (LOOP_DIST_ALIAS): New.
25199 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25200 (fold_loop_internal_call): ... this.
25201 (vect_loop_dist_alias_call): New function.
25202 (set_uid_loop_bbs): Call fold_loop_internal_call.
25203 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25206 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25209 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25210 Require dead FLAGS_REG at the beginning of a peephole.
25212 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25215 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25216 arguments in the call to __builtin_ia32_sbb_u32.
25217 (_subborrow_u64): Swap _X and _Y arguments in the call to
25218 __builtin_ia32_sbb_u64.
25220 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25223 * tree-vrp.c (compare_assert_loc): Turn into a function template
25224 with stable template parameter. Only test if a->e is NULL,
25225 !a->e == !b->e has been verified already. Use e == NULL or
25226 e != NULL instead of e or ! e tests. If stable is true, don't use
25227 iterative_hash_expr, on the other side allow a or b or both NULL
25228 and sort the NULLs last.
25229 (process_assert_insertions): Sort using compare_assert_loc<false>
25230 instead of compare_assert_loc, later sort using
25231 compare_assert_loc<true> before calling process_assert_insertions_for
25232 in a loop. Use break instead of continue once seen NULL pointer.
25234 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25236 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25237 Cortex-R7 and Cortex-R8 processors.
25239 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25241 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25242 uninitialized while src is not.
25244 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25246 * common/config/arm/arm-common.c: Adjust include path for
25248 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25249 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25250 (arm-cpu-data.h): Likewise.
25251 (arm-cpu-cdata.h): Likewise.
25252 * config/arm/arm-cpu.h: Delete.
25253 * config/arm/arm-cpu-cdata.h: Delete.
25254 * config/arm/arm-cpu-data.h: Delete.
25256 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25258 * config/arm/arm-cpus.in (cortex-a55): New.
25259 (cortex-a75): Likewise.
25260 (cortex-a75.cortex-a55): Likewise.
25261 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25263 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25264 * config/arm/arm-cpu-cdata.h: Regenerate.
25265 * config/arm/arm-cpu-data.h: Regenerate.
25266 * config/arm/arm-cpu.h: Regenerate.
25267 * config/arm/arm-tables.opt: Regenerate.
25268 * config/arm/arm-tune.md: Regenerate.
25270 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25272 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25274 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25276 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25279 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25281 PR tree-optimization/81292
25282 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25283 full_string_p, also call adjust_related_strinfos if the adjustment
25284 is simple, otherwise invalidate related strinfos.
25286 2017-07-04 Martin Liska <mliska@suse.cz>
25289 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25290 newly created variable as DECL_IGNORED_P.
25292 2017-07-04 Martin Liska <mliska@suse.cz>
25295 * ipa-inline.c (inline_small_functions):
25296 Use xstrdup_for_dump.
25298 2017-07-04 Tom de Vries <tom@codesourcery.com>
25300 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25302 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25305 * config/darwin.c (darwin_function_switched_text_sections):
25308 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25310 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25312 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25314 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25316 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25318 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25319 min_profitable_iters, and th as inclusive lower bounds.
25320 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25321 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25322 for min_profitable_iters and min_profitable_estimate.
25323 (vect_transform_loop): Treat th as an inclusive lower bound.
25324 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25326 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25329 * config/darwin.c (darwin_function_switched_text_sections):
25330 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25331 in two pieces, and suppress the use of buf.
25333 2017-07-03 Nathan Sidwell <nathan@acm.org>
25335 * hash-table.h (hash_table_mod1): Fix indentation.
25337 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25339 PR middle-end/81290
25340 * predict.c (force_edge_cold): Be more careful about propagation
25342 * profile-count.h (profile_probability::guessed,
25343 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25345 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25347 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25349 * doc/invoke.texi (rcpc architecture extension): Document it.
25351 2017-07-03 Richard Biener <rguenther@suse.de>
25353 PR tree-optimization/60510
25354 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25355 the scalar reduction PHI and use it.
25356 (vectorizable_reduction): Properly guard the single_defuse_cycle
25357 path for non-SLP reduction chains where we cannot use it.
25358 Rework reduc_def/index and vector type deduction. Rework
25359 vector operand gathering during reduction op code-gen.
25360 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25361 chains dissolve the chain and leave it to non-SLP reduction
25364 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25366 * tree-data-ref.h (dr_alignment): Declare.
25367 * tree-data-ref.c (dr_alignment): New function.
25368 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25369 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25371 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25373 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25375 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25376 and base_misalignment fields.
25377 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25378 * tree-data-ref.c: Include builtins.h.
25379 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25380 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25381 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25382 * tree-vect-data-refs.c: Include tree-cfg.h.
25383 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25384 fields instead of calculating an alignment here.
25385 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25386 innermost_loop_behavior fields.
25388 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25390 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25392 (DR_STEP_ALIGNMENT): New macro.
25393 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25394 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25395 (create_data_ref): Print it.
25396 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25397 to tell whether the step preserves vector (mis)alignment.
25398 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25399 Move the check for an integer step and generalise to all INTEGER_CST.
25400 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25401 Print the outer step alignment.
25403 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25405 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25406 with offset_alignment.
25407 (DR_ALIGNED_TO): Delete.
25408 (DR_OFFSET_ALIGNMENT): New macro.
25409 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25410 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25411 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25412 (create_data_ref): Likewise.
25413 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25414 (vect_analyze_data_refs): Likewise.
25415 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25416 creating dummy innermost behavior.
25418 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25420 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25421 with a "innermost_loop_behavior *" and refeence tree.
25422 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25423 (create_data_ref): Update call accordingly.
25424 * tree-predcom.c (find_looparound_phi): Likewise.
25426 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25428 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25429 fields with dr_wrt_vec_loop.
25430 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25431 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25432 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25433 (vect_dr_behavior): New function.
25434 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25435 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25436 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25437 track whether the step preserves the misalignment.
25438 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25439 Use vect_dr_behavior.
25440 (vect_setup_realignment): Update call accordingly.
25441 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25442 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25443 call to vect_create_addr_base_for_vector_ref.
25444 (vect_create_cond_for_align_checks): Likewise.
25445 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25446 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25447 (vect_recog_mask_conversion_pattern): Likewise.
25448 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25449 (new_stmt_vec_info): Remove redundant zeroing.
25451 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25453 * common/config/arm/arm-common.c (arm_be8_option): New function.
25454 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25455 (ISA_ARMv6): Add isa_bit_be8.
25456 * config/arm/arm.h (arm_be8_option): Add prototype.
25457 (BE8_SPEC_FUNCTION): New define.
25458 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25459 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25460 (mlittle-endian): Similarly.
25461 (mbe8, mbe32): New options.
25462 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25463 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25465 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25467 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25469 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25471 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25472 (cleanup_tree_cfg_bb): Use it.
25473 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25475 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25477 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25480 * loop-doloop.c (add_test): Update profile.
25482 2017-07-03 Martin Liska <mliska@suse.cz>
25485 * sanopt.c (rewrite_usage_of_param): New function.
25486 (sanitize_rewrite_addressable_params): Likewise.
25487 (pass_sanopt::execute): Call rewrite_usage_of_param.
25489 2017-07-03 Richard Biener <rguenther@suse.de>
25491 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25492 back to using VIEW_CONVERT_EXPR.
25494 2017-07-03 Martin Liska <mliska@suse.cz>
25497 * doc/extend.texi: Document when a resolver function is
25498 generated for target_clones.
25500 2017-07-03 Martin Liska <mliska@suse.cz>
25502 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25503 and shadow memory of auto variables which are subject of
25504 use-after-scope sanitization.
25505 (asan_expand_mark_ifn): Add do set only when is_poison.
25507 2016-07-03 Richard Biener <rguenther@suse.de>
25509 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25511 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25512 (vectorizable_reduction): Perform reduction PHI creation when
25513 visiting a reduction PHI and adjust and simplify code generation
25514 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25515 (vect_transform_loop): Visit reduction PHIs.
25516 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25517 defs into the SLP tree.
25518 (vect_build_slp_tree): Reduction defs terminate the recursion.
25519 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25521 (vect_get_vec_defs_for_stmt_copy): Export.
25522 (vect_get_vec_defs): Likewise.
25523 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25525 (vect_get_vec_defs_for_stmt_copy): Declare.
25526 (vect_get_vec_defs): Likewise.
25528 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25530 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25531 parameter with a "loop" parameter and use it instead of the
25532 loop containing DR_STMT. Don't check simple_iv when doing
25533 BB analysis. Describe the two analysis modes in the comment.
25535 2017-07-03 Tom de Vries <tom@codesourcery.com>
25537 PR tree-optimization/69468
25538 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25539 (find_same_succ_bb): Handle ignore_edge_flags.
25541 2017-07-03 Tom de Vries <tom@codesourcery.com>
25543 PR tree-optimization/81192
25544 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25546 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25548 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25550 2017-07-03 Tom de Vries <tom@codesourcery.com>
25552 PR tree-optimization/81192
25553 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25554 BB_SAME_SUCC (bb) == NULL.
25556 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25558 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25561 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25563 * dumpfile.c: Include profile-count.h
25564 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25566 (insert_cond_bb): Update profile.
25567 * tree-cfg.h (insert_cond_bb): Update prototype.
25568 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25569 * tree-dump.c: Do not include tree-cfg.
25571 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25573 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25575 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25577 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25580 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25582 * tree-complex.c (expand_complex_div_wide): update profile.
25584 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25585 Alan Hayward <alan.hayward@arm.com>
25586 David Sherwood <david.sherwood@arm.com>
25588 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25589 (CORETYPES_H): New define.
25590 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25591 (insn-modes-inline.h, s-modes-inline-h): New rules.
25592 (generated_files): Add insn-modes-inline.h.
25593 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25594 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25595 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25596 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25597 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25598 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25599 (build/gencodes.o, build/genconditions.o): Likewise.
25600 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25601 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25602 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25603 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25604 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25605 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25606 * coretypes.h: Include everything up to real.h for generators.
25607 Include insn-modes.h first. Include wide-int-print.h after
25608 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25609 * machmode.h: Don't include insn-modes.h here.
25610 * function-tests.c: Remove includes of signop.h, machmode.h,
25611 double-int.h and wide-int.h.
25613 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25615 * optc-save-gen.awk: Likewise.
25616 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25617 * godump.c: Remove include of wide-int-print.h.
25618 * pretty-print.h: Likewise.
25619 * wide-int-print.cc: Likewise.
25620 * wide-int.cc: Likewise.
25621 * hash-map-tests.c: Remove include of signop.h.
25622 * hash-set-tests.c: Likewise.
25623 * rtl-tests.c: Likewise.
25624 * mkconfig.sh: Remove include of machmode.h.
25625 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25627 (emit_insn_modes_inline_h): ...this new function. Emit the code
25628 into an insn-modes-inline.h header file, adding appropriate
25629 include guards and end comments.
25630 (emit_insn_modes_c_header): Remove include of machmode.h.
25631 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25633 (main): Handle -i flag and call emit_insn_modes_inline_h when
25636 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25638 * tree-ssa-strlen.c (strinfo): Rename the length field to
25639 nonzero_chars. Add a full_string_p field.
25640 (compare_nonzero_chars, zero_length_string_p): New functions.
25641 (get_addr_stridx): Add an offset_out parameter.
25642 Use compare_nonzero_chars.
25643 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25644 (new_strinfo): Update after above changes to strinfo.
25645 (set_endptr_and_length): Set full_string_p.
25646 (get_string_length): Update after above changes to strinfo.
25647 (unshare_strinfo): Update call to new_strinfo.
25648 (maybe_invalidate): Likewise.
25649 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25650 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25651 as a uhwi instead of an shwi. Update after above changes to
25652 strinfo and new_strinfo.
25653 (zero_length_string): Assert that chainsi contains full strings.
25654 Use zero_length_string_p. Update call to new_strinfo.
25655 (adjust_related_strinfos): Update after above changes to strinfo.
25656 Copy full_string_p from origsi.
25657 (adjust_last_stmt): Use zero_length_string_p.
25658 (handle_builtin_strlen): Update after above changes to strinfo and
25659 new_strinfo. Install the lhs as the string length if the previous
25660 entry didn't describe a full string.
25661 (handle_builtin_strchr): Update after above changes to strinfo
25663 (handle_builtin_strcpy): Likewise.
25664 (handle_builtin_strcat): Likewise.
25665 (handle_builtin_malloc): Likewise.
25666 (handle_pointer_plus): Likewise.
25667 (handle_builtin_memcpy): Likewise. Track nonzero characters
25668 that aren't necessarily followed by a nul terminator.
25669 (handle_char_store): Likewise.
25671 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25673 PR tree-optimization/80769
25674 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25675 for malloc and calloc. Document the new invariant that all related
25676 strinfos have delayed lengths or none do.
25677 (verify_related_strinfos): Move earlier in file.
25678 (set_endptr_and_length): New function, split out from...
25679 (get_string_length): ...here. Also set the lengths of related
25681 (zero_length_string): Assert that chainsi has known (rather than
25683 (adjust_related_strinfos): Likewise.
25685 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25687 PR tree-optimization/81136
25688 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25689 assert that two references with the same misalignment have the same
25690 compile-time misalignment if those compile-time misalignments
25693 2017-07-01 Andi Kleen <ak@linux.intel.com>
25695 * print-tree.c (print_node): Print all attributes.
25697 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25699 * cfg.c (scale_bbs_frequencies): New function.
25700 * cfg.h (scale_bbs_frequencies): Declare it.
25701 * cfgloopanal.c (single_likely_exit): Cleanup.
25702 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25704 (scale_loop_profile): Likewise.
25705 (loop_version): Likewise.
25706 (create_empty_loop_on_edge): Update.
25707 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25708 scale_loop_frequencies, scale_loop_profile, loopify,
25709 loop_version): Update prototypes.
25710 * modulo-sched.c (sms_schedule): Update.
25711 * predict.c (unlikely_executed_edge_p): Also check probability.
25712 (probably_never_executed_edge_p): Fix typo.
25713 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25714 * tree-parloops.c (gen_parallel_loop): Update.
25715 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25716 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25717 * tree-ssa-loop-split.c (split_loop): Update.
25718 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25719 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25720 (vect_loop_versioning): Update.
25721 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25723 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25725 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25727 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25729 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25730 to keep profile consistent.
25732 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25734 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25735 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25736 * profile-count.h (max_safe_multiplier): Make unsigned.
25737 (profile_count::guessed_zero): New.
25739 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25741 * bb-reorder.c (fix_up_crossing_landing_pad,
25742 fix_crossing_conditional_branches): Use make_single_succ_edge
25743 to keep profile consistent.
25745 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25747 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25750 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25753 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25754 the right scopes, make sure cond_jump isn't preserved between multiple
25755 iterations. Search for fallthru edge whenever there are 3+ edges and
25756 use find_fallthru_edge for it.
25758 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25760 Patch by Alexander Monakov <amonakov@ispras.ru>
25761 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25762 probabilities consistently.
25764 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25766 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25767 emit_cmp_and_jump_insns.
25769 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25772 * tree-inline.c (expand_call_inline): Combine profile statuses.
25774 2017-06-30 Andrew Pinski <apinski@cavium.com>
25776 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25777 fold_stmt returned true.
25779 2017-06-30 Nathan Sidwell <nathan@acm.org>
25781 * ggc.h (empty_string): Delete.
25782 * cfgexpand.c (expand_asm_stmt): Use plain "".
25783 * optabs.c (expand_asm_memory_barrier): Likewise.
25784 * stringpool.c (empty_string): Delete.
25785 (digit_vector, digit_string): Delete.
25786 (ggc_alloc_string): Use plain "", don't optimize single digit
25787 strings. Use ggc_alloc_atomic.
25789 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25791 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25792 comparison set and one other set, use the cost of the non-comparison
25795 2017-06-30 Nathan Sidwell <nathan@acm.org>
25797 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25800 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25802 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25803 loops. Remove now unneeded calls to gimple_switch_set_label() that
25804 just set removed labels to NULL_TREE.
25806 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25808 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25809 (set_range_info): ...here. Only call set_range_info_raw if domain
25811 (set_nonzero_bits): Call set_range_info_raw.
25812 * tree-ssanames.h (set_range_info_raw): New.
25814 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25817 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25818 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25819 of nonimmediate_operand and <store_mask_constraint> instead of m
25820 for the input operand. For V8FI iterator, always split if input
25821 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25822 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25823 <store_mask_predicate> instead of register_operand and
25824 <store_mask_constraint> instead of v for the input operand. Make
25825 sure both operands aren't MEMs for if not <mask_applied>.
25827 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25829 * lto-wrapper.c (copy_file) Close both file descriptors before
25832 2017-06-30 Martin Liska <mliska@suse.cz>
25835 * multiple_target.c (create_dispatcher_calls): Make ifunc
25836 also for function that don't have calls or are not referenced.
25838 2017-06-30 Richard Biener <rguenther@suse.de>
25840 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25841 analyze the first scalar stmt. Move vector type computation
25842 for the BB case here from ...
25843 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25844 live operation processing in the SLP case properly.
25846 2017-06-30 Richard Biener <rguenther@suse.de>
25848 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25850 2017-06-30 Martin Liska <mliska@suse.cz>
25853 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25854 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25856 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25858 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25861 2017-06-30 Richard Biener <rguenther@suse.de>
25863 PR tree-optimization/81249
25864 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25865 condition reduction result to original scalar type.
25867 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25869 * profile-count.h (enum profile_quality): Fix typos and whitespace
25872 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25874 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25875 type for branch probabilities.
25877 2017-06-29 Julian Brown <julian@codesourcery.com>
25878 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25880 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25881 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25882 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25883 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25885 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25887 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25888 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25889 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25890 CC usage from generic code to here.
25891 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25892 CC usage into the target macros.
25894 2017-06-29 Maya Rashish <coypu@sdf.org>
25896 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25899 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25901 * arm/arm-builtins.c: Include profile-count.h
25902 * except.c (sjlj_emit_function_enter): Use
25903 profile_probability::unlikely.
25905 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25907 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25908 and tocrel_offset be pointer args rather than implicitly using
25910 (legitimate_constant_pool_address_p, rs6000_emit_move,
25911 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25912 tocrel_offset and use in toc_relative_expr_p call.
25913 (print_operand, print_operand_address): Use static tocrel_base_oac
25914 and tocrel_offset_oac.
25915 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25918 2017-06-29 Maya Rashish <coypu@sdf.org>
25920 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25922 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25924 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25925 objects, take into account only the alignment of 'op0' and 'mode1' if
25928 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25930 * ccmp.c (ccmp_tree_comparison_p): New function.
25931 (ccmp_candidate_p): Update to use above function.
25932 (get_compare_parts): New function.
25933 (expand_ccmp_next): Update to use new functions.
25934 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25936 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25937 take mode as argument.
25938 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25939 * expr.c (expand_expr_real_1): Pass mode as argument.
25941 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25943 * combine.c (combine_instructions): Print insns to dump_file, together
25946 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25948 * asan.c (asan_emit_stack_protection): Update.
25949 (create_cond_insert_point): Update.
25950 * auto-profile.c (afdo_propagate_circuit): Update.
25951 * basic-block.h (struct edge_def): Turn probability to
25952 profile_probability.
25953 (EDGE_FREQUENCY): Update.
25954 * bb-reorder.c (find_traces_1_round): Update.
25955 (better_edge_p): Update.
25956 (sanitize_hot_paths): Update.
25957 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25958 (make_single_succ_edge): Update.
25959 (check_bb_profile): Update.
25960 (dump_edge_info): Update.
25961 (update_bb_profile_for_threading): Update.
25962 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25963 probabilitycount to 0.
25964 * cfgbuild.c (compute_outgoing_frequencies): Update.
25965 * cfgcleanup.c (try_forward_edges): Update.
25966 (outgoing_edges_match): Update.
25967 (try_crossjump_to_edge): Update.
25968 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25969 (expand_gimple_tailcall): Update.
25970 (construct_init_block): Use make_single_succ_edge.
25971 (construct_exit_block): Use make_single_succ_edge.
25972 * cfghooks.c (verify_flow_info): Update.
25973 (redirect_edge_succ_nodup): Update.
25974 (split_edge): Update.
25975 (account_profile_record): Update.
25976 * cfgloopanal.c (single_likely_exit): Update.
25977 * cfgloopmanip.c (scale_loop_profile): Update.
25978 (set_zero_probability): Remove.
25979 (duplicate_loop_to_header_edge): Update.
25980 * cfgloopmanip.h (loop_version): Update prototype.
25981 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25982 (force_nonfallthru_and_redirect): Update.
25983 (update_br_prob_note): Update.
25984 (rtl_verify_edges): Update.
25985 (purge_dead_edges): Update.
25986 (rtl_lv_add_condition_to_bb): Update.
25987 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25988 * cgraphunit.c (init_lowered_empty_function): Update.
25989 (cgraph_node::expand_thunk): Update.
25990 * cilk-common.c: Include profile-count.h
25991 * dojump.c (inv): Remove.
25992 (jumpifnot): Update.
25993 (jumpifnot_1): Update.
25994 (do_jump_1): Update.
25996 (do_jump_by_parts_greater_rtx): Update.
25997 (do_compare_rtx_and_jump): Update.
25998 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25999 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26000 * dwarf2cfi.c: Include profile-count.h
26001 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26002 (sjlj_emit_dispatch_table): Likewise.
26003 * explow.c: Include profile-count.h
26004 * expmed.c (emit_store_flag_force): Update.
26005 (do_cmp_and_jump): Update.
26006 * expr.c (compare_by_pieces_d::generate): Update.
26007 (compare_by_pieces_d::finish_mode): Update.
26008 (emit_block_move_via_loop): Update.
26009 (store_expr_with_bounds): Update.
26010 (store_constructor): Update.
26011 (expand_expr_real_2): Update.
26012 (expand_expr_real_1): Update.
26013 * expr.h (try_casesi, try_tablejump): Update prototypes.
26014 * gimple-pretty-print.c (dump_probability): Update.
26015 (dump_profile): New.
26016 (dump_gimple_label): Update.
26017 (dump_gimple_bb_header): Update.
26018 * graph.c (draw_cfg_node_succ_edges): Update.
26019 * hsa-gen.c (convert_switch_statements): Update.
26020 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26021 (find_if_case_1): Update.
26022 (find_if_case_2): Update.
26023 * internal-fn.c (expand_arith_overflow_result_store): Update.
26024 (expand_addsub_overflow): Update.
26025 (expand_neg_overflow): Update.
26026 (expand_mul_overflow): Update.
26027 (expand_vector_ubsan_overflow): Update.
26028 * ipa-cp.c (good_cloning_opportunity_p): Update.
26029 * ipa-split.c (split_function): Use make_single_succ_edge.
26030 * ipa-utils.c (ipa_merge_profiles): Update.
26031 * loop-doloop.c (add_test): Update.
26032 (doloop_modify): Update.
26033 * loop-unroll.c (compare_and_jump_seq): Update.
26034 (unroll_loop_runtime_iterations): Update.
26035 * lra-constraints.c (lra_inheritance): Update.
26036 * lto-streamer-in.c (input_cfg): Update.
26037 * lto-streamer-out.c (output_cfg): Update.
26038 * mcf.c (adjust_cfg_counts): Update.
26039 * modulo-sched.c (sms_schedule): Update.
26040 * omp-expand.c (expand_omp_for_init_counts): Update.
26041 (extract_omp_for_update_vars): Update.
26042 (expand_omp_ordered_sink): Update.
26043 (expand_omp_for_ordered_loops): Update.
26044 (expand_omp_for_generic): Update.
26045 (expand_omp_for_static_nochunk): Update.
26046 (expand_omp_for_static_chunk): Update.
26047 (expand_cilk_for): Update.
26048 (expand_omp_simd): Update.
26049 (expand_omp_taskloop_for_outer): Update.
26050 (expand_omp_taskloop_for_inner): Update.
26051 * omp-simd-clone.c (simd_clone_adjust): Update.
26052 * optabs.c (expand_doubleword_shift): Update.
26053 (expand_abs): Update.
26054 (emit_cmp_and_jump_insn_1): Update.
26055 (expand_compare_and_swap_loop): Update.
26056 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26057 * predict.c (predictable_edge_p): Update.
26058 (edge_probability_reliable_p): Update.
26059 (set_even_probabilities): Update.
26060 (combine_predictions_for_insn): Update.
26061 (combine_predictions_for_bb): Update.
26062 (propagate_freq): Update.
26063 (estimate_bb_frequencies): Update.
26064 (force_edge_cold): Update.
26065 * profile-count.c (profile_count::dump): Add missing space into dump.
26066 (profile_count::debug): Add newline.
26067 (profile_count::differs_from_p): Explicitly convert to unsigned.
26068 (profile_count::stream_in): Update.
26069 (profile_probability::dump): New member function.
26070 (profile_probability::debug): New member function.
26071 (profile_probability::differs_from_p): New member function.
26072 (profile_probability::differs_lot_from_p): New member function.
26073 (profile_probability::stream_in): New member function.
26074 (profile_probability::stream_out): New member function.
26075 * profile-count.h (profile_count_quality): Rename to ...
26076 (profile_quality): ... this one.
26077 (profile_probability): New.
26078 (profile_count): Update.
26079 * profile.c (compute_branch_probabilities): Update.
26080 * recog.c (peep2_attempt): Update.
26081 * sched-ebb.c (schedule_ebbs): Update.
26082 * sched-rgn.c (find_single_block_region): Update.
26083 (compute_dom_prob_ps): Update.
26084 (schedule_region): Update.
26085 * sel-sched-ir.c (compute_succs_info): Update.
26086 * stmt.c (struct case_node): Update.
26087 (do_jump_if_equal): Update.
26088 (get_outgoing_edge_probs): Update.
26089 (conditional_probability): Update.
26090 (emit_case_dispatch_table): Update.
26091 (expand_case): Update.
26092 (expand_sjlj_dispatch_table): Update.
26093 (emit_case_nodes): Update.
26094 * targhooks.c: Update.
26095 * tracer.c (better_p): Update.
26096 (find_best_successor): Update.
26097 * trans-mem.c (expand_transaction): Update.
26098 * tree-call-cdce.c: Update.
26099 * tree-cfg.c (gimple_split_edge): Upate.
26100 (move_sese_region_to_fn): Upate.
26101 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26102 * tree-eh.c (lower_resx): Upate.
26103 (cleanup_empty_eh_move_lp): Upate.
26104 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26105 * tree-inline.c (copy_edges_for_bb): Update.
26106 (copy_cfg_body): Update.
26107 * tree-parloops.c (gen_parallel_loop): Update.
26108 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26109 (gimple_gen_time_profiler): Update.
26110 * tree-ssa-dce.c (remove_dead_stmt): Update.
26111 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26112 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26113 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26114 (unloop_loops): Update.
26115 (try_peel_loop): Update.
26116 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26117 * tree-ssa-loop-split.c (connect_loops): Update.
26118 (split_loop): Update.
26119 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26120 (hoist_guard): Update.
26121 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26122 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26123 (value_replacement): Update.
26124 * tree-ssa-reassoc.c (branch_fixup): Update.
26125 * tree-ssa-tail-merge.c (replace_block_by): Update.
26126 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26127 (create_edge_and_update_destination_phis): Update.
26128 (compute_path_counts): Update.
26129 (recompute_probabilities): Update.
26130 (update_joiner_offpath_counts): Update.
26131 (freqs_to_counts_path): Update.
26132 (duplicate_thread_path): Update.
26133 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26134 (struct switch_conv_info): Update.
26135 (gen_inbound_check): Update.
26136 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26137 (vect_do_peeling): Update.
26138 (vect_loop_versioning): Update.
26139 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26140 (optimize_mask_stores): Update.
26141 * ubsan.c (ubsan_expand_null_ifn): Update.
26142 * value-prof.c (gimple_divmod_fixed_value): Update.
26143 (gimple_divmod_fixed_value_transform): Update.
26144 (gimple_mod_pow2): Update.
26145 (gimple_mod_pow2_value_transform): Update.
26146 (gimple_mod_subtract): Update.
26147 (gimple_mod_subtract_transform): Update.
26148 (gimple_ic): Update.
26149 (gimple_stringop_fixed_value): Update.
26150 (gimple_stringops_transform): Update.
26151 * value-prof.h: Update.
26153 2017-06-29 Carl Love <cel@us.ibm.com>
26155 * config/rs6000/rs6000-c.c: Add support for built-in functions
26156 vector signed int vec_signed (vector float);
26157 vector signed long long vec_signed (vector double);
26158 vector signed int vec_signed2 (vector double, vector double);
26159 vector signed int vec_signede (vector double);
26160 vector signed int vec_signedo (vector double);
26161 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26162 instruction generator.
26163 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26164 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26165 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26167 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26168 vunsignede_v2df): Add define_expands.
26169 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26170 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26171 VEC_UNSIGNEDO): Add definitions.
26172 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26173 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26174 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26175 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26176 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26177 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26178 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26179 vec_unsignede, vec_unsignedo): Add builtin defines.
26180 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26182 * doc/extend.texi: Update the built-in documentation file for the
26183 new built-in functions.
26185 2017-06-29 Richard Biener <rguenther@suse.de>
26187 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26188 reduction chains to LOOP_VINFO_REDUCTIONS.
26189 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26190 SLP reductions after processing reduction chains.
26192 2017-06-29 Nathan Sidwell <nathan@acm.org>
26194 * builtins.c (fold_builtin_FUNCTION): Use
26195 lang_hooks.decl_printable_name.
26197 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26199 PR middle-end/81194
26200 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26201 with only one label.
26202 * stmt.c (expand_case): Assert NCASES is greater than one.
26204 2017-06-29 Richard Biener <rguenther@suse.de>
26206 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26208 (group_case_labels): Likewise.
26209 (find_taken_edge): Push sanity checking on val to workers...
26210 (find_taken_edge_cond_expr): ... here
26211 (find_taken_edge_switch_expr): ... and here, handle cases
26212 with just a default label.
26213 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26214 (group_case_labels): Likewise.
26215 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26216 group_case_labels does anything cleanup the CFG again.
26218 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26220 PR tree-optimization/81196
26221 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26222 exit condition comparing two IVs.
26224 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26226 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26227 profile to the dummy entry at the end of the list of architectures.
26228 * config/arm/arm-cpu-cdata.h: Regenerated.
26230 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26231 Michael Collison <michael.collison@arm.com>
26234 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26236 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26237 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26238 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26239 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26240 of shift when the shift amount is masked with constant equal to
26241 the size of the mode.
26242 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26245 2017-06-29 Martin Liska <mliska@suse.cz>
26247 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26249 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26252 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26255 (edge_growth_cache_entry::edge_growth_cache_entry): New
26257 (reset_edge_growth_cache): Update to use constructor.
26259 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26261 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26262 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26263 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26265 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26267 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26268 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26270 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26272 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26273 (*-linux-uclibc*): Add t-uclibc tmake_file.
26274 * config/t-musl: New.
26275 * config/t-uclibc: New.
26277 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26279 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26281 (gen_comm_data): Emit architectural setting of arch_prof.
26282 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26284 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26285 (armv8-m.base, armv8-m.main): Likewise.
26286 * arm-protos.h (arm_build_target): Add profile field.
26287 (arch_option): Likewise.
26288 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26290 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26291 arm_active_target.profile.
26293 2017-06-28 Richard Biener <rguenther@suse.de>
26295 PR middle-end/81227
26296 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26297 TYPE_OVERFLOW_WRAPS.
26298 * match.pd (negate_expr_p): Likewise.
26299 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26300 fold_build2, not fold_binary.
26302 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26304 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26305 Convert memory address to Pmode.
26306 (aarch64_print_operand): Assert MEM operands are always Pmode.
26308 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26311 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26312 Remove redundant if.
26313 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26314 * config/arm/aarch-common-protos.h
26315 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26316 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26318 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26321 * multiple_target.c (create_dispatcher_calls): Set the default
26322 clone to be static, not public.
26324 2017-06-28 Richard Biener <rguenther@suse.de>
26326 * tree-vect-loop.c (vectorizable_reduction): Move special
26327 cond reduction IV var creation ...
26328 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26329 parameter. Use STMT_VINFO_VECTYPE.
26330 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26333 2017-06-28 Martin Liska <mliska@suse.cz>
26336 * ipa-visibility.c (non_local_p): Handle visibility.
26338 2017-06-28 Martin Liska <mliska@suse.cz>
26341 * common.opt: Add IntegerRange to various options.
26342 * opt-functions.awk (integer_range_info): New function.
26343 * optc-gen.awk: Add integer_range_info to cl_options struct.
26344 * opts-common.c (decode_cmdline_option): Handle
26345 CL_ERR_INT_RANGE_ARG.
26346 (cmdline_handle_error): Likewise.
26347 * opts.c (print_filtered_help): Show valid interval in
26348 when --help is provided.
26349 * opts.h (struct cl_option): Add range_min and range_max fields.
26350 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26352 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26354 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26355 (x * C EQ/NE y * C): New transformation.
26357 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26359 * genmultilib (combination_space): Accept '+' in option names.
26361 2017-06-28 Martin Liska <mliska@suse.cz>
26364 * asan.c (instrument_derefs): Bail out inner references
26365 that are hard register variables.
26367 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26370 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26371 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26373 2017-06-28 Richard Biener <rguenther@suse.de>
26375 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26376 (vect_get_slp_defs): Adjust.
26377 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26379 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26381 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26382 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26383 (vectorizable_reduction): Adjust.
26384 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26386 (vect_get_slp_defs): Likewise.
26387 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26388 (vectorizable_bswap): Adjust.
26389 (vectorizable_call): Likewise.
26390 (vectorizable_conversion): Likewise.
26391 (vectorizable_assignment): Likewise.
26392 (vectorizable_shift): Likewise.
26393 (vectorizable_operation): Likewise.
26394 (vectorizable_store): Likewise.
26395 (vectorizable_condition): Likewise.
26396 (vectorizable_comparison): Likewise.
26398 2017-06-28 Michael Collison <michael.collison@arm.com>
26401 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26403 (arm_gen_movmemqi): Removed unused variable 'i'.
26404 Convert 'for' loop into 'while' loop.
26405 (arm_expand_prologue): Remove last unnecessary set of insn.
26406 (thumb_pop): Remove unused variable 'pushed_words'.
26407 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26409 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26411 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26412 * config/s390/s390-protos.h: Add prototype of
26413 s390_rel_address_ok_p.
26414 * config/s390/s390.c (s390_got_symbol): New function.
26415 (s390_rel_address_ok_p): New function.
26416 (legitimize_pic_address): Use s390_rel_address_ok_p.
26417 (s390_load_got): Use s390_got_symbol.
26418 (s390_option_override): Issue error if
26419 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26420 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26422 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26424 2017-06-27 Andrew Pinski <apinski@cavium.com>
26426 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26427 (X * copysign (1.0, X)): New pattern.
26428 (X * copysign (1.0, -X)): New pattern.
26429 (copysign (-1.0, CST)): New pattern.
26431 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26433 * genmultilib (combination_space): Remove variable.
26434 Validate reuse rules against regular expression for any sequence
26435 of multilib options in any order.
26437 2017-06-27 Michael Collison <michael.collison@arm.com>
26439 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26440 call aarch64_split_simd_combine.
26441 * (aarch64_combine_internal<mode>): Delete pattern.
26442 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26443 Allow register and subreg operands.
26445 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26447 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26448 specific need, just fallback on defaults.
26449 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26451 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26452 Olivier Hainque <hainque@adacore.com>
26454 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26456 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26457 targets. Pick a default if no particular attempt applied.
26458 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26461 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26463 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26464 (x86_64-wrs-vxworks7): Likewise.
26466 2017-06-27 Marek Polacek <polacek@redhat.com>
26469 * ubsan.c (instrument_null): Check get_base_address's result for null.
26471 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26473 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26475 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26477 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26478 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26479 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26480 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26481 New function types.
26482 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26483 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26484 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26485 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26486 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26487 BUILT_IN_FEUPDATEENV): New builtins.
26488 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26489 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26490 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26491 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26493 (builtin_structptr_types): Adjust size.
26494 * tree.c (builtin_structptr_types): Add four entries.
26496 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26497 Olivier Hainque <hainque@adacore.com>
26499 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26500 (TLS_SYM): New local macro, forcing reference to __tls__ on
26501 link command lines for VxWorks 7 RTPs, triggering initialization
26503 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26504 OS features TLS support, true for RTPs on VxWorks 7.
26505 * config/vxworks.c (vxworks_override_options): Setup emutls
26508 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26510 * predict.c (test_prediction_value_range): Use -1U instead of -1
26511 to avoid narrowing conversion warning.
26512 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26513 to avoid narrowing conversion warning.
26514 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26516 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26518 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26520 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26521 64bit configurations.
26522 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26523 (SIZE_TYPE): Likewise.
26524 * config/vxworks.c (vxworks_emutls_var_fields): Use
26525 long_unsigned_type_node instead of unsigned_type_node as the offset
26526 field type, which is "pointer" mode in emutls.c.
26528 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26531 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26533 PR middle-end/81207
26534 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26535 gimple_vuse copying separately from gimple_vdef copying.
26537 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26539 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26542 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26543 Olivier Hainque <hainque@adacore.com>
26545 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26546 for all vxworks7 targets.
26547 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26548 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26549 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26550 variations for VX6/VX7 and 32/64bits later on in ...
26551 (VXWORKS_LIB_SPEC): Leverage new macros.
26552 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26553 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26555 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26557 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26558 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26560 2017-06-26 Carl Love <cel@us.ibm.com>
26562 * config/rs6000/rs6000-c.c: Add support for built-in functions
26563 vector bool char vec_reve (vector bool char);
26564 vector signed char vec_reve (vector signed char);
26565 vector unsigned char vec_reve (vector unsigned char);
26566 vector bool int vec_reve (vector bool int);
26567 vector signed int vec_reve (vector signed int);
26568 vector unsigned int vec_reve (vector unsigned int);
26569 vector bool long long vec_reve (vector bool long long);
26570 vector signed long long vec_reve (vector signed long long);
26571 vector unsigned long long vec_reve (vector unsigned long long);
26572 vector bool short vec_reve (vector bool short);
26573 vector signed short vec_reve (vector signed short);
26574 vector double vec_reve (vector double);
26575 vector float vec_reve (vector float);
26576 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26577 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26578 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26579 (altivec_vreve): New pattern.
26580 * config/rs6000/altivec.h (vec_reve): New define.
26581 * doc/extend.texi (vec_rev): Update the built-in documentation file
26582 for the new built-in functions.
26584 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26586 PR tree-optimization/71815
26587 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26589 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26591 (slsr_process_phi): Likewise.
26592 (replace_uncond_cands_and_profitable_phis): Don't replace a
26593 multiply candidate with a stride of 1 (copy or cast).
26594 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26596 (lowest_cost_path): Likewise.
26597 (total_savings): Likewise.
26599 2017-06-26 Richard Biener <rguenther@suse.de>
26602 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26603 Use def_builtin_pure for all gather builtins.
26605 2017-06-26 Richard Biener <rguenther@suse.de>
26607 PR tree-optimization/81203
26608 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26609 non-dominating BBs.
26611 2017-06-26 Marek Polacek <polacek@redhat.com>
26614 * doc/invoke.texi: Document -Wmultistatement-macros.
26616 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26618 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26619 arm_neon_ok_no_float_abi effective target.
26621 2017-06-26 Richard Biener <rguenther@suse.de>
26623 PR tree-optimization/80928
26624 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26625 (copy_bbs): Set BB_DUPLICATED flag early.
26626 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26628 (execute_on_shrinking_pred): Likewise.
26629 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26630 BB_DUPLICATED blocks.
26631 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26632 iterate over all PHIs considering removal of *gsi.
26634 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26636 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26639 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26641 * config/rs6000/rs6000-string.c: (expand_block_clear,
26642 do_load_for_compare, select_block_compare_mode,
26643 compute_current_alignment, expand_block_compare,
26644 expand_strncmp_align_check, expand_strn_compare,
26645 expand_block_move, rs6000_output_load_multiple)
26646 Move functions related to string/block move/compare
26647 to a separate file.
26648 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26649 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26650 for this function which is now used in two files.
26651 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26652 * config.gcc: Add rs6000-string.o to extra_objs for
26653 targets powerpc*-*-* and rs6000*-*-*.
26655 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26658 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26659 32-bit, since indexed is not valid for DImode.
26660 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26661 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26662 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26663 (define_peephole2 for Altivec d-form store): Likewise.
26666 * multiple_target.c (create_dispatcher_calls): Only create the
26667 dispatcher call if the function is the default clone of a
26668 versioned function.
26670 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26672 PR middle-end/80902
26673 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26674 a call, force the call to not be a tail call.
26676 2017-06-23 Jeff Law <law@redhat.com>
26678 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26681 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26683 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26684 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26685 handling for early expansion of vector shifts (sl,sr,sra,rl).
26686 (builtin_function_type): Add vector shift right instructions
26687 to the unsigned argument list.
26689 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26691 rtl-optimizatoin/79286
26692 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26693 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26694 trap. PIC register plus a const unspec without offset can never trap.
26696 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26698 * tree.h (builtin_structptr_type): New type.
26699 (builtin_structptr_types): Declare new array.
26700 * tree.c (builtin_structptr_types): New array.
26701 (free_lang_data, build_common_tree_nodes): Use it.
26703 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26706 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26709 2017-06-22 Matt Turner <mattst88@gmail.com>
26711 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26712 Lake models to skylake case. Assume skylake for unknown
26713 models with clflushopt.
26715 2017-06-22 Jeff Law <law@redhat.com>
26717 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26718 frame sizes that do not satisfy aarch64_uimm12_shift.
26720 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26722 * profile-count.h (apply_probability,
26723 apply_scale, probability_in): Fix checks for zero.
26725 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26727 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26728 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26730 2016-06-22 Richard Biener <rguenther@suse.de>
26732 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26733 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26734 REDUC_MAX_EXPR support.
26735 (vectorizable_reduction): Likewise.
26736 (vect_create_epilog_for_reduction): Likewise.
26738 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26740 * match.pd (A / (1 << B) -> A >> B): New.
26741 * generic-match-head.c: Include optabs-tree.h.
26742 * gimple-match-head.c: Likewise.
26743 * optabs-tree.h (target_supports_op_p): New.
26744 * optabs-tree.c (target_supports_op_p): New.
26746 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26748 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26749 $gcc_cv_ld --help output.
26750 (gcc_cv_ld_demangle): Likewise.
26751 (gcc_cv_ld_eh_frame_hdr): Likewise.
26752 (gcc_cv_ld_pie): Likewise.
26753 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26754 (gcc_cv_ld_buildid): Likewise.
26755 (gcc_cv_ld_sysroot): Likewise.
26756 (ld_bndplt_support): Likewise.
26757 (ld_pushpopstate_support): Likewise.
26758 * configure: Regenerate.
26759 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26761 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26764 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26765 operands indexes to avoid gap between operands and match_dups.
26767 2017-06-21 Andrew Pinski <apinski@cavium.com>
26769 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26770 Increment Arith_shift and Arith_shift_reg by 1.
26771 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26773 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26774 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26775 (aarch64_strip_extend): Add new argument and test for it.
26776 (aarch64_cheap_mult_shift_p): New function.
26777 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26778 add a cost if it is true.
26779 Update calls to aarch64_strip_extend.
26780 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26782 2017-06-21 Andrew Pinski <apinski@cavium.com>
26784 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26786 (thunderxt88): Likewise.
26787 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26788 (thunderx_prefetch_tune): New variable.
26789 (thunderx2t99_prefetch_tune): Update for the correct values.
26790 (thunderxt88_tunings): New variable.
26791 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26792 generic_prefetch_tune.
26793 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26795 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26797 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26798 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26799 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26800 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26801 (aarch64_atomic_cas<mode>, GPI): Likewise.
26803 2017-06-21 Martin Liska <mliska@suse.cz>
26805 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26806 statements on cold and hot labels.
26807 * predict.c (tree_estimate_probability_bb): Remove the
26808 prediction from this place.
26810 2017-06-21 Martin Liska <mliska@suse.cz>
26812 PR tree-optimization/79489
26813 * gimplify.c (maybe_add_early_return_predict_stmt): New
26815 (gimplify_return_expr): Call the function.
26816 * predict.c (tree_estimate_probability_bb): Remove handling
26818 * predict.def: Update comment about early return predictor.
26819 * gimple-predict.h (is_gimple_predict): New function.
26820 * predict.def: Change default value of early return to 66.
26821 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26823 * passes.def: Put pass_strip_predict_hints to the beginning of
26826 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26828 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26829 FUNCTION_DECL declarations.
26830 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26832 (dwaf2out_decl): Likewise.
26833 * godump.c (go_early_global_decl): Skip call to the real debug hook
26834 for FUNCTION_DECL declarations.
26835 * passes.c (rest_of_decl_compilation): Skip call to the
26836 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26837 -fdump-go-spec is passed.
26839 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26841 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26842 Reorder for compactness.
26843 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26844 (def_builtin_pure, def_builtin_pure2): New functions.
26845 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26847 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26849 * match.pd (nop_convert): New predicate.
26850 ((A +- CST1) +- CST2): Allow some NOP conversions.
26852 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26855 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26856 with ctors/dtors if GOVD_SHARED is set.
26858 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26860 * config/aarch64/aarch64.md (movti_aarch64):
26861 Emit mov rather than orr.
26862 (movtf_aarch64): Likewise.
26863 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26864 Emit mov rather than orr.
26866 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26868 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26869 Swap alternatives, make integer dup more expensive.
26871 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26873 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26874 Return true for non-tls symbols.
26876 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26878 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26879 (cortex-a75): Likewise.
26880 (cortex-a75.cortex-a55): Likewise.
26881 * config/aarch64/aarch64-tune.md: Regenerate.
26882 * doc/invoke.texi (-mtune): Document new values for -mtune.
26884 2017-06-21 Tom de Vries <tom@codesourcery.com>
26886 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26887 stack_size feature.
26888 (Effective-Target Keywords, Other attributes): Suggest using
26889 dg-add-options stack_size feature to get stack limit in stack_size
26890 effective target documentation.
26892 2017-06-21 Julian Brown <julian@codesourcery.com>
26893 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26895 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26896 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26897 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26899 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26900 attribute type list for neon_multiply.
26901 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26902 attribute type list for neon_multiply.
26903 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26904 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26905 attribute type list for neon_multiply.
26906 * config/arm/types.md (crypto_pmull): Add.
26907 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26908 attribute type list.
26910 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26912 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26915 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26917 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26918 to make sure not to dereference a NULL cost_classes_ptr pointer.
26920 2017-06-20 Carl Love <cel@us.ibm.com>
26922 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26923 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26924 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26925 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26926 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26927 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26928 VMULOSW): New enum "unspec" values.
26929 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26930 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26931 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26932 altivec_vmulosw): New patterns.
26933 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26934 VMULOSW): Add definitions.
26936 2017-06-20 Julia Koval <julia.koval@intel.com>
26938 * config/i386/i386.c: Fix rounding expand for new pattern.
26939 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26941 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26943 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26944 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26946 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26948 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26951 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26953 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26954 architecture, then by implementer ID.
26955 * config/aarch64/aarch64-tune.md: Regenerate.
26957 2017-06-20 Richard Biener <rguenther@suse.de>
26959 PR middle-end/81097
26960 * fold-const.c (split_tree): Fold to type before negating.
26962 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26964 * diagnostic-show-locus.c
26965 (selftest::test_fixit_deletion_affecting_newline): New function.
26966 (selftest::diagnostic_show_locus_c_tests): Call it.
26968 2017-06-20 Andreas Schwab <schwab@suse.de>
26971 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26974 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26976 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26977 __ARM_FEATURE_COPROC according to support.
26979 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26981 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26982 Rewritten to avoid overflow for > 32-bit pointers.
26985 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26986 by removing enum keyword.
26987 (ubsan_type_descriptor): Likewise. Formatting fix.
26990 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26991 splitter): Require TARGET_SSE2 in the condition.
26993 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26996 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26997 for doing vector set of SFmode on ISA 3.0.
26998 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26999 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27001 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27002 SFmode value into a V4SF variable that was extracted from another
27003 V4SF variable without converting the element to double precision
27004 and back to single precision vector format.
27005 (vsx_insert_extract_v4sf_p9_2): Likewise.
27007 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27009 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27010 in UWHI to avoid undefined overflow.
27013 * ubsan.h (enum ubsan_encode_value_phase): New.
27014 (ubsan_encode_value): Change second argument to
27015 enum ubsan_encode_value_phase with default value of
27016 UBSAN_ENCODE_VALUE_GENERIC.
27017 * ubsan.c (ubsan_encode_value): Change second argument to
27018 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27019 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27020 create_tmp_var_raw instead of create_tmp_var and use a
27022 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27023 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27024 ubsan_encode_value callers.
27027 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27028 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27029 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27031 2017-06-19 Richard Biener <rguenther@suse.de>
27033 PR middle-end/81118
27034 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27035 estimates if we changed anything.
27037 2017-06-19 Richard Biener <rguenther@suse.de>
27039 PR tree-optimization/80887
27040 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27041 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27042 simplified lookups, then reset mprts_hook.
27043 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27045 (try_to_simplify): Likewise.
27047 2017-06-19 Martin Liska <mliska@suse.cz>
27050 * gimplify.c (gimplify_switch_expr):
27051 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27053 2017-06-19 Martin Liska <mliska@suse.cz>
27055 * doc/install.texi: Document that PGO runs in 4 stages.
27057 2017-06-19 Martin Liska <mliska@suse.cz>
27060 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27061 to dispatcher function name.
27062 * multiple_target.c (replace_function_decl): New function.
27063 (create_dispatcher_calls): Redirect both edges and references.
27065 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27067 * profile-count.c (profile_count::dump): Dump quality.
27068 (profile_count::differs_from_p): Update for unsigned val.
27069 * profile-count.h (profile_count_quality): New enum.
27070 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27072 2017-06-19 Richard Biener <rguenther@suse.de>
27074 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27075 struct function as arg.
27076 (estimate_numbers_of_iterations): Export overload with loop arg.
27077 (free_numbers_of_iterations_estimates_loop): Use an overload of
27078 free_numbers_of_iterations_estimates instead.
27079 * tree-cfg.c (remove_bb): Adjust.
27080 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27081 * tree-parloops.c (gen_parallel_loop): Likewise.
27082 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27084 (tree_unroll_loops_completely): Likewise.
27085 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27086 Use an overload instead and export.
27087 (estimated_loop_iterations): Adjust.
27088 (max_loop_iterations): Likewise.
27089 (likely_max_loop_iterations): Likewise.
27090 (estimate_numbers_of_iterations): Take struct function as arg
27092 (loop_exits_before_overflow): Adjust.
27093 (free_numbers_of_iterations_estimates_loop): Use an overload.
27094 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27095 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27097 2017-06-19 Richard Biener <rguenther@suse.de>
27100 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27101 RANGE_EXPR conservatively.
27103 2017-06-16 Carl Love <cel@us.ibm.com>
27105 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27106 definitions for vec_float, vec_float2, vec_floato,
27107 vec_floate built-ins.
27108 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27109 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27111 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27112 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27113 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27114 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27115 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27116 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27117 vec_floato): Add builtin defines.
27118 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27119 Update the built-in documentation file for the new built-in
27122 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27124 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27125 (mthumb): Mark as the negative of -marm.
27127 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27129 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27131 (ARM Options, -mtune): Document that this accepts the same
27132 extension options as -mcpu.
27133 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27135 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27137 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27138 permitted extensions.
27140 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27142 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27143 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27144 (armv8-m.main): Add option +nodsp.
27145 * config/arm/arm-cpu-cdata.h: Regenerated.
27147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27149 * config/arm/t-fuchsia: New file.
27150 * config.gcc (arm*-*-fuchsia*): Use it.
27152 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27154 * config/arm/t-symbian: Rewrite for new option infrastructure.
27156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27158 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27159 (MULTILIB_REQUIRED): Likewise.
27161 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27163 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27164 (MULTILIB_RESUE): Likewise.
27165 (MULTILIB_MATCHES): Likewise.
27166 (MULTLIB_REQUIRED): Likewise.
27168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27170 * config/arm/t-rtems: Rewrite for new option framework.
27172 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27174 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27175 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27176 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27177 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27178 * config/arm/t-multilib: ... here.
27179 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27180 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27181 armv7-a and armv8*-a when A-profile libraries have not been built.
27182 * config/arm/t-rmprofile: Rewrite.
27184 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27186 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27187 with a backslash. Remove the backslash after substituting unescaped
27189 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27191 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27193 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27194 the driver to pass through the default setting of -mfloat-abi.
27195 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27197 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27198 (all_feat_combs): New rule.
27199 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27201 * config/arm/t-aprofile: Rewrite.
27203 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27205 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27206 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27207 fpu is not specified by the user/command-line.
27208 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27209 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27210 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27211 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27212 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27213 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27215 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27217 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27218 * config/arm/t-arm-elf: Rewritten.
27220 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27222 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27223 have some floating-point instructions.
27224 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27225 (TARGET_MAYBE_HARD_FLOAT): New macro.
27226 * config/arm/arm-builtins.c (arm_init_builtins): Use
27227 TARGET_MAYBE_HARD_FLOAT.
27228 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27230 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27232 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27233 (configargs.h): Include it.
27234 (arm_print_hint_for_fpu_option): New function.
27235 (arm_parse_fpu_option): New function.
27236 (candidate_extension): New class.
27237 (arm_canon_for_multilib): New function.
27238 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27239 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27240 (ARCH_CANONICAL_SPECS): New macro.
27241 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27243 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27245 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27246 are set after handling multilib fragments. Set target_cpu_default2
27249 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27251 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27253 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27254 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27256 (check_arch): Likewise.
27257 * config/arm/arm.c (arm_configure_build_target): Handle
27258 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27259 options in the default.
27261 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27263 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27264 when an option is an alias of another.
27265 * config/arm/parsecpu.awk (optalias): New parser token.
27266 (gen_comm_data): Mark non-alias options as such. Emit entries
27267 for extension aliases.
27268 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27269 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27270 (armv6kz, armv6zk, armv6t2): Likewise.
27271 (armv7): Make vfpv3-d16 an alias.
27272 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27274 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27275 Sort in canonical order.
27276 (armv8-a): Sort in canonical order.
27277 (armv8.1-a, armv8.2-a): Likewise.
27278 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27280 (cortex-a9): Sort in canonical order.
27281 * config/arm/arm.c (selftests.h): Include it.
27282 (arm_test_cpu_arch_data): New function.
27283 (arm_run_self_tests): New function.
27284 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27285 (targetm): Move declaration to the end of the file.
27286 * arm-cpu-cdata.h: Regenerated.
27288 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27290 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27291 call to target_mode_check describing the type of option passed.
27292 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27293 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27294 arm_parse_cpu_option_name to match parameters against list of
27296 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27297 arm_arch_core_flags data structure.
27298 * config/arm/arm-cpu_cdata.h: Regenerated.
27300 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27302 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27304 (arm_print_hint_for_cpu_option): Likewise.
27305 (arm_print_hint_for_arch_option): Likewise.
27306 (arm_parse_cpu_option_name): Likewise.
27307 (arm_parse_arch_option_name): Likewise.
27308 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27309 of entries in the all_fpus list.
27310 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27311 (arm_parse_cpu_option_name): Declare.
27312 (arm_parse_arch_option_name): Declare.
27313 (arm_parse_option_features): Declare.
27314 (arm_intialize_isa): Declare.
27315 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27317 (gen_comm_data): ... here. Make definitions non-static.
27318 * config/arm/arm-cpu-data.h: Regenerated.
27319 * config/arm/arm-cpu-cdata.h: Regenerated.
27321 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27323 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27324 (cpu_arch_extension): New structure.
27325 (cpu_arch_option, arch_option, cpu_option): New structures.
27326 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27327 architecture types.
27328 (gen_data): Generate new format data tables.
27329 * config/arm/arm.c (cpu_tune): New structure.
27330 (cpu_option, processors): Delete.
27331 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27332 (arm_print_hint_for_cpu_option): ... this and ...
27333 (arm_print_hint_for_arch_option): ... this.
27334 (arm_parse_arch_cpu_name): Delete. Replace with ...
27335 (arm_parse_cpu_option_name): ... this and ...
27336 (arm_parse_arch_option_name): ... this.
27337 (arm_unrecognized_feature): Change type of target parameter to
27339 (arm_parse_arch_cpu_features): Delete. Replace with ...
27340 (arm_parse_option_features): ... this.
27341 (arm_configure_build_target): Rework to use new configuration data
27343 (arm_print_tune_info): Rework for new configuration data tables.
27344 * config/arm/arm-cpu-data.h: Regenerated.
27345 * config/arm/arm-cpu.h: Regenerated.
27347 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27349 * Makefile.in (OBJS): Move sbitmap.o from here ...
27350 (OBJS-libcommon): ... to here.
27352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27354 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27355 (ISA_ALL_CRYPTO): New macro.
27356 (ISA_ALL_SIMD): New macro
27357 (ISA_ALL_FP): New macro.
27358 * config/arm/arm.c (fpu_bitlist): Update initializer.
27359 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27361 (arm9e): Add fpu. Add option for nofp
27362 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27363 (arm926ej-s, arm1026ej-s): Likewise.
27364 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27365 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27366 neon-fp16, neon-vfpv4, nofp and nosimd.
27367 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27368 (cortex-a8): Add fpu. Add option for nofp.
27369 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27370 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27371 (cortex-r4f): Add fpu.
27372 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27373 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27375 (cortex-r8): Likewise.
27376 (cortex-m4): Add fpu. Add option for nofp.
27377 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27378 (cortex-a17.cortex-a7): Likewise.
27379 (cortex-a32): Add fpu. Add options for crypto and nofp.
27380 (cortex-a35, cortex-a53): Likewise.
27381 (cortex-a57): Add fpu. Add option for crypto.
27382 (cortex-a72, cortex-a73): Likewise.
27383 (exynos-m1): Likewise.
27384 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27385 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27386 (cortex-m33): Add fpu. Add option for nofp.
27387 * config/arm/arm-cpu-cdata.h: Regenerated
27388 * config/arm/arm-cpu-data.h: Regenerated.
27390 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27392 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27393 (armv5te, armv5tej): Likewise.
27394 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27395 (armv7): Add options fp and vfpv3-d16.
27396 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27397 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27399 (armv7ve): Likewise.
27400 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27401 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27402 (armv8-a): Add nocrypto option.
27403 (armv8.1-a, armv8.2-a): Likewise.
27404 (armv8-m.main): add options fp, fp.dp and nofp.
27406 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27408 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27410 (armv8-a+crc): Delete.
27411 (armv8.1-a): Add options simd, crypto and nofp.
27412 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27413 (armv8.2-a+fp16): Delete.
27414 (armv8-m.main): Add option dsp.
27415 (armv8-m.main+dsp): Delete.
27416 (cortex-a8): Add fpu. Add nofp option.
27417 (cortex-a9): Add fpu. Add nofp and nosimd options.
27418 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27419 link to main cpu and architecture data structures.
27420 (gen_comm_data): Only put isa attributes from the main architecture
27422 (option): New statement for architecture and CPU entries.
27423 * arm.c (struct cpu_option): New structure.
27424 (struct processors): Add entry for options.
27425 (arm_unrecognized_feature): New function.
27426 (arm_parse_arch_cpu_name): Ignore any characters after the first
27428 (arm_parse_arch_cpu_feature): New function.
27429 (arm_configure_build_target): Separate out any CPU and architecture
27430 features and parse separately. Don't error out if -mfpu=auto is
27431 used with only an architecture string.
27432 (arm_print_asm_arch_directives): New function.
27433 (arm_file_start): Call it.
27434 * config/arm/arm-cpu-cdata.h: Regenerated.
27435 * config/arm/arm-cpu-data.h: Likewise.
27436 * config/arm/arm-tables.opt: Likewise.
27438 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27440 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27441 assembler when it is not -mfpu=auto.
27443 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27445 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27446 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27447 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27448 (ASM_CPU_SPEC): Rewrite.
27449 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27450 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27451 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27452 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27453 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27454 copied string is NUL-terminated. Also strip any characters prefixed
27456 (arm_rewrite_selected_arch): New function.
27457 (arm_rewrite_march): New function.
27459 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27461 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27462 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27463 (march, mcpu, mtune): Convert to string-based options.
27464 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27465 (arm_parse_arch_cpu_name): New function.
27466 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27467 identify selected architecture or CPU.
27468 (arm_option_save): New function.
27469 (TARGET_OPTION_SAVE): Redefine.
27470 (arm_option_restore): Restore string options.
27471 (arm_option_print): Print string options.
27473 2017-06-16 Martin Sebor <msebor@redhat.com>
27475 PR tree-optimization/80933
27476 PR tree-optimization/80934
27477 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27478 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27479 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27480 (gimple_fold_builtin): Call them.
27482 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27484 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27485 as unlikely; update profile.
27487 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27489 * predict.c (force_edge_cold): Handle declaring edges impossible
27492 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27494 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27496 (try_unroll_loop_completely): Fix reporting.
27498 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27500 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27502 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27505 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27506 if given a non-constant argument for an intrinsic which requires a
27509 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27511 * profile.c (compare_freqs): New function.
27512 (branch_prob): Sort edge list.
27513 (find_spanning_tree): Assume that the list is priority sorted.
27515 2017-06-16 Richard Biener <rguenther@suse.de>
27517 PR tree-optimization/81090
27518 * passes.def (pass_record_bounds): Remove.
27519 * tree-pass.h (make_pass_record_bounds): Likewise.
27520 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27521 make_pass_record_bounds): Likewise.
27522 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27523 not free niter estimates at the beginning but at the end.
27524 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27526 2017-06-16 Richard Biener <rguenther@suse.de>
27528 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27529 initializer to workaround ICE in host GCC 4.8.
27531 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27533 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27535 (clone_inlined_nodes): Update.
27537 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27539 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27540 prefetch settings, and enable prefetching by default at -O3.
27542 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27544 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27545 Set flag_prefetch_loop_arrays according to tuning data.
27547 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27549 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27550 New tune structure.
27551 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27552 [Unrelated to main purpose of the patch] Place the pointer field last
27553 to enable type checking errors when tune structure are wrongly merged.
27554 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27555 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27556 (thunderx2t99_prefetch_tune): New tune constants.
27557 (tune_params *_tunings): Update all tunings (no functional change).
27558 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27559 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27560 from tunings structures.
27562 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27565 * ubsan.c (instrument_null): Add T argument, use it instead
27566 of computing it based on IS_LHS.
27567 (instrument_object_size): Likewise.
27568 (pass_ubsan::execute): Adjust instrument_null and
27569 instrument_object_size callers to pass gimple_get_lhs or
27570 gimple_assign_rhs1 result to it. Use instrument_null instead of
27571 calling get_base_address and instrument_mem_ref. Handle
27572 aggregate call arguments for object-size sanitization.
27574 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27576 PR tree-optimization/81089
27577 * tree-vrp.c (is_masked_range_test): Validate operands of
27580 2017-06-15 Martin Sebor <msebor@redhat.com>
27583 * dumpfile.c (dump_register): Avoid calling memset to initialize
27584 a class with a default ctor.
27585 * gcc.c (struct compiler): Remove const qualification.
27586 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27587 * hash-table.h: Ditto.
27588 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27590 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27591 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27593 * params.h (struct param_info): Make struct members non-const.
27594 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27595 with default initialization.
27596 * vec.h (vec_copy_construct, vec_default_construct): New helper
27598 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27599 with vec_copy_construct.
27600 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27601 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27602 * doc/invoke.texi (-Wclass-memaccess): Document.
27604 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27606 * emit-rtl.h (is_leaf): Update comment about local
27607 register allocator.
27609 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27612 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27613 for a variable to have a section before checking if the section has a
27615 Set section to.persistent if persistent attribute is set.
27616 Warn if .persistent attribute is used on an automatic variable.
27618 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27620 PR rtl-optimization/80474
27621 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27623 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27625 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27628 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27630 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27631 (casesi): Emit bounds checking as RTL.
27632 (casesi_internal_mips16_<mode>): Remove bounds checking.
27634 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27636 * config/xtensa/xtensa.c (xtensa_option_override): Append
27637 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27638 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27639 xtensa_doloop_hooks): Define unconditionally.
27640 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27642 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27643 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27644 for it in xtensa_option_override.
27645 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27646 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27648 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27650 * doc/cppopts.texi: Document '-' special value to -MF.
27652 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27654 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27655 (cortex_a53_fconst): Likewise.
27656 (cortex_a53_fpmul): Likewise.
27657 (cortex_a53_f_load_64): Likewise.
27658 (cortex_a53_f_load_many): Likewise.
27659 (cortex_a53_advsimd_alu): Likewise.
27660 (cortex_a53_advsimd_alu_q): Likewise.
27661 (cortex_a53_advsimd_mul): Likewise.
27662 (cortex_a53_advsimd_mul_q): Likewise.
27663 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27664 Add missing fmul, r2f_cvt and fconst cases.
27666 2017-06-14 Richard Biener <rguenther@suse.de>
27668 PR middle-end/81088
27669 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27671 (fold_binary_loc): When associating do not treat pre-existing
27672 TREE_OVERFLOW on literal constants as a reason to allow
27673 TREE_OVERFLOW on associated literal constants.
27675 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27677 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27678 (MASK_FEATURES): New macro.
27679 * config/sparc/sparc.c (sparc_option_override): Remove the special
27680 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27682 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27684 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27685 a division of 0 if non-call exceptions are enabled.
27687 2017-06-14 Andrew Pinski <apinski@cavium.com>
27688 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27691 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27692 Improve vector initialization code gen for only variable case.
27694 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27696 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27698 2017-06-14 Richard Biener <rguenther@suse.de>
27700 PR tree-optimization/81083
27701 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27704 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27706 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27707 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27708 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27709 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27710 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27711 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27713 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27715 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27716 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27718 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27720 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27722 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27724 * config/rs6000/t-rtems: Don't handle SPE.
27726 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27728 * config/rs6000/t-linux: Don't handle SPE.
27730 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27732 * config/rs6000/eabispe.h: Delete file.
27734 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27736 * config/rs6000/t-spe: Delete file.
27738 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27740 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27741 (rs6000_legitimate_offset_address_p): Return false for anything in
27742 V2SImode or V2SFmode.
27744 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27746 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27747 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27748 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27750 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27751 (rs6000_legitimate_offset_address_p): Ditto.
27752 (rs6000_emit_move): Ditto.
27753 (rs6000_init_builtins): Remove V4HI_type_node.
27755 2017-06-13 Martin Liska <mliska@suse.cz>
27758 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27759 (gate_asan): Likewise.
27760 * asan.h (asan_no_sanitize_address_p): Remove the function.
27761 (sanitize_flags_p): New function.
27762 * builtins.def: Fix coding style.
27763 * common.opt: Use renamed enum value.
27764 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27765 * doc/extend.texi: Document no_sanitize attribute.
27766 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27767 to SANITIZE_UNDEFINED_NONDEFAULT.
27768 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27769 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27770 Use sanitize_flags_p.
27771 * gimplify.c (gimplify_function_tree): Likewise.
27772 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27773 * opts.c (parse_no_sanitize_attribute): New function.
27774 (common_handle_option): Use renamed enum value.
27775 * opts.h (parse_no_sanitize_attribute): Declare.
27776 * tree.c (sanitize_flags_p): New function.
27777 * tree.h: Declared here.
27778 * tsan.c: Use sanitize_flags_p.
27779 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27780 (instrument_mem_ref): Likewise.
27781 (instrument_bool_enum_load): Likewise.
27782 (do_ubsan_in_current_function): Remove the function.
27783 (pass_ubsan::execute): Use sanitize_flags_p.
27784 * ubsan.h: Remove do_ubsan_in_current_function
27785 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27786 (dump_function_to_file): Use it here.
27788 2017-06-13 Martin Jambor <mjambor@suse.cz>
27790 PR tree-optimization/80803
27791 PR tree-optimization/81063
27792 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27793 (propagate_subaccesses_across_link): Enqueue subtree whenever
27794 necessary instead of relying on the caller.
27796 2017-06-13 Martin Jambor <mjambor@suse.cz>
27798 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27799 that have a first_link.
27800 (sort_and_splice_var_accesses): Do not check first_link before
27802 (subtree_mark_written_and_enqueue): Likewise.
27803 (propagate_all_subaccesses): Likewise and do not stop at first
27804 parent with a first_link.
27806 2017-06-13 Martin Jambor <mjambor@suse.cz>
27808 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27811 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27813 * match.pd: New pattern.
27815 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27817 * tree-vrp.c (is_masked_range_test): New function.
27818 (register_edge_assert_for): Determine ranges for
27821 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27823 PR tree-optimization/67328
27824 * fold-const.c (maskable_range_p): New function.
27825 (build_range_check): Generate bittests if possible.
27827 2017-06-13 Martin Liska <mliska@suse.cz>
27829 * gimple-pretty-print.c (dump_probability): Add new argument.
27830 (dump_edge_probability): Dump both probability and count.
27831 (dump_gimple_label): Likewise.
27832 (dump_gimple_bb_header): Likewise.
27834 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27837 * config/avr/avr-devices.c: Fix indentation.
27838 * config/avr/gen-avr-mmcu-specs.c: Dito.
27840 2017-06-13 Richard Biener <rguenther@suse.de>
27842 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27843 instead get vector type from stmt_info.
27844 (vectorizable_reduction): Adjust. Remove dead code.
27846 2017-06-13 Richard Biener <rguenther@suse.de>
27848 PR middle-end/81065
27849 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27850 case of C * (x * C2 + C3).
27851 (fold_addr_of_array_ref_difference): Properly fold index difference.
27853 2017-06-12 David S. Miller <davem@davemloft.net>
27856 * config/sparc/sparc.md (return expander): Emit frame blockage if
27857 function uses alloca.
27859 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27861 * combine.c (make_field_assignment): Check len rather than the mode
27862 precision when calling force_to_mode.
27864 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27866 Support multilibs and devices that see flash in RAM address range.
27869 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27870 (avr_mcu_t) <flash_pm_offset>: New field.
27871 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27872 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27873 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27874 (AVR_TINY_PM_OFFSET): Remove macro.
27875 * config/avr/avr.opt (-mshort-calls): New option.
27876 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27877 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27878 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27879 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27880 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27881 instead of avr_arch->have_jmp_call.
27882 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27883 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27884 avr_arch->flash_pm_offset to define.
27885 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27886 new field flash_pm_offset. Add entry for avrxmega3.
27887 (avr_texinfo): Add entry for avrxmega3.
27888 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27889 attiny212, attiny214,
27890 attiny412, attiny414, attiny416, attiny417,
27891 attiny814, attiny816, attiny817,
27892 attiny1614, attiny1616, attiny1617,
27893 attiny3214, attiny3216, attiny3217.
27894 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27895 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27896 (avr_print_operand_address) [AVR_TINY]: Same.
27897 (avr_asm_init_sections) <readonly_data_section>: Only patch
27898 callback if avr_arch->flash_pm_offset = 0.
27899 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27900 for rodata if avr_arch->flash_pm_offset != 0.
27901 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27902 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27903 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27904 (m_options): Append opt_rcall.
27905 (m_dirnames): Append dir_rcall.
27906 * config/avr/t-multilib: Regenerate.
27908 * configure.ac [target=avr]: Check whether avrxmega3 default
27909 linker description file works as needed.
27910 * configure: Regenerate.
27911 * doc/avr-mmcu.texi: Regenerate.
27912 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27913 <__AVR_ARCH__>: Document avrxmega3 and 103.
27914 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27915 <__AVR_SHORT_CALLS__>: Document it.
27916 <__AVR_PM_BASE_ADDRESS__>: Document it.
27917 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27918 (AVR Variable Attributes) <progmem>: Document this is
27919 not needed for avrxmega3.
27920 (AVR Named Address Spaces) <__flash>: Dito.
27922 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27924 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27926 2017-06-12 Doug Rupp <rupp@adacore.com>
27928 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27929 Append vxworks-stdint.h to the tm_file list.
27930 * config/vxworks-stdint.h: New file.
27932 2017-06-12 Martin Liska <mliska@suse.cz>
27934 PR tree-optimization/81041
27935 * tree-profile.c (gimple_gen_ic_func_profiler):
27936 Create an extra BB in profile-generate
27937 (gimple_gen_time_profiler): Likewise.
27939 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27941 PR tree-optimization/81003
27942 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27943 (update_range_test): Use it instead of force_gimple_operand_gsi.
27945 2017-06-12 Richard Biener <rguenther@suse.de>
27947 PR tree-optimization/81053
27948 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27949 with backedge value not defined in loop. Simplify def stmt
27952 2017-06-11 Tom de Vries <tom@codesourcery.com>
27955 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27957 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27958 nvptx_cannot_force_const_mem.
27960 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27962 * opts.c (finish_options): Move test for flag_split_stack after
27963 it has been initialized.
27965 2017-06-11 Jason Merrill <jason@redhat.com>
27967 * tree.h (id_equal): New.
27968 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27969 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27970 instead of strcmp of IDENTIFIER_POINTER.
27972 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27974 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27975 (mark_all_inlined_calls_cdtor): Fix formating.
27976 (inline_transform): Rescale profile before inlining.
27978 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27980 * cgraph.h (cgraph_edge::clone): Update prototype.
27981 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27982 (cgraph_node::create_clone): Update.
27983 (cgraph_node::create_version_clone): Update.
27984 * tree-inline.c (copy_bb): Update.
27985 (expand_call_inline): Update.
27987 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27989 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27990 factored out from ...
27991 (rs6000_emit_prologue): ... here.
27993 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27995 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27996 factored out from ...
27997 (rs6000_emit_prologue): ... here.
27999 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28001 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28003 (handle_missing_profiles): Fix computation of tp_first_run.
28004 (counts_to_freqs): Do not touch freqs when count is 0.
28006 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28008 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28011 2017-06-10 Tom de Vries <tom@codesourcery.com>
28013 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28014 attributes): Document signal effective target.
28016 2017-06-10 Tom de Vries <tom@codesourcery.com>
28018 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28019 Document effective target stack_size.
28021 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28023 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28024 to the edit_context if they can be auto-applied.
28026 2017-06-9 Ian Lance Taylor <iant@golang.org>
28028 * opts.c (finish_options): If -fsplit-stack, disable implicit
28029 -forder-blocks-and-partition.
28030 * doc/invoke.texi (Optimize Options): Document that when using
28031 -fsplit-stack -forder-blocks-and-partition is not implicitly
28034 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28036 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28037 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28038 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28039 * builtins.def (abort, trap, unreachable): Declare cold.
28040 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28041 * tree-core.h (ECF_COLD): New.
28042 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28043 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28045 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28047 * predict.c (unlikely_executed_stmt_p): Cleanup.
28049 2017-06-09 Richard Biener <rguenther@suse.de>
28051 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28052 model if the ref is always written to.
28054 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28056 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28058 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28060 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28063 2017-06-09 Tom de Vries <tom@codesourcery.com>
28066 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28067 "target cannot support label values" when encountering LABEL_REF.
28069 2017-06-09 Martin Liska <mliska@suse.cz>
28071 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28072 (gimple_gen_ic_func_profiler): Emit direct comparison
28073 of __gcov_indirect_call_callee with NULL.
28074 (gimple_gen_time_profiler): Change probability from
28075 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28077 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28079 * profile.c (edge_gcov_counts): Turn to pointer.
28080 (compute_branch_probabilities, compute_branch_probabilities): Update.
28081 (branch_prob): Do not clear edge_gcov_count.
28082 * profile.h (edge_gcov_counts): Turn to pointer.
28083 (edge_gcov_count): Update.
28085 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28087 * gimple.h (gimple_check_failed): Mark cold.
28089 2017-06-09 Richard Biener <rguenther@suse.de>
28091 PR tree-optimization/66623
28092 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28093 refactor check_reduction into two parts, properly computing
28094 whether we have to check reduction validity for outer loop
28097 2017-06-09 Richard Biener <rguenther@suse.de>
28099 PR tree-optimization/79483
28100 * graphite-scop-detection.c (order): New global.
28101 (get_order): Compute bb to order mapping that satisfies code
28102 generation constraints.
28103 (cmp_pbbs): New helper.
28104 (build_scops): Start domwalk at entry block, sort generated
28107 2017-06-09 Richard Biener <rguenther@suse.de>
28109 PR middle-end/81007
28110 * ipa-polymorphic-call.c
28111 (ipa_polymorphic_call_context::restrict_to_inner_class):
28112 Skip FIELD_DECLs with error_mark_node type.
28113 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28116 2017-06-09 Martin Liska <mliska@suse.cz>
28118 * predict.c (struct branch_predictor): New struct.
28119 (test_prediction_value_range): New test.
28120 (predict_c_tests): New function.
28121 * selftest-run-tests.c (selftest::run_tests): Run the function.
28122 * selftest.h: Declare new tests.
28124 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28127 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28128 gen_add3_insn did not fail.
28129 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28130 r0, construct that number in a temporary reg and add that reg to r0.
28131 If asked to put the result in r0 as well, fail.
28133 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28135 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28136 for early expansion of vec_eqv.
28138 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28140 PR middle-end/81005
28141 * ubsan.c (instrument_null): Avoid pointless code temporary.
28142 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28144 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28148 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28151 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28154 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28156 * predict.c (unlikely_executed_edge_p): Move ahead.
28157 (probably_never_executed_edge_p): Use it.
28159 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28161 PR middle-end/79988
28162 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28163 gimple_call_builtin_p call.
28165 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28167 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28168 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28169 rtl_check_failed_type2, rtl_check_failed_code1,
28170 rtl_check_failed_code2, rtl_check_failed_code_mode,
28171 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28172 rtvec_check_failed_bounds, rtl_check_failed_flag,
28173 _fatal_insn_not_found, _fatal_insn): Likewise.
28174 * tree.h (tree_contains_struct_check_failed,
28175 tree_check_failed, tree_not_check_failed,
28176 tree_class_check_failed, tree_range_check_failed,
28177 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28178 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28179 tree_operand_check_failed, omp_clause_check_failed,
28180 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28183 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28185 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28186 flag_branch_probabilities.
28187 * ipa-inline.c (edge_badness): Likewise.
28188 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28189 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28190 * predict.c (maybe_hot_frequency_p): Likewise.
28191 (probably_never_executed): Likewise.
28192 * sched-ebb.c (schedule_ebbs): Likewise.
28193 * sched-rgn.c (find_single_block_region): Likewise.
28194 * tracer.c (tail_duplicate): Likewise.
28196 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28198 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28199 longer requires x_flag_profile_use.
28201 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28203 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28204 instead of flag_reorder_blocks_and_partition.
28205 * dbxout.c (dbxout_function_end): Likewise.
28206 * dwarf2out.c (gen_subprogram_die): Likewise.
28207 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28208 * hw-doloop.c (reorg_loops): Likewise.
28209 * varasm.c (assemble_start_function,
28210 assemble_end_function): Likewise.
28211 (decide_function_section): Do not check for
28212 flag_reorder_blocks_and_partition.
28214 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28216 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28218 (chkp_get_hard_register_fake_addr_expr): Ditto.
28219 (chkp_build_addr_expr): Add check for hard reg case.
28220 (chkp_parse_array_and_component_ref): Ditto.
28221 (chkp_find_bounds_1): Ditto.
28222 (chkp_process_stmt): Don't generate bounds store for
28225 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28227 * predict.c (maybe_hot_bb_p): Do not check profile status.
28228 (maybe_hot_edge_p): Likewise.
28229 (probably_never_executed): Check for zero counts even if profile
28231 (unlikely_executed_edge_p): New function.
28232 (unlikely_executed_stmt_p): New function.
28233 (unlikely_executed_bb_p): New function.
28234 (set_even_probabilities): Use unlikely predicates.
28235 (combine_predictions_for_bb): Likewise.
28236 (predict_paths_for_bb): Likewise.
28237 (predict_paths_leading_to_edge): Likewise.
28238 (determine_unlikely_bbs): New function.
28239 (estimate_bb_frequencies): Use it.
28240 (compute_function_frequency): Use zero counts even if profile is
28242 * profile-count.h: Fix typo.
28244 2017-08-08 Julia Koval <julia.koval@intel.com>
28246 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28247 _mm512_mask_cvtsepi16_storeu_epi8,
28248 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28249 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28250 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28251 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28252 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28253 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28254 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28255 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28256 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28257 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28258 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28259 __builtin_ia32_pmovuswb256mem_mask,
28260 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28261 __builtin_ia32_pmovwb512mem_mask): New builtins.
28263 2017-08-08 Julia Koval <julia.koval@intel.com>
28265 PR target/73350,80862
28266 * config/i386/subst.md (round): Fix round pattern.
28267 * config/i386/i386.c (ix86_erase_embedded_rounding):
28268 Fix erasing rounding for the fixed pattern.
28270 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28272 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28274 2017-06-08 Martin Liska <mliska@suse.cz>
28276 PR gcov-profile/80911
28277 * gcov.c (block_info::block_info): New constructor.
28279 2017-06-07 Carl Love <cel@us.ibm.com>
28281 * config/rs6000/rs6000-c: The return type of the following
28282 built-in functions was implemented as int not long long. Fix sign
28283 of return value for the unsigned version of vec_mulo and vec_mule.
28284 vector unsigned long long vec_bperm (vector unsigned long long,
28285 vector unsigned char)
28286 vector signed long long vec_mule (vector signed int,
28288 vector unsigned long long vec_mule (vector unsigned int,
28289 vector unsigned int)
28290 vector signed long long vec_mulo (vector signed int,
28292 vector unsigned long long vec_mulo (vector unsigned int,
28293 vector unsigned int)
28294 * doc/extend.texi: Fix the documentation for the built-in
28297 2017-06-07 Carl Love <cel@us.ibm.com>
28300 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28303 2017-06-07 Carl Love <cel@us.ibm.com>
28305 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28306 support, Generate doublehv for signed int/float for BE case only.
28308 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28310 * doc/invoke.texi (mcx16): Rewrite.
28312 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28314 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28315 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28316 *mov<mode>_softfloat, and an anonymous splitter): Use
28317 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28319 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28321 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28323 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28324 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28325 (rs6000_debug_reg_global): Adjust.
28326 (rs6000_init_hard_regno_mode_ok): Adjust.
28327 (rs6000_dbx_register_number): Adjust.
28328 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28329 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28330 Remove SPE_ACC and SPEFSCR.
28331 (REG_ALLOC_ORDER): Ditto.
28332 (FRAME_POINTER_REGNUM): Change to 111.
28333 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28334 (REG_CLASS_NAMES): Ditto.
28335 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28336 (REGISTER_NAMES): Ditto.
28337 (ADDITIONAL_REG_NAMES): Ditto.
28338 (rs6000_reg_names): Ditto.
28339 * config/rs6000/rs6000.md: Renumber some register number
28342 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28344 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28346 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28347 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28349 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28350 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28351 Delete the SPE high registers.
28352 (REG_ALLOC_ORDER): Ditto.
28353 (enum reg_class): Remove SPE_HIGH_REGS.
28354 (REG_CLASS_NAMES): Ditto.
28355 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28356 (REGISTER_NAMES): Ditto.
28357 (rs6000_reg_names): Ditto.
28358 * doc/tm.texi.in: Remove SPE as example.
28359 * doc/tm.texi: Regenerate.
28361 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28363 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28364 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28365 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28366 * config/rs6000/rs6000.md (type): Remove "brinc".
28368 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28370 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28371 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28372 * config/rs6000/linuxspe.h: Delete file.
28373 * config/rs6000/rs6000.md: Don't include spe.md.
28374 * config/rs6000/spe.h: Delete file.
28375 * config/rs6000/spe.md: Delete file.
28376 * config/rs6000/t-rs6000: Remove spe.md.
28378 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28380 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28381 (reg_or_none500mem_operand): Delete.
28382 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28383 instead of reg_or_none500mem_operand.
28385 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28387 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28388 handling of SPE flags.
28389 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28391 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28393 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28395 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28396 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28397 paired_divv2sf3): Similar.
28398 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28399 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28400 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28402 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28403 Rename the paired_* instruction patterns.
28404 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28406 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28407 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28408 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28409 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28410 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28411 PAIRED_VECTOR_MODE.
28412 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28413 (spe_func_has_64bit_regs_p): Delete.
28414 (spe_expand_predicate_builtin): Delete.
28415 (spe_expand_evsel_builtin): Delete.
28416 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28417 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28418 (invalid_e500_subreg): Delete.
28419 (rs6000_legitimize_address): Always force_reg op2 as well, for
28420 paired single memory accesses.
28421 (rs6000_member_type_forces_blk): Delete.
28422 (rs6000_spe_function_arg): Delete.
28423 (rs6000_expand_unop_builtin): Delete SPE handling.
28424 (rs6000_expand_binop_builtin): Ditto.
28425 (spe_expand_stv_builtin): Delete.
28426 (bdesc_2arg_spe): Delete.
28427 (spe_expand_builtin): Delete.
28428 (spe_expand_predicate_builtin): Delete.
28429 (spe_expand_evsel_builtin): Delete.
28430 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28431 (spe_init_builtins): Delete.
28432 (spe_func_has_64bit_regs_p): Delete.
28433 (savres_routine_name): Delete "info" parameter. Adjust callers.
28434 (rs6000_emit_stack_reset): Ditto.
28435 (rs6000_dwarf_register_span): Delete.
28436 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28437 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28438 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28440 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28442 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28443 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28444 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28445 mulv2sf3, divv2sf3): Delete expanders.
28447 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28449 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28451 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28453 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28454 * config/rs6000/rs6000.c: Ditto.
28456 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28458 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28459 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28460 comparison_operator.
28462 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28464 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28465 * config/rs6000/rs6000.opt: Ditto.
28466 * config/rs6000/t-rtems: Ditto.
28468 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28470 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28471 TARGET_E500_SINGLE by 0, simplify.
28472 * config/rs6000/rs6000.c: Ditto.
28473 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28474 (spe_build_register_parallel): Delete.
28475 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28476 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28477 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28478 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28479 (E500_CONVERT): Delete.
28480 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28482 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28484 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28485 * config/rs6000/dfp.md: Ditto.
28486 (negdd2, *negdd2_fpr): Merge.
28487 (absdd2, *absdd2_fpr): Merge.
28488 (negtd2, *negtd2_fpr): Merge.
28489 (abstd2, *abstd2_fpr): Merge.
28490 * config/rs6000/e500.h: Delete file.
28491 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28492 TARGET_FPRS by 1 and simplify.
28493 * config/rs6000/rs6000-c.c: Ditto.
28494 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28495 TARGET_DF_SPE by 0.
28496 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28498 * config/rs6000/rs6000.md: Ditto.
28499 (floatdidf2, *floatdidf2_fpr): Merge.
28500 (move_from_CR_gt_bit): Delete.
28501 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28502 (E500_CR_IOR_COMPARE): Delete.
28503 (All patterns that require !TARGET_FPRS): Delete.
28504 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28506 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28508 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28510 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28512 * graphds.c (add_edge): Intitialize edge's attached data.
28513 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28514 pointer parameter. Call pointed function on each edge during
28515 graph traversing. Skip traversing the edge when the function
28517 (graphds_dfs, graphds_scc): Ditto.
28518 (for_each_edge): New parameter. Pass the new parameter to callback
28520 * graphds.h (skip_edge_callback): New function pointer type.
28521 (graphds_dfs, graphds_scc): New function pointer parameter.
28522 (graphds_edge_callback, for_each_edge): New parameter.
28524 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28526 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28527 out code checking if runtime alias check is possible to below ...
28528 Call the new function.
28529 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28530 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28532 2017-06-07 Marek Polacek <polacek@redhat.com>
28535 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28536 TYPE_OVERFLOW_WRAPS check.
28538 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28540 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28541 if versioning is required.
28542 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28543 peeling with the check for versioning.
28545 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28547 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28548 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28549 Set true to new parameter if new ssa variable is defined.
28550 (vect_gen_vector_loop_niters): Refactor. Set range information
28551 for the new vector loop bound variable.
28552 (vect_do_peeling): Ditto.
28554 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28556 * tree-affine.c (ssa.h): Include header file.
28557 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28558 has wrapping overflow behavior.
28560 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28562 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28564 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28566 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28567 (tree_to_aff_combination): ... here.
28569 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28571 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28572 reg_pressure model function.
28573 (ivopts_global_cost_for_size): Delete.
28574 (determine_set_costs, iv_ca_recount_cost): Call new model function
28575 ivopts_estimate_reg_pressure.
28577 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28579 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28580 expensive than udiv. Remove floating point cases from mod.
28582 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28584 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28585 Increase idiv cost.
28587 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28589 * config/aarch64/aarch64.md
28590 (copysignsf3): Fix mask generation.
28592 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28594 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28596 (class dump_manager): Add register_dumps method.
28597 * dumpfile.c: Include langhooks.h.
28598 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28599 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28600 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28601 (dump_manager::dump_register): Start with 512 entries instead of 32.
28602 (dump_manager::register_dumps): New method.
28603 * toplev.c (general_init): Instead of invoking register_dumps
28604 langhook, invoke register_dumps method on the dump manager.
28605 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28608 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28610 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28611 Say that # requires an associated define_split to exist, and that
28612 the define_split must be suitable for use after register allocation.
28614 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28616 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28617 (compute_outgoing_frequencies): Also initialize zero counts.
28618 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28619 around loops; preserve more of profile when nothing changes.
28621 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28623 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28625 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28626 * config/arm/arm-cpu-cdata.h: Regenerate.
28627 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28628 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28629 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28630 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28631 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28633 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28635 * config/arm/t-rmprofile: Likewise.
28636 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28638 2017-06-06 David S. Miller <davem@davemloft.net>
28641 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28642 blockage if function uses alloca.
28644 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28646 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28647 New "uid" fields to hold pretty-print IDs of group and ref.
28648 Memory references are now identified as <group_id>:<ref_id>
28649 instead of using [random] addresses.
28650 (dump_mem_details): Simplify, no functional change.
28651 (dump_mem_ref): Simplify and make output more concise.
28652 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28653 (find_or_create_group): Initialize group uid.
28654 (record_ref): Initialize ref uid. Improve debug output.
28655 (prune_group_by_reuse, should_issue_prefetch_p,)
28656 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28657 (mark_nontemporal_store, determine_loop_nest_reuse):
28658 Improve debug output.
28660 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28662 * dbgcnt.def (prefetch): New debug counter.
28663 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28664 (schedule_prefetches): Stop issueing prefetches if debug counter
28667 2017-06-06 Tom de Vries <tom@codesourcery.com>
28669 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28670 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28672 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28674 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28675 Use aarch64_reg_or_zero predicate for operand 4.
28676 (aarch64_compare_and_swap<mode> define_insn_and_split):
28677 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28678 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28680 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28682 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28683 (arm_compute_save_core_reg_mask): This.
28684 (thumb1_compute_save_reg_mask): Rename into ...
28685 (thumb1_compute_save_core_reg_mask): This.
28686 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28687 (arm_compute_frame_layout): Likewise.
28689 2017-06-06 Richard Biener <rguenther@suse.de>
28691 PR tree-optimization/80974
28692 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28693 keep or clear leaders SSA info.
28695 2017-06-06 Tom de Vries <tom@codesourcery.com>
28697 * config/nvptx/nvptx.c (split_mode_p): New function.
28698 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28700 2017-06-06 Tom de Vries <tom@codesourcery.com>
28702 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28704 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28707 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28710 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28712 * shrink-wrap.c (handle_simple_exit): Update profile.
28713 (try_shrink_wrapping): Upate profile.
28715 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28717 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28718 (tree_guess_outgoing_edge_probabilities): New.
28719 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28720 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28722 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28724 * ipa-split.c (split_function): Initialize return bb profile.
28726 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28728 * profile.c (compute_branch_probabilities): Also initialize
28729 EXIT_BLOCK profile.
28731 2017-06-06 Richard Biener <rguenther@suse.de>
28733 PR tree-optimization/80928
28734 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28735 (vect_analyze_loop_operations): Properly guard analysis for
28737 (vect_transform_loop): Likewise.
28738 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28739 (vect_model_induction_cost): Do not cost for pure SLP.
28740 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28741 of induction in inner loop vectorization.
28742 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28743 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28744 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28746 (vect_analyze_slp_cost_1): Cost induction.
28747 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28748 (vect_get_slp_vect_defs): Likewise.
28749 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28750 (vect_transform_stmt): Handle SLP reductions.
28751 * tree-vectorizer.h (vectorizable_induction): Adjust.
28753 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28755 * config/rs6000/rs6000.c (make_resolver_func): Update
28756 init_lowered_empty_function call.
28758 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28760 * doc/invoke.texi: Document the -fprofile-abs-path option.
28761 * common.opt (fprofile-abs-path): New option.
28762 * gcov-io.h (gcov_write_filename): Declare.
28763 * gcov-io.c (gcov_write_filename): New function.
28764 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28765 * profile.c (output_location): Likewise.
28767 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28769 * shring-wrap.c: Revert accidental commit.
28771 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28773 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28775 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28777 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28779 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28780 profile in callgraph edge.
28781 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28782 (apply_scale): Likewise.
28783 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28784 Also scale profile when inlining function with zero profile.
28785 (initialize_cfun): Update exit block profile even when it is zero.
28786 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28787 when profile is read.
28789 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28791 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28792 (CLONE_*): New constants to define the processors we can generate
28793 code for with the target_clone attribute.
28794 (rs6000_clone_map): New array to identify which clone processors
28795 the current program is running on.
28796 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28797 target_clone attribute.
28798 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28799 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28800 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28801 (cpu_expand_builtin): Add support for target_clone attribute.
28802 (rs6000_valid_attribute_p): Allow "default" attribute.
28803 (get_decl_name): New debug function to simplify printing the
28804 current function name in debugging statements.
28805 (rs6000_clone_priority): New functions to support the target_clone
28806 attribute, and be able to generate code to switch between ISA 2.05
28807 through ISA 3.0 (power6 through power9).
28808 (rs6000_compare_version_priority): Likewise.
28809 (rs6000_get_function_versions_dispatcher): Likewise.
28810 (make_resolver_func): Likewise.
28811 (add_condition_to_bb): Likewise.
28812 (dispatch_function_versions): Likewise.
28813 (rs6000_generate_version_dispatcher_body): Likewise.
28814 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28815 (fusion_gpr_load_p): Fix a spacing issue.
28816 * doc/extend.texi (Common Function Attributes): Document that the
28817 PowerPC supports the target_clone attribute.
28819 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28821 * config/arm/arm.h: explain F symbol found in description of ARM
28822 register allocation in its legend.
28824 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28826 * config/mips/frame-header-opt.c: Include profile-count.h.
28827 * config/riscv/riscv.c: Include profile-count.h
28829 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28831 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28833 (sm_set_flag_if_changed): Add bbs field.
28834 (execute_sm_if_changed_flag_set): Pass BBS.
28835 (execute_sm): Update.
28837 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28839 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28842 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28844 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28845 (peephole2): New peephole2 to emit the above.
28846 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28848 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28850 * config/aarch64/aarch64.c (define_peephole2 above
28851 *sub_<shift>_<mode>): New peephole.
28853 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28855 * config/i386/i386.c (make_resolver_func): Update.
28856 * Makefile.in: Add profile-count.h and profile-count.o
28857 * auto-profile.c (afdo_indirect_call): Update to new API.
28858 (afdo_set_bb_count): Update.
28859 (afdo_propagate_edge): Update.
28860 (afdo_propagate_circuit): Update.
28861 (afdo_calculate_branch_prob): Update.
28862 (afdo_annotate_cfg): Update.
28863 * basic-block.h: Include profile-count.h
28864 (struct edge_def): Turn count to profile_count.
28865 (struct basic_block_def): Likewie.
28866 (REG_BR_PROB_BASE): Move to profile-count.h
28867 (RDIV): Move to profile-count.h
28868 * bb-reorder.c (max_entry_count): Turn to profile_count.
28869 (find_traces): Update.
28870 (rotate_loop):Update.
28871 (connect_traces):Update.
28872 (sanitize_hot_paths):Update.
28873 * bt-load.c (migrate_btr_defs): Update.
28874 * cfg.c (RDIV): Remove.
28875 (init_flow): Use alloc_block.
28876 (alloc_block): Uninitialize count.
28877 (unchecked_make_edge): Uninitialize count.
28878 (check_bb_profile): Update.
28879 (dump_edge_info): Update.
28880 (dump_bb_info): Update.
28881 (update_bb_profile_for_threading): Update.
28882 (scale_bbs_frequencies_int): Update.
28883 (scale_bbs_frequencies_gcov_type): Update.
28884 (scale_bbs_frequencies_profile_count): New.
28885 * cfg.h (update_bb_profile_for_threading): Update.
28886 (scale_bbs_frequencies_profile_count): Declare.
28887 * cfgbuild.c (compute_outgoing_frequencies): Update.
28888 (find_many_sub_basic_blocks): Update.
28889 * cfgcleanup.c (try_forward_edges): Update.
28890 (try_crossjump_to_edge): Update.
28891 * cfgexpand.c (expand_gimple_tailcall): Update.
28892 (construct_exit_block): Update.
28893 * cfghooks.c (verify_flow_info): Update.
28894 (dump_bb_for_graph): Update.
28895 (split_edge): Update.
28896 (make_forwarder_block): Update.
28897 (duplicate_block): Update.
28898 (account_profile_record): Update.
28899 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28900 (get_estimated_loop_iterations): Update.
28901 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28902 (single_likely_exit): Update.
28903 * cfgloopmanip.c (scale_loop_profile): Update.
28905 (set_zero_probability): Update.
28906 (lv_adjust_loop_entry_edge): Update.
28907 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28908 (purge_dead_edges): Update.
28909 (rtl_account_profile_record): Update.
28910 * cgraph.c (cgraph_node::create): Uninitialize count.
28911 (symbol_table::create_edge): Uninitialize count.
28912 (cgraph_update_edges_for_call_stmt_node): Update.
28913 (cgraph_edge::dump_edge_flags): Update.
28914 (cgraph_node::dump): Update.
28915 (cgraph_edge::maybe_hot_p): Update.
28916 * cgraph.h: Include profile-count.h
28917 (create_clone), create_edge, create_indirect_edge): Update.
28918 (cgraph_node): Turn count to profile_count.
28919 (cgraph_edge0: Likewise.
28920 (make_speculative, clone): Update.
28921 (create_edge): Update.
28922 (init_lowered_empty_function): Update.
28923 * cgraphclones.c (cgraph_edge::clone): Update.
28924 (duplicate_thunk_for_node): Update.
28925 (cgraph_node::create_clone): Update.
28926 * cgraphunit.c (cgraph_node::analyze): Update.
28927 (cgraph_node::expand_thunk): Update.
28928 * final.c (dump_basic_block_info): Update.
28929 * gimple-streamer-in.c (input_bb): Update.
28930 * gimple-streamer-out.c (output_bb): Update.
28931 * graphite.c (print_global_statistics): Update.
28932 (print_graphite_scop_statistics): Update.
28933 * hsa-brig.c: Include basic-block.h.
28934 * hsa-dump.c: Include basic-block.h.
28935 * hsa-gen.c (T sum_slice): Update.
28936 (convert_switch_statements):Update.
28937 * hsa-regalloc.c: Include basic-block.h.
28938 * ipa-chkp.c (chkp_produce_thunks): Update.
28939 * ipa-cp.c (struct caller_statistics): Update.
28940 (init_caller_stats): Update.
28941 (gather_caller_stats): Update.
28942 (ipcp_cloning_candidate_p): Update.
28943 (good_cloning_opportunity_p): Update.
28944 (get_info_about_necessary_edges): Update.
28945 (dump_profile_updates): Update.
28946 (update_profiling_info): Update.
28947 (update_specialized_profile): Update.
28948 (perhaps_add_new_callers): Update.
28949 (decide_about_value): Update.
28950 (ipa_cp_c_finalize): Update.
28951 * ipa-devirt.c (struct odr_type_warn_count): Update.
28952 (struct decl_warn_count): Update.
28953 (struct final_warning_record): Update.
28954 (possible_polymorphic_call_targets): Update.
28955 (ipa_devirt): Update.
28956 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28957 * ipa-icf.c (sem_function::merge): Update.
28958 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28959 * ipa-inline.c (compute_uninlined_call_time): Update.
28960 (compute_inlined_call_time): Update.
28961 (want_inline_small_function_p): Update.
28962 (want_inline_self_recursive_call_p): Update.
28963 (edge_badness): Update.
28964 (lookup_recursive_calls): Update.
28965 (recursive_inlining): Update.
28966 (inline_small_functions): Update.
28967 (dump_overall_stats): Update.
28968 (dump_inline_stats): Update.
28969 * ipa-profile.c (ipa_profile_generate_summary): Update.
28970 (ipa_propagate_frequency): Update.
28971 (ipa_profile): Update.
28972 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28973 * ipa-utils.c (ipa_merge_profiles): Update.
28974 * loop-doloop.c (doloop_modify): Update.
28975 * loop-unroll.c (report_unroll): Update.
28976 (unroll_loop_runtime_iterations): Update.
28977 * lto-cgraph.c (lto_output_edge): Update.
28978 (lto_output_node): Update.
28979 (input_node): Update.
28980 (input_edge): Update.
28981 (merge_profile_summaries): Update.
28982 * lto-streamer-in.c (input_cfg): Update.
28983 * lto-streamer-out.c (output_cfg): Update.
28984 * mcf.c (create_fixup_graph): Update.
28985 (adjust_cfg_counts): Update.
28986 (sum_edge_counts): Update.
28987 * modulo-sched.c (sms_schedule): Update.
28988 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28989 * predict.c (maybe_hot_count_p): Update.
28990 (probably_never_executed): Update.
28991 (dump_prediction): Update.
28992 (combine_predictions_for_bb): Update.
28993 (propagate_freq): Update.
28994 (handle_missing_profiles): Update.
28995 (counts_to_freqs): Update.
28996 (rebuild_frequencies): Update.
28997 (force_edge_cold): Update.
28998 * predict.h: Include profile-count.h
28999 (maybe_hot_count_p, counts_to_freqs): UPdate.
29000 * print-rtl-function.c: Do not include cfg.h
29001 * print-rtl.c: Include basic-block.h
29002 * profile-count.c: New file.
29003 * profile-count.h: New file.
29004 * profile.c (is_edge_inconsistent): Update.
29005 (correct_negative_edge_counts): Update.
29006 (is_inconsistent): Update.
29007 (set_bb_counts): Update.
29008 (read_profile_edge_counts): Update.
29009 (compute_frequency_overlap): Update.
29010 (compute_branch_probabilities): Update; Initialize and deinitialize
29012 (branch_prob): Update.
29013 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29014 (edge_gcov_count): New.
29015 (bb_gcov_count): New.
29016 * shrink-wrap.c (try_shrink_wrapping): Update.
29017 * tracer.c (better_p): Update.
29018 * trans-mem.c (expand_transaction): Update.
29019 (ipa_tm_insert_irr_call): Update.
29020 (ipa_tm_insert_gettmclone_call): Update.
29021 * tree-call-cdce.c: Update.
29022 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29023 (gimple_duplicate_sese_tail): Update.
29024 (gimple_account_profile_record): Update.
29025 (execute_fixup_cfg): Update.
29026 * tree-inline.c (copy_bb): Update.
29027 (copy_edges_for_bb): Update.
29028 (initialize_cfun): Update.
29029 (freqs_to_counts): Update.
29030 (copy_cfg_body): Update.
29031 (expand_call_inline): Update.
29032 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29033 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29034 (try_unroll_loop_completely): Update.
29035 (try_peel_loop): Update.
29036 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29037 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29038 * tree-ssa-loop-split.c (connect_loops): Update.
29039 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29040 * tree-ssa-reassoc.c (branch_fixup): Update.
29041 * tree-ssa-tail-merge.c (replace_block_by): Update.
29042 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29043 (compute_path_counts): Update.
29044 (update_profile): Update.
29045 (recompute_probabilities): Update.
29046 (update_joiner_offpath_counts): Update.
29047 (estimated_freqs_path): Update.
29048 (freqs_to_counts_path): Update.
29049 (clear_counts_path): Update.
29050 (ssa_fix_duplicate_block_edges): Update.
29051 (duplicate_thread_path): Update.
29052 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29053 (struct switch_conv_info): Update.
29054 * tree-tailcall.c (decrease_profile): Update.
29055 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29056 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29057 * value-prof.c (check_counter): Update.
29058 (gimple_divmod_fixed_value): Update.
29059 (gimple_mod_pow2): Update.
29060 (gimple_mod_subtract): Update.
29061 (gimple_ic_transform): Update.
29062 (gimple_stringop_fixed_value): Update.
29063 * value-prof.h (gimple_ic): Update.
29065 2017-06-02 Carl Love <cel@us.ibm.com>
29067 * config/rs6000/rs6000-c: Add support for built-in functions
29068 vector double vec_doublee (vector signed int);
29069 vector double vec_doublee (vector unsigned int);
29070 vector double vec_doublee (vector float);
29071 vector double vec_doubleh (vector signed int);
29072 vector double vec_doubleh (vector unsigned int);
29073 vector double vec_doubleh (vector float);
29074 vector double vec_doublel (vector signed int);
29075 vector double vec_doublel (vector unsigned int);
29076 vector double vec_doublel (vector float);
29077 vector double vec_doubleo (vector signed int);
29078 vector double vec_doubleo (vector unsigned int);
29079 vector double vec_doubleo (vector float);.
29080 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29081 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29083 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29084 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29085 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29087 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29088 vec_doublel, vec_doubleh.
29089 * doc/extend.texi: Update the built-in documentation file for the
29090 new built-in functions.
29092 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29095 * ipa-inline-analysis.c (free_growth_caches): Set
29096 edge_removal_hook_holder to NULL after removing it.
29098 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29100 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29101 comparision with zero.
29103 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29104 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29105 for early expansion of vec_min and vec_max builtins.
29106 (builtin_function_type): Add min/max unsigned variants to those
29107 identified as having unsigned arguments.
29109 2017-06-02 Olivier Hainque <hainque@adacore.com>
29111 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29113 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29115 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29116 Use VALL_F16 iterator rather than VALL.
29118 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29120 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29121 Emit CBNZ inside loop when doing a strong exchange and comparing
29122 against zero. Generate the CC flags after the loop.
29124 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29126 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29127 (dl_section_ref): New.
29128 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29129 On AIX, append an expression to subtract the size of the
29130 section length to dl_section_ref.
29132 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29134 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29135 for early expansion of vector absolute builtins.
29137 2017-06-02 Richard Biener <rguenther@suse.de>
29139 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29140 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29142 2017-06-02 Richard Biener <rguenther@suse.de>
29144 PR tree-optimization/80948
29145 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29146 stmt order as well.
29148 2017-06-02 Richard Biener <rguenther@suse.de>
29150 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29152 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29153 for inductions as relevant.
29155 2017-06-02 Richard Biener <rguenther@suse.de>
29157 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29158 (vectorizable_induction): ... this. Remove dead code.
29160 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29162 * builtins. (expand_builtin_alloca): Remove second parameter and
29163 infer its value from the first parameter instead.
29164 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29166 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29168 PR rtl-optimization/80903
29169 * loop-doloop.c (add_test): Unshare sequence.
29171 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29173 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29175 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29177 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29179 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29180 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29181 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29182 (xlogue_layout::get_stub_rtx): Make static.
29183 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29184 (xlogue_layout::compute_stub_managed_regs): Rename to...
29185 (xlogue_layout::count_stub_managed_regs): ...this.
29186 (xlogue_layout::is_stub_managed_reg): New function.
29187 (xlogue_layout::m_stub_names): Rename to...
29188 (xlogue_layout::s_stub_names): ...this, make static.
29189 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29190 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29191 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29192 xlogue_layout::s_stub_names): Instantiate statics.
29193 (stub_managed_regs): Remove.
29194 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29195 (disable_call_ms2sysv_xlogues): Rename to...
29196 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29197 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29199 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29200 change after reload_completed.
29201 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29203 (ix86_expand_prologue): Likewise.
29204 (ix86_expand_epilogue): Likewise.
29205 (ix86_expand_split_stack_prologue): Likewise.
29206 (ix86_compute_frame_layout): Remove frame parameter ...
29207 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29208 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29210 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29211 (ix86_frame): Move from here ...
29212 * config/i386/i386.h (ix86_frame): ... to here.
29213 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29214 complete ix86_frame data structure instead. Remove some_ld_name.
29216 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29218 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29219 symbols that hold a DECL_VALUE_EXPR.
29221 2017-06-01 Martin Jambor <mjambor@suse.cz>
29223 PR tree-optimization/80898
29224 * tree-sra.c (process_subtree_disqualification): Removed.
29225 (disqualify_candidate): Do not acll
29226 process_subtree_disqualification.
29227 (subtree_mark_written_and_enqueue): New function.
29228 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29229 RHS has been disqualified and re-queue LHS if necessary. Apart
29230 from that, ignore disqualified RHS.
29232 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29234 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29235 address fetch for z10 or later.
29237 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29239 * config/arc/arc.md (tst_movb): Add guard when splitting.
29241 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29243 * config/arc/arc.c (arc_can_eliminate): Test against
29244 arc_frame_pointer_needed.
29246 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29248 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29249 to prevent store reordering.
29250 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29251 (type): Add block type.
29252 (stack_tie): Define special instruction to be used in
29255 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29257 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29258 constraint. It is not valid for the pattern.
29259 (noncommutative_binary_comparison): Likewise.
29261 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29263 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29266 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29268 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29269 be used by the reg-alloc.
29271 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29273 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29274 reg-alloc when having mul64 or mul32x16 instructions.
29275 (mulsidi3): Likewise.
29276 (umulsidi3): Likewise.
29277 (mulsi32x16): New pattern.
29278 (mulsi64): Likewise.
29279 (mulsidi64): Likewise.
29280 (umulsidi64): Likewise.
29281 (MUL32x16_REG): Define.
29282 (mul64_600): Use MUL32x16_REG.
29283 (mac64_600): Likewise.
29284 (umul64_600): Likewise.
29285 (umac64_600): Likewise.
29287 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29289 * config/arc/arc.md (mulsi3_700): Make it commutative.
29291 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29293 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29295 (*sign_extendsidi2_insn): Likewise for movstosw.
29297 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29299 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29300 the type of the input discriminant value. Convert the
29301 discriminant value of signedness vary.
29303 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29305 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29306 Add to -Wall section.
29308 2017-06-01 Richard Biener <rguenther@suse.de>
29310 PR middle-end/66313
29311 * fold-const.c (fold_plusminus_mult_expr): If the factored
29312 factor may be zero use a wrapping type for the inner operation.
29313 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29314 and handle moved defs.
29315 (process_assignment): Properly guard the unary op case. Return a
29316 tri-state indicating that moving the stmt before the call may allow
29317 to continue. Pass through to_move.
29318 (find_tail_calls): Handle moving unrelated defs before
29321 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29324 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29325 splitter result in the canonical way.
29327 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29329 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29330 also for 32bit target. Update insn attributes.
29331 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29333 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29335 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29336 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29337 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29338 (_mm_maskz_min_ss): New intrinsics.
29340 2017-05-31 Martin Liska <mliska@suse.cz>
29342 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29343 Change comment style to one we normally use.
29344 (vectorizable_reduction): Likewise.
29345 (vectorizable_induction): Likewise.
29346 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29347 (vectorizable_call): Likewise.
29348 (vectorizable_simd_clone_call): Likewise.
29349 (vectorizable_conversion): Likewise.
29350 (vectorizable_assignment): Likewise.
29351 (vectorizable_shift): Likewise.
29352 (vectorizable_operation): Likewise.
29353 (vectorizable_store): Likewise.
29354 (vectorizable_load): Likewise.
29355 * tree-vectorizer.h: Likewise.
29357 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29359 * passes.c (emergency_dump_function): New.
29360 * tree-pass.h (emergency_dump_function): Declare.
29361 * plugin.c (plugins_internal_error_function): Remove.
29362 * plugin.h (plugins_internal_error_function): Remove declaration.
29363 * toplev.c (internal_error_function): New static function. Use it...
29364 (general_init): ...here.
29366 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29368 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29369 (arc_rtx_costs): Add costs for new patterns.
29370 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29371 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29373 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29375 * tree-ssa-strlen.c (get_next_strinfo): New function.
29376 (get_stridx_plus_constant): Use it.
29377 (zero_length_string): Likewise.
29378 (adjust_related_strinfos): Likewise.
29379 (adjust_last_stmt): Likewise.
29381 2017-05-31 Richard Biener <rguenther@suse.de>
29384 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29385 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29387 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29389 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29390 loop_vinfo argument and use of dependence distance vectors.
29391 Check instead whether the two references differ only in their
29392 initial value and assume that they have the same alignment if the
29393 difference is a multiple of the vector alignment.
29394 (vect_analyze_data_refs_alignment): Update call accordingly.
29396 2017-05-31 Martin Liska <mliska@suse.cz>
29399 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29401 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29403 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29404 (create_intersect_range_checks): Move from ...
29405 * tree-data-ref.c (create_intersect_range_checks_index)
29406 (create_intersect_range_checks): ... to here.
29407 (create_runtime_alias_checks): New function factored from ...
29408 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29409 here. Call above function.
29410 * tree-data-ref.h (create_runtime_alias_checks): New function.
29412 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29414 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29415 segment length for dr_b and compute it in wide_int.
29417 2017-05-31 Richard Biener <rguenther@suse.de>
29419 PR tree-optimization/80906
29420 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29421 and pass through iv_map.
29422 (copy_bb_and_scalar_dependences): Adjust.
29423 (translate_pending_phi_nodes): Likewise.
29424 (copy_loop_close_phi_args): Handle code-generating IVs instead
29427 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29429 * diagnostic-color.c (color_dict): Add "type-diff".
29430 (parse_gcc_colors): Update comment.
29431 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29432 -fdiagnostics-show-template-tree and -fno-elide-type.
29433 (GCC_COLORS): Add type-diff to example.
29435 (-fdiagnostics-show-template-tree): New.
29436 (-fno-elide-type): New.
29437 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29438 the pp_format_decoder callback. Call any m_format_postprocessor's
29440 (pretty_printer::pretty_printer): Initialize
29441 m_format_postprocessor.
29442 (pretty_printer::~pretty_printer): Delete any
29443 m_format_postprocessor.
29444 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29445 (class format_postprocessor): New class.
29446 (struct pretty_printer::format_decoder): Document the new parameters.
29447 (struct pretty_printer::m_format_postprocessor): New field.
29448 * tree-diagnostic.c (default_tree_printer): Update for new
29449 bool and const char ** params.
29450 * tree-diagnostic.h (default_tree_printer): Likewise.
29452 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29454 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29455 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29456 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29457 rs6000_gen_cell_microcode code.
29458 (rs6000_final_prescan_insn): Delete.
29459 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29460 "warn-cell-microcode" entries.
29461 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29462 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29463 throughout. Change cc_reg_not_micro_cr0_operand to
29464 cc_reg_not_cr0_operand throughout.
29465 (*extendhi<mode>2_noload): Delete.
29466 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29467 (mwarn-cell-microcode): Delete.
29468 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29469 -mgen-cell-microcode and -mwarn-cell-microcode.
29471 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29474 * config/i386/constraints.md (Yd): New constraint.
29476 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29477 and (?Yd, r) alternatives. Update insn attributes.
29478 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29479 and (?*Yd, r) alternatives. Update insn attributes.
29480 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29482 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29484 * gimplify.c (gimplify_modify_expr): Don't create a
29485 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29488 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29490 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29492 2017-05-30 Richard Biener <rguenther@suse.de>
29494 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29495 and reduc_def fields.
29496 (STMT_VINFO_REDUC_TYPE): New define.
29497 (STMT_VINFO_REDUC_DEF): Likewise.
29498 (vect_force_simple_reduction): Adjust prototype.
29499 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29500 (vect_is_simple_reduction): Remove check_reduction argument.
29501 (vect_force_simple_reduction): Adjust and set
29502 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29503 (vectorizable_reduction): Do not re-do reduction analysis
29504 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29505 * tree-parloops.c (gather_scalar_reductions): Adjust.
29507 2017-05-30 Richard Biener <rguenther@suse.de>
29509 PR middle-end/80901
29510 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29513 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29515 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29516 Introduce unknown_misalignment parameter and remove vf.
29517 (vect_peeling_hash_get_lowest_cost):
29518 Pass unknown_misalignment parameter.
29519 (vect_enhance_data_refs_alignment):
29520 Fix unsupportable data ref treatment.
29522 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29524 * tree-vect-data-refs.c (vect_get_data_access_cost):
29525 Workaround for SLP handling.
29526 (vect_enhance_data_refs_alignment):
29527 Compute costs for doing no peeling at all, compare to the best
29528 peeling costs so far and avoid peeling if cheaper.
29530 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29532 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29533 Return peeling info and set costs to zero for unlimited cost
29535 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29536 with unknown misalignment. Compute and costs for unknown
29537 misalignment, compare them to the costs for known misalignment
29538 and choose the cheapest for peeling.
29540 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29542 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29543 (vect_get_peeling_costs_all_drs): Create function.
29544 (vect_peeling_hash_get_lowest_cost):
29545 Use vect_get_peeling_costs_all_drs.
29546 (vect_peeling_supportable): Create function.
29547 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29549 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29551 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29552 DR_HAS_NEGATIVE_STEP.
29553 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29554 (vect_enhance_data_refs_alignment): Use.
29555 (vect_duplicate_ssa_name_ptr_info): Use.
29556 * tree-vectorizer.h (dr_misalignment): Use.
29557 (known_alignment_for_access_p): Use.
29559 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29562 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29564 (has_section_name): New function.
29566 2017-05-30 Martin Liska <mliska@suse.cz>
29569 * auto-profile.c (get_function_decl_from_block): Fix
29572 2017-05-30 Richard Biener <rguenther@suse.de>
29574 PR middle-end/80876
29575 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29577 2017-05-30 Martin Liska <mliska@suse.cz>
29579 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29580 * dumpfile.h (struct dump_file_info): Remove ctors.
29582 2017-05-30 Martin Liska <mliska@suse.cz>
29584 * predict.def: Fix GNU coding style.
29586 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29588 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29589 Mark 'to' argument with ATTRIBUTE_UNUSED.
29591 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29593 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29594 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29595 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29598 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29600 * doc/install.texi (Options specification): Restore entry of
29601 --enable-sjlj-exceptions.
29603 2017-05-27 Michael Eager <eager@eagercon.com>
29606 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29608 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29610 * config/microblaze/microblaze.h
29611 (FIXED_REGISTERS): Update in macro.
29612 (CALL_USED_REGISTERS): Update in macro.
29614 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29616 * doc/install.texi: Add links to macOS binary distributions.
29618 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29622 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29624 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29626 2017-05-26 Martin Liska <mliska@suse.cz>
29628 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29630 2017-05-26 Martin Liska <mliska@suse.cz>
29632 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29633 always leading ';; '.
29634 (dump_bb_info): Likewise.
29635 (brief_dump_cfg): Likewise.
29636 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29637 * dumpfile.c: Remove usage of TDF_VERBOSE.
29638 * dumpfile.h (enum dump_kind): Likewise.
29639 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29640 * print-tree.c (debug_verbose): Remove.
29641 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29642 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29643 * tree-diagnostic.c (default_tree_printer): Replace
29644 TDF_DIAGNOSTIC with TDF_SLIM.
29646 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29648 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29649 in parameter loop, rather than loop_vinfo.
29650 (create_intersect_range_checks): Ditto.
29651 (vect_create_cond_for_alias_checks): Update call to above functions.
29653 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29655 PR tree-optimization/80815
29656 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29657 for merging runtime alias checks. Handle negative DR_STEPs.
29659 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29661 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29663 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29664 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29665 out code pruning runtime alias checks.
29666 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29667 factored out from above.
29668 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29670 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29672 (prune_runtime_alias_test_list): New decalaration.
29674 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29676 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29677 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29678 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29679 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29680 (operator==, comp_dr_with_seg_len_pair): Ditto.
29681 (vect_prune_runtime_alias_test_list): Ditto.
29683 2017-05-26 Martin Liska <mliska@suse.cz>
29686 * params.def: Bound partial-inlining-entry-probability param.
29688 2017-05-26 Marek Polacek <polacek@redhat.com>
29691 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29694 2017-05-26 Richard Biener <rguenther@suse.de>
29696 PR tree-optimization/80842
29697 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29700 2017-05-26 Richard Biener <rguenther@suse.de>
29702 PR tree-optimization/80844
29703 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29705 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29707 * doc/md.texi (Machine Constraints): Update x86 family
29708 machine constraints section to match 'config/i386/constraints.md'.
29710 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29712 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29714 2017-05-25 Nathan Sidwell <nathan@acm.org>
29716 * doc/invoke.texi (--enable-languages): Update documentation.
29718 2017-05-25 Martin Liska <mliska@suse.cz>
29720 * dumpfile.c: Add TDF_FOLDING.
29721 * dumpfile.h (enum dump_kind): Likewise.
29722 * genmatch.c (dt_simplify::gen_1): Use it.
29724 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29726 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29728 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29730 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29731 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29733 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29735 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29736 * match.pd (X == C): Rewrite it here.
29737 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29738 with_certain_nonzero_bits2): New predicates.
29739 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29741 2017-05-24 Nathan Sidwell <nathan@acm.org>
29743 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29746 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29749 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29751 * config/powerpcspe: New port. Files are copied from the rs6000
29752 port, with "rs6000" in filenames replaced by "powerpcspe".
29754 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29756 PR rtl-optimization/80754
29757 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29759 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29761 * config/sparc/sparc.md (length): Return the correct value for -mflat
29762 sibcalls to match output_sibcall.
29764 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29768 * config/rs6000/rs6000.c (struct machine_function): Add new field
29770 (rs6000_get_separate_components): Init that field, use it.
29771 (rs6000_components_for_bb): Use the field.
29773 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29775 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29777 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29779 PR middle-end/80823
29780 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29782 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29785 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29786 address against address_operand predicate.
29787 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29789 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29791 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29792 parameters passed indirectly.
29794 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29796 * config/i386/i386.md (*movdi_internal): Remove SSE4
29797 alternative 18 (?r, *v). Update insn attributes.
29798 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29799 Update insn attributes.
29800 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29801 Update insn attributes.
29802 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29803 alternative 1 (r, v). Remove isa attribute.
29804 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29805 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29806 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29808 2017-05-23 Tom de Vries <tom@codesourcery.com>
29810 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29813 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29815 * cgraphunit.c (symbol_table::process_new_functions): Update.
29816 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29817 (inline_generate_summary): Rename to ...
29818 (ipa_fn_summary_generate): ... this one.
29819 (inline_read_summary): Rename to ...
29820 (ipa_fn_summary_read): ... this one.
29821 (inline_write_summary): Rename to ...
29822 (ipa_fn_summary_write): ... this one.
29823 (inline_free_summary): Rename to ...
29824 (ipa_free_fn_summary): ... this one.
29825 (pass_data_local_fn_summary, pass_local_fn_summary,
29826 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29827 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29828 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29829 make_pass_ipa_fn_summary): New.
29830 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29831 inline_write_summary, inline_free_summary): Remove.
29832 (ipa_free_fn_summary) : New.
29833 * ipa-inline.c (ipa_inline): Update.
29834 (pass_ipa_inline): Do not generate summaries.
29835 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29837 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29838 and add pass_ipa_fn_summary.
29839 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29841 (make_pass_inline_parameters): Remove.
29843 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29845 * omp-low.c (struct omp_context): Remove "default_kind" member.
29848 * omp-offload.c (execute_oacc_device_lower): Remove the
29849 parallelism dimensions function attributes for unparallelized
29850 OpenACC kernels constructs.
29852 2017-05-23 Martin Liska <mliska@suse.cz>
29854 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29856 (cgraph_edge::make_speculative): Likewise.
29857 (cgraph_edge::resolve_speculation): Likewise.
29858 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29859 (cgraph_node::dump): Likewise.
29860 * cgraph.h: Likewise.
29861 * cgraphunit.c (analyze_functions): Likewise.
29862 (symbol_table::compile): Likewise.
29863 * ipa-cp.c (print_all_lattices): Likewise.
29864 (determine_versionability): Likewise.
29865 (initialize_node_lattices): Likewise.
29866 (ipcp_verify_propagated_values): Likewise.
29867 (estimate_local_effects): Likewise.
29868 (update_profiling_info): Likewise.
29869 (create_specialized_node): Likewise.
29870 (perhaps_add_new_callers): Likewise.
29871 (decide_about_value): Likewise.
29872 (decide_whether_version_node): Likewise.
29873 (identify_dead_nodes): Likewise.
29874 (ipcp_store_bits_results): Likewise.
29875 * ipa-devirt.c (dump_targets): Likewise.
29876 (ipa_devirt): Likewise.
29877 * ipa-icf.c (sem_item::dump): Likewise.
29878 (sem_function::equals): Likewise.
29879 (sem_variable::equals): Likewise.
29880 (sem_item_optimizer::read_section): Likewise.
29881 (sem_item_optimizer::execute): Likewise.
29882 (congruence_class::dump): Likewise.
29883 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29884 (dump_inline_summary): Likewise.
29885 (estimate_node_size_and_time): Likewise.
29886 (inline_analyze_function): Likewise.
29887 * ipa-inline-transform.c (inline_call): Likewise.
29888 * ipa-inline.c (report_inline_failed_reason): Likewise.
29889 (want_early_inline_function_p): Likewise.
29890 (edge_badness): Likewise.
29891 (update_edge_key): Likewise.
29892 (inline_small_functions): Likewise.
29893 * ipa-profile.c (ipa_profile): Likewise.
29894 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29895 (ipa_make_edge_direct_to_target): Likewise.
29896 (remove_described_reference): Likewise.
29897 (ipa_impossible_devirt_target): Likewise.
29898 (propagate_controlled_uses): Likewise.
29899 (ipa_print_node_params): Likewise.
29900 (ipcp_transform_function): Likewise.
29901 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29902 (propagate_pure_const): Likewise.
29903 * ipa-reference.c (generate_summary): Likewise.
29904 (read_write_all_from_decl): Likewise.
29905 (propagate): Likewise.
29906 (ipa_reference_read_optimization_summary): Likewise.
29907 * ipa-utils.c (ipa_merge_profiles): Likewise.
29908 * ipa.c (walk_polymorphic_call_targets): Likewise.
29909 (symbol_table::remove_unreachable_nodes): Likewise.
29910 (ipa_single_use): Likewise.
29911 * passes.c (execute_todo): Likewise.
29912 * predict.c (drop_profile): Likewise.
29913 * symtab.c (symtab_node::get_dump_name): New function.
29914 (symtab_node::dump_name): Likewise.
29915 (symtab_node::dump_asm_name): Likewise.
29916 (symtab_node::dump_references): Likewise.
29917 (symtab_node::dump_referring): Likewise.
29918 (symtab_node::dump_base): Likewise.
29919 (symtab_node::debug_symtab): Likewise.
29920 * tree-sra.c (convert_callers_for_node): Likewise.
29921 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29922 * value-prof.c (init_node_map): Likewise.
29924 2017-05-23 Martin Liska <mliska@suse.cz>
29926 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29927 and symtab_node::debug_symtab to symbol_table::debug.
29928 * cgraphunit.c (analyze_functions): Use the renamed function.
29929 (symbol_table::compile): Likewise.
29930 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29931 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29932 * passes.c (execute_todo): Likewise.
29933 * symtab.c (symbol_table::dump): New function.
29934 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29936 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29938 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29939 that nonconst implies exec.
29941 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29943 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29944 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29945 (inline_edge_summary_vec): Turn into ...
29946 (ipa_call_summaries): ... this one.
29947 (redirect_to_unreachable, edge_set_predicate,
29948 evaluate_properties_for_edge, inline_summary_alloc,
29949 reset_ipa_call_summary, reset_inline_summary,
29950 inline_summary_t::duplicate): Update.
29951 (inline_edge_duplication_hook): Turn to ...
29952 (ipa_call_summary_t::duplicate): ... this one.
29953 (inline_edge_removal_hook): Turn to ...
29954 (ipa_call_summary_t::remove): ... this one.
29955 (dump_inline_edge_summary): Turn to ...
29956 (dump_ipa_call_summary): ... this one.
29957 (estimate_function_body_sizes): Update.
29958 (inline_update_callee_summaries): Update.
29959 (remap_edge_change_prob): Update.
29960 (remap_edge_summaries): Update.
29961 (inline_merge_summary): Update.
29962 (do_estimate_edge_time): Update.
29963 (inline_generate_summary): Update.
29964 (inline_read_section): Update.
29965 (inline_read_summary): Update.
29966 (inline_free_summary): Update.
29967 * ipa-inline.c (can_inline_edge_p): Update.
29968 (compute_inlined_call_time): Update.
29969 (want_inline_small_function_p): Update.
29970 (edge_badness): Update.
29971 (early_inliner): Update.
29972 * ipa-inline.h (inline_edge_summary): Turn to ...
29973 (ipa_call_summary): ... this one.
29974 (ipa_call_summary_t): New class.
29975 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29976 (ipa_call_summaries): New.
29977 (inline_edge_summary): Remove.
29978 (estimate_edge_growth): Update.
29979 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29980 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29981 * ipa-split.c (execute_split_functions): Update.
29982 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29984 2017-05-23 Tom de Vries <tom@codesourcery.com>
29986 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29987 attributes): Document rdrand effective target.
29989 2017-05-23 Tom de Vries <tom@codesourcery.com>
29991 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29992 attributes): Sort alphabetically.
29994 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29996 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29998 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30001 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30002 V2DF/V2DI splat into two separate patterns, one that handles
30003 registers, and the other that only handles memory. Drop support
30004 for splatting from a GPR on ISA 2.07 and then splitting the
30005 splat into direct move and splat.
30006 (vsx_splat_<mode>_reg): Likewise.
30007 (vsx_splat_<mode>_mem): Likewise.
30009 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30011 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30013 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30015 PR middle-end/80809
30016 * omp-low.c (finish_taskreg_remap): New function.
30017 (finish_taskreg_scan): If unit size of ctx->record_type
30018 is non-constant, unshare the size expression and replace
30019 decls in it with possible outer var refs.
30021 PR middle-end/80809
30022 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30023 GOVD_SHARED rather than GOVD_PRIVATE with it.
30024 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30025 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30027 PR middle-end/80853
30028 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30029 as last argument to build_outer_var_ref for pointer bases of array
30030 section reductions.
30032 2017-05-19 Martin Sebor <msebor@redhat.com>
30034 * print-tree.c (print_node): Print DECL_READ_P flag.
30036 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30038 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30039 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30040 * cgraph.c: Likewise.
30041 * cgraphunit.c: Likewise.
30042 * gengtype.c: Likewise.
30043 * ipa-cp.c: Likewise.
30044 * ipa-devirt.c: Likewise.
30045 * ipa-icf.c: Likewise.
30046 * ipa-predicate.c: Likewise.
30047 * ipa-profile.c: Likewise.
30048 * ipa-prop.c: Likewise.
30049 * ipa-split.c: Likewise.
30051 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30052 edge_predicate_pool, dump_inline_hints,
30053 inline_summary::account_size_time, redirect_to_unreachable,
30054 edge_set_predicate, set_hint_predicate,
30055 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30056 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30057 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30058 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30059 ipa_call_summary_t::remove, initialize_growth_caches,
30060 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30061 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30062 mark_modified, unmodified_parm_1, unmodified_parm,
30063 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30064 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30065 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30066 will_be_nonconstant_predicate, record_modified_bb_info,
30067 get_minimal_bb, record_modified, param_change_prob,
30068 phi_result_unknown_predicate, predicate_for_phi_result,
30069 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30070 estimate_function_body_sizes, compute_inline_parameters,
30071 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30072 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30073 inline_update_callee_summaries, remap_edge_change_prob,
30074 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30075 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30076 inline_analyze_function, inline_summary_t::insert,
30077 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30078 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30079 inline_free_summary): Move to ipa-fnsummary.h
30080 (predicate_t): Remove.
30081 * ipa-fnsummary.c: New file.
30082 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30083 (enum inline_hints_vals, inline_hints, agg_position_info,
30084 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30085 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30086 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30087 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30088 inline_read_summary, inline_write_summary, inline_free_summary,
30089 inline_analyze_function, initialize_inline_failed,
30090 inline_merge_summary, inline_update_overall_summary,
30091 compute_inline_parameters): Move to ipa-fnsummary.h
30092 * ipa-fnsummary.h: New file.
30093 * ipa-inline-transform.h: Include ipa-inline.h.
30094 * ipa-inline.c: LIkewise.
30096 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30098 * ipa-inline.c (edge_badness): Use inlined_time instead of
30099 inline_summaries->get.
30101 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30103 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30105 2017-05-22 Nathan Sidwell <nathan@acm.org>
30107 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30108 (fdump-lang): Document 'raw' option.
30109 * dumpfile.h (TDI_tu): Delete.
30110 * dumpfile.c (dump_files): Remove translation-unit.
30111 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30113 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30115 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30116 command option from $(AWK) call.
30117 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30118 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30119 [FORMAT]: Remove handling of variable.
30120 * config/avr/t-multilib: Regenerate.
30122 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30124 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30126 (dump_inline_summary): Do not print self_time.
30127 (estimate_function_body_sizes): Do not set self_time.
30128 (compute_inline_parameters): Likewise.
30129 (inline_read_section, inline_write_summary): Do not stream self_time.
30130 * ipa-inline.h (inline_summary): Drop self_time.
30132 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30134 * ipa-inline-analysis.c (account_size_time): Rename to ...
30135 (inline_summary::account_size_time): ... this one.
30136 (reset_ipa_call_summary): Turn to ...
30137 (ipa_call_summary::reset): ... this one.
30138 (reset_inline_summary): Turn to ...
30139 (inline_summary::reset): ... this one.
30140 (inline_summary_t::remove): Update.
30141 (inline_summary_t::duplicate): Update.
30142 (ipa_call_summary_t::remove): Update.
30143 (dump_inline_summary): Update.
30144 (estimate_function_body_sizes): Update.
30145 (compute_inline_parameters): Update.
30146 (estimate_node_size_and_time): Update.
30147 (inline_merge_summary): Update.
30148 (inline_update_overall_summary): Update.
30149 (inline_read_section): Update.
30150 (inline_write_summary): Update.
30151 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30152 add account_size_time and reset member functions.
30153 (ipa_call_summary): Add reset function.
30154 * ipa-predicate.h (predicate::operator &): Constify.
30156 2017-05-22 Richard Biener <rguenther@suse.de>
30158 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30160 2017-05-19 Jason Merrill <jason@redhat.com>
30162 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30164 2017-05-19 Marek Polacek <polacek@redhat.com>
30167 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30168 TYPE_OVERFLOW_WRAPS checks.
30170 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30172 * tree-core.h (enum omp_clause_default_kind): Add
30173 "OMP_CLAUSE_DEFAULT_PRESENT".
30174 * tree-pretty-print.c (dump_omp_clause): Handle it.
30175 * gimplify.c (enum gimplify_omp_var_data): Add
30176 "GOVD_MAP_FORCE_PRESENT".
30177 (gimplify_adjust_omp_clauses_1): Map it to
30178 "GOMP_MAP_FORCE_PRESENT".
30179 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30181 * gimplify.c (oacc_default_clause): Clarify.
30183 2017-05-19 Nathan Sidwell <nathan@acm.org>
30185 LANG_HOOK_REGISTER_DUMPS
30186 * toplev.c (general_init): Call register dump lang hook.
30187 * doc/invoke.texi: Document -fdump-lang option family.
30188 * dumpfile.c (dump_files): Remove class dump here.
30189 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30190 * dumpfile.h (tree_dump_index): Remove TDI_class.
30191 * langhooks-def.h (lhd_register_dumps): Declare.
30192 (LANG_HOOKS_REGISTER_DUMPS): Define.
30193 (LANG_HOOKS_INITIALIZER): Add it.
30194 * langhooks.c (lhd_register_dumps): Define.
30195 * langhooks.h (struct lang_hooks): Add register_dumps.
30197 2017-05-19 Nathan Sidwell <nathan@acm.org>
30199 * context.h (context::set_passes): New.
30200 * context.c (context::context): Do not create pass manager.
30201 * toplev.c (general_init): Create pass manager here.
30203 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30205 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30206 use this splitter if two add or or instructions would also work for
30207 the constant we want to generate.
30209 2017-05-19 Richard Biener <rguenther@suse.de>
30212 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30213 predicate evaluation.
30215 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30217 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30219 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30222 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30224 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30225 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30226 (inline_edge_summary_vec): Turn into ...
30227 (ipa_call_summaries): ... this one.
30228 (redirect_to_unreachable, edge_set_predicate,
30229 evaluate_properties_for_edge, inline_summary_alloc,
30230 reset_ipa_call_summary, reset_inline_summary,
30231 inline_summary_t::duplicate): Update.
30232 (inline_edge_duplication_hook): Turn to ...
30233 (ipa_call_summary_t::duplicate): ... this one.
30234 (inline_edge_removal_hook): Turn to ...
30235 (ipa_call_summary_t::remove): ... this one.
30236 (dump_inline_edge_summary): Turn to ...
30237 (dump_ipa_call_summary): ... this one.
30238 (estimate_function_body_sizes): Update.
30239 (inline_update_callee_summaries): Update.
30240 (remap_edge_change_prob): Update.
30241 (remap_edge_summaries): Update.
30242 (inline_merge_summary): Update.
30243 (do_estimate_edge_time): Update.
30244 (inline_generate_summary): Update.
30245 (inline_read_section): Update.
30246 (inline_read_summary): Update.
30247 (inline_free_summary): Update.
30248 * ipa-inline.c (can_inline_edge_p): Update.
30249 (compute_inlined_call_time): Update.
30250 (want_inline_small_function_p): Update.
30251 (edge_badness): Update.
30252 (early_inliner): Update.
30253 * ipa-inline.h (inline_edge_summary): Turn to ...
30254 (ipa_call_summary): ... this one.
30255 (ipa_call_summary_t): New class.
30256 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30257 (ipa_call_summaries): New.
30258 (inline_edge_summary): Remove.
30259 (estimate_edge_growth): Update.
30260 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30261 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30262 * ipa-split.c (execute_split_functions): Update.
30263 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30265 2017-05-19 Richard Biener <rguenther@suse.de>
30267 PR middle-end/80764
30268 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30270 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30272 * config/rs6000/rs6000.c (struct machine_function): Add field
30273 fpr_is_wrapped_separately.
30274 (rs6000_get_separate_components): Use 64 components. Handle the
30275 new FPR components.
30276 (rs6000_components_for_bb): Handle the FPR components.
30277 (rs6000_emit_prologue_components): Handle the FPR components.
30278 (rs6000_emit_epilogue_components): Handle the FPR components.
30279 (rs6000_set_handled_components): Handle the FPR components.
30280 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30281 that are already separately shrink-wrapped.
30282 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30283 that are already separately shrink-wrapped.
30285 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30288 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30291 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30292 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30293 cases where the register allocator uses a move and an offsettable
30294 memory operation to/from a FPR register on ISA 2.06/2.07.
30295 (define_peephole2 for Altivec d-form store): Likewise.
30297 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30300 * config/i386/mmx.md (*mov<mode>_internal): Enable
30301 alternatives 11, 12, 13 and 14 also for 32bit targets.
30302 Remove alternatives 15, 16, 17 and 18.
30303 * config/i386/sse.md (vec_concatv2di): Change
30304 alternative (!x, *y) to (x, ?!*Yn).
30306 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30308 * dumpfile.h (enum dump_kind): Remove stray comma.
30310 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30312 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30313 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30314 predicate::num_conditions
30315 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30316 (CHANGED): turn into predicate::changed.
30317 (agg_position_info): Move to ipa-predicate.h
30318 (add_condition, predicate::add_clause, predicate::operator &=,
30319 predicate::or_with, predicate::evaluate, predicate::probability,
30320 dump_condition, dump_clause, predicate::dump,
30321 predicate::remap_after_duplication, predicate::remap_after_inlining,
30322 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30323 (evaluate_conditions_for_known_args): Update.
30324 (set_cond_stmt_execution_predicate): Update.
30325 * ipa-inline.h: Include ipa-predicate.h
30326 (condition, inline_param_summary, conditions, agg_position_info,
30327 predicate): Move to ipa-predicate.h
30328 * ipa-predicate.c: New file.
30329 * ipa-predicate.h: New file.
30331 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30333 * final.c (leaf_function_p): Check we are not in a sequence.
30335 2017-05-18 Martin Liska <mliska@suse.cz>
30337 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30338 * dumpfile.c (dump_register): Use new enum dump_kind.
30339 (get_dump_file_name): Likewise.
30340 (dump_enable_all): Likewise.
30341 (dump_switch_p_1): Likewise.
30342 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30343 * dumpfile.h (enum dump_kind): New enum type.
30344 (struct dump_file_info): Create constructor and
30345 format fields and comments.
30346 * passes.c (pass_manager::register_one_dump_file):
30348 * statistics.c (statistics_early_init): Likewise.
30349 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30350 TDF_TREE with TDF_SLIM.
30351 (gather_memory_references_ref): Likewise.
30353 2017-05-18 Martin Liska <mliska@suse.cz>
30355 * vec.h (struct vnull): Use it.
30357 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30359 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30360 (true_predicate, false_predicate, true_predicate_p,
30361 false_predicate_p): Remove.
30362 (single_cond_predicate, not_inlined_predicate): Turn to member function
30364 (add_condition): Update.
30365 (add_clause): Turn to...
30366 (predicate::add_clause): ... this one; update; allow passing NULL
30368 (and_predicates): Turn to ...
30369 (predicate::operator &=): ... this one.
30370 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30371 (or_predicates): Turn to ...
30372 (predicate::or_with): ... this one.
30373 (evaluate_predicate): Turn to ...
30374 (predicate::evaluate): ... this one.
30375 (predicate_probability): Turn to ...
30376 (predicate::probability): ... this one.
30377 (dump_condition): Update.
30378 (dump_predicate): Turn to ...
30379 (predicate::dump): ... this one.
30380 (account_size_time): Update.
30381 (edge_set_predicate): Update.
30382 (set_hint_predicate): UPdate.
30383 (evaluate_conditions_for_known_args): Update.
30384 (evaluate_properties_for_edge): Update.
30385 (remap_predicate_after_duplication): Turn to...
30386 (predicate::remap_after_duplication): ... this one.
30387 (remap_hint_predicate_after_duplication): Update.
30388 (inline_summary_t::duplicate): UPdate.
30389 (dump_inline_edge_summary): Update.
30390 (dump_inline_summary): Update.
30391 (set_cond_stmt_execution_predicate): Update.
30392 (set_switch_stmt_execution_predicate): Update.
30393 (compute_bb_predicates): Update.
30394 (will_be_nonconstant_expr_predicate): Update.
30395 (will_be_nonconstant_predicate): Update.
30396 (phi_result_unknown_predicate): Update.
30397 (predicate_for_phi_result): Update.
30398 (array_index_predicate): Update.
30399 (estimate_function_body_sizes): Update.
30400 (estimate_node_size_and_time): Update.
30401 (estimate_ipcp_clone_size_and_time): Update.
30402 (remap_predicate): Rename to ...
30403 (predicate::remap_after_inlining): ... this one.
30404 (remap_hint_predicate): Update.
30405 (inline_merge_summary): Update.
30406 (inline_update_overall_summary): Update.
30407 (estimate_size_after_inlining): Update.
30408 (read_predicate): Rename to ...
30409 (predicate::stream_in): ... this one.
30410 (read_inline_edge_summary): Update.
30411 (write_predicate): Rename to ...
30412 (predicate::stream_out): ... this one.
30413 (write_inline_edge_summary): Update.
30414 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30415 (clause_t): Turn to uint32_t
30416 (predicate): Turn to class; implement constructor and operators
30418 (size_time_entry): Update.
30419 (inline_summary): Update.
30420 (inline_edge_summary): Update.
30422 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30424 * fold-const.c (fold_binary_loc): Move transformation...
30425 * match.pd (C - X CMP X): ... here.
30427 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30429 * config/sparc/sparc.c (sparc_option_override): Set function
30430 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30431 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30433 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30435 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30437 2017-05-18 Marek Polacek <polacek@redhat.com>
30440 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30441 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30442 gimple_assign_load_p.
30444 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30446 PR middle-end/80692
30447 * real.c (do_compare): Give decimal_do_compare preference over
30448 comparing just the signs.
30450 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30452 * doc/md.texi (Canonicalization of Instructions): Describe the
30453 canonical form of instructions that inherently set a condition
30456 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30458 PR middle-end/80775
30459 * tree-cfg.c: Move deletion of unreachable case statements to after
30460 the merging of consecutive case labels.
30462 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30464 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30465 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30466 restoring of callee-saved registers.
30468 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30470 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30471 * config/visium/visium.c (single_set_and_flags): Likewise.
30472 * config/visium/visium.md (Substitutions): Likewise.
30474 2017-05-17 Martin Liska <mliska@suse.cz>
30476 * cfg.c: Introduce dump_flags_t type and
30477 use it instead of int type.
30479 * cfghooks.c: Likewise.
30480 * cfghooks.h (struct cfg_hooks): Likewise.
30481 * cfgrtl.c: Likewise.
30482 * cfgrtl.h: Likewise.
30483 * cgraph.c (cgraph_node::get_body): Likewise.
30484 * coretypes.h: Likewise.
30485 * domwalk.c: Likewise.
30486 * domwalk.h: Likewise.
30487 * dumpfile.c (struct dump_option_value_info): Likewise.
30488 (dump_enable_all): Likewise.
30489 (dump_switch_p_1): Likewise.
30490 (opt_info_switch_p): Likewise.
30491 * dumpfile.h (enum tree_dump_index): Likewise.
30492 (struct dump_file_info): Likewise.
30493 * genemit.c: Likewise.
30494 * generic-match-head.c: Likewise.
30495 * gengtype.c (open_base_files): Likewise.
30496 * gimple-pretty-print.c: Likewise.
30497 * gimple-pretty-print.h: Likewise.
30498 * graph.c (print_graph_cfg): Likewise.
30499 * graphite-scop-detection.c (dot_all_sese): Likewise.
30500 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30501 * loop-unroll.c (report_unroll): Likewise.
30502 * passes.c (pass_manager::register_one_dump_file): Likewise.
30503 * print-tree.c: Likewise.
30504 * statistics.c: Likewise.
30505 * tree-cfg.c: Likewise.
30506 * tree-cfg.h: Likewise.
30507 * tree-dfa.c: Likewise.
30508 * tree-dfa.h: Likewise.
30509 * tree-dump.c (dump_function): Likewise.
30510 * tree-dump.h (struct dump_info): Likewise.
30511 * tree-pretty-print.c: Likewise.
30512 * tree-pretty-print.h: Likewise.
30513 * tree-ssa-live.c: Likewise.
30514 * tree-ssa-live.h: Likewise.
30515 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30516 * tree-vect-loop.c: Likewise.
30517 * tree-vect-slp.c: Likewise.
30519 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30520 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30522 PR tree-optimization/80457
30523 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30524 of all arguments to a statement as scalar_to_vec operations.
30525 (vectorizable_call): Adjust call to vect_model_simple_cost for
30527 (vectorizable_conversion): Likewise.
30528 (vectorizable_assignment): Likewise.
30529 (vectorizable_shift): Likewise.
30530 (vectorizable_operation): Likewise.
30531 (vectorizable_comparison): Likewise.
30532 (vect_is_simple_cond): Record the def types for operands.
30533 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30534 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30535 for statement argument count.
30537 2017-05-16 Carl Love <cel@us.ibm.com>
30539 * config/rs6000/rs6000-c: Add support for built-in functions
30540 vector unsigned long long vec_bperm (vector unsigned long long,
30541 vector unsigned char)
30542 vector signed long long vec_mule (vector signed int,
30544 vector unsigned long long vec_mule (vector unsigned int,
30545 vector unsigned int)
30546 vector signed long long vec_mulo (vector signed int,
30548 vector unsigned long long vec_mulo (vector unsigned int,
30549 vector unsigned int)
30550 vector signed char vec_sldw (vector signed char,
30551 vector signed char,
30553 vector unsigned char vec_sldw (vector unsigned char,
30554 vector unsigned char,
30556 vector signed short vec_sldw (vector signed short,
30557 vector signed short,
30559 vector unsigned short vec_sldw (vector unsigned short,
30560 vector unsigned short,
30562 vector signed int vec_sldw (vector signed int,
30565 vector unsigned int vec_sldw (vector unsigned int,
30566 vector unsigned int,
30568 vector signed long long vec_sldw (vector signed long long,
30569 vector signed long long,
30571 vector unsigned long long vec_sldw (vector unsigned long long,
30572 vector unsigned long long,
30574 * config/rs6000/rs6000-c: Add support for built-in functions
30575 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30576 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30577 * doc/extend.texi: Update the built-in documentation for the
30578 new built-in functions.
30580 2017-05-16 Marek Polacek <polacek@redhat.com>
30584 * tree.c (save_expr): Don't fold the expression.
30586 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30588 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30589 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30590 and (?*y,m). Update insn attributes.
30592 2017-05-16 Martin Liska <mliska@suse.cz>
30594 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30595 flags argument of print_gimple_stmt, print_gimple_expr,
30596 print_generic_stmt and print_generic_expr.
30597 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30598 * coretypes.h: Likewise.
30599 * except.c (dump_eh_tree): Likewise.
30600 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30601 * gimple-pretty-print.h: Likewise.
30602 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30603 (backprop::push_to_worklist): Likewise.
30604 (backprop::pop_from_worklist): Likewise.
30605 (backprop::process_use): Likewise.
30606 (backprop::intersect_uses): Likewise.
30607 (note_replacement): Likewise.
30608 * gimple-ssa-store-merging.c
30609 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30610 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30611 (pass_store_merging::execute): Likewise.
30612 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30613 (ssa_base_cand_dump_callback): Likewise.
30614 (dump_incr_vec): Likewise.
30615 (replace_refs): Likewise.
30616 (replace_mult_candidate): Likewise.
30617 (create_add_on_incoming_edge): Likewise.
30618 (create_phi_basis): Likewise.
30619 (insert_initializers): Likewise.
30620 (all_phi_incrs_profitable): Likewise.
30621 (introduce_cast_before_cand): Likewise.
30622 (replace_one_candidate): Likewise.
30623 * gimplify.c (gimplify_expr): Likewise.
30624 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30625 (set_rename): Likewise.
30626 (rename_uses): Likewise.
30627 (copy_loop_phi_nodes): Likewise.
30628 (add_close_phis_to_merge_points): Likewise.
30629 (copy_loop_close_phi_args): Likewise.
30630 (copy_cond_phi_args): Likewise.
30631 (graphite_copy_stmts_from_block): Likewise.
30632 (translate_pending_phi_nodes): Likewise.
30633 * graphite-poly.c (print_pdr): Likewise.
30634 (dump_gbb_cases): Likewise.
30635 (dump_gbb_conditions): Likewise.
30636 (print_scop_params): Likewise.
30637 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30638 (build_cross_bb_scalars_use): Likewise.
30639 (gather_bbs::before_dom_children): Likewise.
30640 * hsa-dump.c (dump_hsa_immed): Likewise.
30641 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30642 (get_replacement_map): Likewise.
30643 * ipa-inline-analysis.c (dump_condition): Likewise.
30644 (estimate_function_body_sizes): Likewise.
30645 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30646 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30647 * ipa-prop.c (ipa_dump_param): Likewise.
30648 (ipa_print_node_jump_functions_for_edge): Likewise.
30649 (ipa_modify_call_arguments): Likewise.
30650 (ipa_modify_expr): Likewise.
30651 (ipa_dump_param_adjustments): Likewise.
30652 (ipa_dump_agg_replacement_values): Likewise.
30653 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30654 * ipa-pure-const.c (check_stmt): Likewise.
30655 (pass_nothrow::execute): Likewise.
30656 * ipa-split.c (execute_split_functions): Likewise.
30657 * omp-offload.c (dump_oacc_loop_part): Likewise.
30658 (dump_oacc_loop): Likewise.
30659 * trans-mem.c (tm_log_emit): Likewise.
30660 (tm_memopt_accumulate_memops): Likewise.
30661 (dump_tm_memopt_set): Likewise.
30662 (dump_tm_memopt_transform): Likewise.
30663 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30664 (print_loop): Likewise.
30665 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30666 (chkp_gather_checks_info): Likewise.
30667 (chkp_get_check_result): Likewise.
30668 (chkp_remove_check_if_pass): Likewise.
30669 (chkp_use_outer_bounds_if_possible): Likewise.
30670 (chkp_reduce_bounds_lifetime): Likewise.
30671 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30672 (chkp_mark_completed_bounds): Likewise.
30673 (chkp_register_incomplete_bounds): Likewise.
30674 (chkp_mark_invalid_bounds): Likewise.
30675 (chkp_maybe_copy_and_register_bounds): Likewise.
30676 (chkp_build_returned_bound): Likewise.
30677 (chkp_get_bound_for_parm): Likewise.
30678 (chkp_build_bndldx): Likewise.
30679 (chkp_get_bounds_by_definition): Likewise.
30680 (chkp_generate_extern_var_bounds): Likewise.
30681 (chkp_get_bounds_for_decl_addr): Likewise.
30682 * tree-chrec.c (chrec_apply): Likewise.
30683 * tree-data-ref.c (dump_data_reference): Likewise.
30684 (dump_subscript): Likewise.
30685 (dump_data_dependence_relation): Likewise.
30686 (analyze_overlapping_iterations): Likewise.
30687 * tree-inline.c (expand_call_inline): Likewise.
30688 (tree_function_versioning): Likewise.
30689 * tree-into-ssa.c (dump_defs_stack): Likewise.
30690 (dump_currdefs): Likewise.
30691 (dump_names_replaced_by): Likewise.
30692 (dump_update_ssa): Likewise.
30693 (update_ssa): Likewise.
30694 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30695 * tree-parloops.c (build_new_reduction): Likewise.
30696 (try_create_reduction_list): Likewise.
30697 (ref_conflicts_with_region): Likewise.
30698 (oacc_entry_exit_ok_1): Likewise.
30699 (oacc_entry_exit_single_gang): Likewise.
30700 * tree-pretty-print.h: Likewise.
30701 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30702 (get_scalar_evolution): Likewise.
30703 (add_to_evolution): Likewise.
30704 (get_loop_exit_condition): Likewise.
30705 (analyze_evolution_in_loop): Likewise.
30706 (analyze_initial_condition): Likewise.
30707 (analyze_scalar_evolution): Likewise.
30708 (instantiate_scev): Likewise.
30709 (number_of_latch_executions): Likewise.
30710 (gather_chrec_stats): Likewise.
30711 (final_value_replacement_loop): Likewise.
30712 (scev_const_prop): Likewise.
30713 * tree-sra.c (dump_access): Likewise.
30714 (disqualify_candidate): Likewise.
30715 (create_access): Likewise.
30716 (reject): Likewise.
30717 (maybe_add_sra_candidate): Likewise.
30718 (create_access_replacement): Likewise.
30719 (analyze_access_subtree): Likewise.
30720 (analyze_all_variable_accesses): Likewise.
30721 (sra_modify_assign): Likewise.
30722 (initialize_constant_pool_replacements): Likewise.
30723 (find_param_candidates): Likewise.
30724 (decide_one_param_reduction): Likewise.
30725 (replace_removed_params_ssa_names): Likewise.
30726 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30727 * tree-ssa-copy.c (dump_copy_of): Likewise.
30728 (copy_prop_visit_cond_stmt): Likewise.
30729 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30730 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30731 (record_equivalences_from_stmt): Likewise.
30732 * tree-ssa-dse.c (compute_trims): Likewise.
30733 (delete_dead_call): Likewise.
30734 (delete_dead_assignment): Likewise.
30735 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30736 (forward_propagate_into_cond): Likewise.
30737 (pass_forwprop::execute): Likewise.
30738 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30739 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30741 (move_computations_worker): Likewise.
30742 (execute_sm): Likewise.
30743 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30744 (remove_exits_and_undefined_stmts): Likewise.
30745 (remove_redundant_iv_tests): Likewise.
30746 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30747 (adjust_iv_update_pos): Likewise.
30748 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30749 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30750 (value_replacement): Likewise.
30751 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30752 * tree-ssa-pre.c (print_pre_expr): Likewise.
30753 (get_representative_for): Likewise.
30754 (create_expression_by_pieces): Likewise.
30755 (insert_into_preds_of_block): Likewise.
30756 (eliminate_insert): Likewise.
30757 (eliminate_dom_walker::before_dom_children): Likewise.
30758 (eliminate): Likewise.
30759 (remove_dead_inserted_code): Likewise.
30760 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30761 * tree-ssa-reassoc.c (get_rank): Likewise.
30762 (eliminate_duplicate_pair): Likewise.
30763 (eliminate_plus_minus_pair): Likewise.
30764 (eliminate_not_pairs): Likewise.
30765 (undistribute_ops_list): Likewise.
30766 (eliminate_redundant_comparison): Likewise.
30767 (update_range_test): Likewise.
30768 (optimize_range_tests_var_bound): Likewise.
30769 (optimize_vec_cond_expr): Likewise.
30770 (rewrite_expr_tree): Likewise.
30771 (rewrite_expr_tree_parallel): Likewise.
30772 (linearize_expr): Likewise.
30773 (break_up_subtract): Likewise.
30774 (linearize_expr_tree): Likewise.
30775 (attempt_builtin_powi): Likewise.
30776 (attempt_builtin_copysign): Likewise.
30777 (transform_stmt_to_copy): Likewise.
30778 (transform_stmt_to_multiply): Likewise.
30779 (dump_ops_vector): Likewise.
30780 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30781 (print_scc): Likewise.
30782 (set_ssa_val_to): Likewise.
30783 (visit_reference_op_store): Likewise.
30784 (visit_use): Likewise.
30785 (sccvn_dom_walker::before_dom_children): Likewise.
30786 (run_scc_vn): Likewise.
30787 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30789 (expr_hash_elt::print): Likewise.
30790 (const_and_copies::pop_to_marker): Likewise.
30791 (const_and_copies::record_const_or_copy_raw): Likewise.
30792 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30793 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30794 (dump_predicates): Likewise.
30795 (find_uninit_use): Likewise.
30796 (warn_uninitialized_phi): Likewise.
30797 (pass_late_warn_uninitialized::execute): Likewise.
30798 * tree-ssa.c (verify_vssa): Likewise.
30799 (verify_ssa): Likewise.
30800 (maybe_optimize_var): Likewise.
30801 * tree-vrp.c (dump_value_range): Likewise.
30802 (dump_all_value_ranges): Likewise.
30803 (dump_asserts_for): Likewise.
30804 (register_edge_assert_for_2): Likewise.
30805 (vrp_visit_cond_stmt): Likewise.
30806 (vrp_visit_switch_stmt): Likewise.
30807 (vrp_visit_stmt): Likewise.
30808 (vrp_visit_phi_node): Likewise.
30809 (simplify_cond_using_ranges_1): Likewise.
30810 (fold_predicate_in): Likewise.
30811 (evrp_dom_walker::before_dom_children): Likewise.
30812 (evrp_dom_walker::push_value_range): Likewise.
30813 (evrp_dom_walker::pop_value_range): Likewise.
30814 (execute_early_vrp): Likewise.
30816 2017-05-16 Richard Biener <rguenther@suse.de>
30818 * dwarf2out.c (loc_list_from_tree_1): Do not create
30819 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30821 2017-05-16 Richard Biener <rguenther@suse.de>
30823 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30825 (note_variable_value_in_expr): If we resolved the decl ref
30826 do not push to the stack.
30828 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30830 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30831 operations in fast-math mode.
30832 (vaddq_f16): Likewise.
30833 (vmul_f16): Likewise.
30834 (vmulq_f16): Likewise.
30835 (vsub_f16): Likewise.
30836 (vsubq_f16): Likewise.
30837 * config/arm/neon.md (add<mode>3): New.
30839 (fma:<VH:mode>3): New. Also remove outdated comment.
30842 2017-05-16 Martin Liska <mliska@suse.cz>
30846 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30847 (odr_types_equivalent_p): Likewise.
30849 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30851 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30853 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30856 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30857 non-interunit SSE move alternatives with '?'.
30858 (zero-extendsidi peephole2): New peephole to skip intermediate
30859 general register in SSE zero-extend sequence.
30861 2017-05-15 Jeff Law <law@redhat.com>
30863 * reorg.c (relax_delay_slots): Create a new variable to hold
30864 the temporary target rather than clobbering TARGET_LABEL.
30866 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30867 missing argument to extract_bit_field call.
30868 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30870 2017-05-15 Martin Liska <mliska@suse.cz>
30873 * gcc.c (process_command): Do not allow empty argument of -o option.
30875 2017-05-15 Renlin Li <renlin.li@arm.com>
30877 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30878 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30879 * config/aarch64/constraints.md (Usf): Add long call check.
30880 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30881 (call_value): Likewise.
30882 (sibcall): Likewise.
30883 (sibcall_value): Likewise.
30885 (call_value_insn): New.
30886 (sibcall_insn): Update rtx pattern.
30887 (sibcall_value_insn): Likewise.
30888 (call_internal): Remove.
30889 (call_value_internal): Likewise.
30890 (sibcall_internal): Likewise.
30891 (sibcall_value_internal): Likewise.
30892 (call_reg): Likewise.
30893 (call_symbol): Likewise.
30894 (call_value_reg): Likewise.
30895 (call_value_symbol): Likewise.
30897 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30900 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30902 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30904 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30905 compatible with CCGOCmode and with CCZmode.
30907 2017-05-14 Martin Sebor <msebor@redhat.com>
30909 PR middle-end/77671
30910 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30911 (gimple_fold_builtin_snprintf): Same.
30912 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30913 (gimple_fold_builtin_snprintf): Same.
30914 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30915 of character types.
30916 (is_call_safe): New function.
30917 (try_substitute_return_value): Call it.
30918 (try_simplify_call): New function.
30919 (pass_sprintf_length::handle_gimple_call): Call it.
30921 2017-05-14 Martin Sebor <msebor@redhat.com>
30923 PR middle-end/80669
30924 * builtins.c (expand_builtin_stpncpy): Simplify.
30926 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30928 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30929 * config/i386/i386.h
30930 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30931 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30932 (struct machine_function): Add new members call_ms2sysv,
30933 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30934 (struct machine_frame_state): New fields sp_realigned and
30935 sp_realigned_offset.
30936 * config/i386/i386.c
30937 (enum xlogue_stub): New enum.
30938 (enum xlogue_stub_sets): New enum.
30939 (class xlogue_layout): New class.
30940 (struct ix86_frame): New fields stack_realign_allocate_offset,
30941 stack_realign_offset and outlined_save_offset. Modify comments to
30942 detail stack layout when using out-of-line stubs.
30943 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30944 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30945 -mcall-ms2sysv-xlogues.
30946 (stub_managed_regs): New static variable.
30947 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30948 registers managed by out-of-line stub.
30949 (disable_call_ms2sysv_xlogues): New function.
30950 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30951 m->call_ms2sysv when appropriate and compute frame layout for
30953 (sp_valid_at, fp_valid_at): New inline functions.
30954 (choose_basereg): New function.
30955 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30957 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30958 Use align parameter of choose_baseaddr to generated aligned SSE movs
30960 (pro_epilogue_adjust_stack): Modify to track
30961 machine_frame_state::sp_realigned.
30962 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30963 (ix86_nsaved_sseregs): Likewise.
30964 (ix86_emit_save_regs): Likewise.
30965 (ix86_emit_save_regs_using_mov): Likewise.
30966 (ix86_emit_save_sse_regs_using_mov): Likewise.
30967 (get_scratch_register_on_entry): Likewise.
30968 (gen_frame_set): New function.
30969 (gen_frame_load): Likewise.
30970 (gen_frame_store): Likewise.
30971 (emit_outlined_ms2sysv_save): Likewise.
30972 (emit_outlined_ms2sysv_restore): Likewise.
30973 (ix86_expand_prologue): Modify stack re-alignment code and call
30974 emit_outlined_ms2sysv_save when appropriate.
30975 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30976 parameter rtx_insn *insn, which allows the function to be used to only
30977 generate the notes.
30978 (ix86_expand_epilogue): Modify validity checks of frame and stack
30979 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30980 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30981 * config/i386/predicates.md
30982 (save_multiple): New predicate.
30983 (restore_multiple): Likewise.
30984 * config/i386/sse.md
30985 (save_multiple<mode>): New pattern.
30986 (save_multiple_realign<mode>): Likewise.
30987 (restore_multiple<mode>): Likewise.
30988 (restore_multiple_and_return<mode>): Likewise.
30989 (restore_multiple_leave_return<mode>): Likewise.
30990 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30992 2017-05-14 Julia Koval <julia.koval@intel.com>
30994 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30995 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30996 (__builtin_ia32_xsetbv): New builtins.
30997 * config/i386/i386.c (ix86_expand_special_args_builtin):
30999 (ix86_expand_builtin): Special expand for new intrinsics.
31000 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31001 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31002 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31004 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31006 * cfganal.c (inverted_post_order_compute): Change argument type
31008 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31009 * df-core.c (rest_of_handle_df_initialize): Adjust.
31010 (rest_of_handle_df_finish): Likewise.
31011 (df_analyze_1): Likewise.
31012 (df_analyze): Likewise.
31013 (loop_inverted_post_order_compute): Change argument to be a vec *.
31014 (df_analyze_loop): Adjust.
31015 (df_get_n_blocks): Likewise.
31016 (df_get_postorder): Likewise.
31017 * df.h (struct df_d): Change field to be a vec.
31018 * lcm.c (compute_laterin): Adjust.
31019 (compute_available): Likewise.
31020 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31021 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31022 * tree-ssa-pre.c (compute_antic): Likewise.
31024 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31026 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31027 (depth_first_search::depth_first_search): Change structure init
31028 function to this constructor.
31029 (depth_first_search::add_bb): Rename function to this member.
31030 (depth_first_search::execute): Likewise.
31031 (flow_dfs_compute_reverse_finish): Adjust.
31033 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31035 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31036 (longest_simple_path): Likewise.
31037 * shrink-wrap.c (spread_components): Likewise.
31038 (disqualify_problematic_components): Likewise.
31039 (emit_common_heads_for_components): Likewise.
31040 (emit_common_tails_for_components): Likewise.
31041 (insert_prologue_epilogue_for_components): Likewise.
31043 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31045 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31048 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31050 * df-core.c (df_set_blocks): Start using auto_bitmap.
31051 (df_compact_blocks): Likewise.
31052 * df-problems.c (df_rd_confluence_n): Likewise.
31053 * df-scan.c (df_insn_rescan_all): Likewise.
31054 (df_process_deferred_rescans): Likewise.
31055 (df_update_entry_block_defs): Likewise.
31056 (df_update_exit_block_uses): Likewise.
31057 (df_entry_block_bitmap_verify): Likewise.
31058 (df_exit_block_bitmap_verify): Likewise.
31059 (df_scan_verify): Likewise.
31060 * lra-constraints.c (lra_constraints): Likewise.
31061 (undo_optional_reloads): Likewise.
31062 (lra_undo_inheritance): Likewise.
31063 * lra-remat.c (calculate_gen_cands): Likewise.
31064 (do_remat): Likewise.
31065 * lra-spills.c (assign_spill_hard_regs): Likewise.
31066 (spill_pseudos): Likewise.
31067 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31068 (bitmap_set_subtract_values): Likewise.
31070 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31072 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31073 management with auto_bitmap.
31074 (fix_inter_tick): Likewise.
31075 (fix_recovery_deps): Likewise.
31076 * ira.c (add_store_equivs): Likewise.
31077 (find_moveable_pseudos): Likewise.
31078 (split_live_ranges_for_shrink_wrap): Likewise.
31079 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31080 (rtx_reuse_manager::seen_def_p): Likewise.
31081 (rtx_reuse_manager::set_seen_def): Likewise.
31082 * print-rtl.h (class rtx_reuse_manager): Likewise.
31084 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31086 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31088 (migrate_btr_def): Likewise.
31089 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31090 * df-core.c (loop_post_order_compute): Likewise.
31091 (loop_inverted_post_order_compute): Likewise.
31092 * hsa-common.h: Likewise.
31093 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31094 * init-regs.c (initialize_uninitialized_regs): Likewise.
31095 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31096 (inline_small_functions): Likewise.
31097 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31098 * ira.c (combine_and_move_insns): Likewise.
31099 (build_insn_chain): Likewise.
31100 * loop-invariant.c (find_invariants): Likewise.
31101 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31102 * predict.c (tree_predict_by_opcode): Likewise.
31103 (predict_paths_leading_to): Likewise.
31104 (predict_paths_leading_to_edge): Likewise.
31105 (estimate_loops_at_level): Likewise.
31106 (estimate_loops): Likewise.
31107 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31108 (spread_components): Likewise.
31109 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31110 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31111 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31112 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31113 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31114 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31115 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31116 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31117 (mark_threaded_blocks): Likewise.
31118 (thread_through_all_blocks): Likewise.
31119 * tree-ssa.c (verify_ssa): Likewise.
31120 (execute_update_addresses_taken): Likewise.
31121 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31123 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31125 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31127 (post_order_compute): Likewise.
31128 (inverted_post_order_compute): Likewise.
31129 (pre_and_rev_post_order_compute_fn): Likewise.
31131 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31133 * genrecog.c (int_set::int_set): Explicitly construct our
31134 auto_vec base class.
31135 * vec.h (auto_vec::auto_vec): New constructor.
31137 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31139 * bitmap.h (class auto_bitmap): New constructor taking
31140 bitmap_obstack * argument.
31142 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31144 * bitmap.h (class auto_bitmap): Change type of m_bits to
31145 bitmap_head, and adjust ctor / dtor and member operators.
31147 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31149 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31150 when returned register mode doesn't match original mode.
31152 2017-05-12 Jeff Law <law@redhat.com>
31153 Jakub Jelinek <jakub@redhat.com>
31155 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31156 we look for cc setter after the compare-elim changes.
31157 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31158 within the vector to match what compare-elim now expects.
31159 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31160 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31162 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31163 after the compare-elim changes.
31164 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31165 the vector to match what compare-elim now expects.
31166 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31167 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31168 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31169 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31170 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31172 * config/visium/visium.c (single_set_and_flags): Fix where
31173 we look for cc setter after the compare-elim changes.
31174 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31175 with the vector to match what compare-elim now expects.
31176 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31177 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31178 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31179 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31180 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31181 (neg<mode>2_insn_set_overflow): Likewise.
31183 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31185 PR middle-end/79794
31186 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31187 maybe_expand_insn call, set ops[0].target. If still set after call,
31188 set alt_rtl. Add extra arg to recursive calls.
31189 (extract_bit_field): Add alt_rtl argument. Pass to
31191 * expmed.h (extract_bit_field): Fix prototype.
31192 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31193 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31194 to extract_bit_field_calls.
31195 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31196 Pass alt_rtl to extract_bit_field calls.
31197 * calls.c (store_unaligned_arguments_into_psuedos)
31198 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31199 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31201 * optabs.h (struct expand_operand): Add target bitfield.
31203 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31205 * compare-elim.c (try_eliminate_compare): Canonicalize
31206 operation with embedded compare to
31207 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31208 (set (reg) (operation)].
31210 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31212 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31215 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31216 cost of adding a carry flag for ADC instruction.
31217 [case MINUS]: Ignore the cost of subtracting a carry flag
31218 for SBB instruction.
31220 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31222 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31224 * config/rs6000/bmiintrin.h: New file.
31225 * config/rs6000/bmi2intrin.h: New file.
31226 * config/rs6000/x86intrin.h: New file.
31228 2017-05-12 Jeff Law <law@redhat.com>
31230 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31233 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31235 PR middle-end/80707
31236 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31238 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31240 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31241 early expansion of vector divide builtins.
31242 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31243 builtins identified as having unsigned arguments.
31245 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31247 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31248 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31249 expansion of vector logical operations (and, andc, or, xor,
31252 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31254 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31255 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31257 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31259 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31260 early GIMPLE expansion of vector multiplies.
31262 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31264 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31265 TARGET_HAVE_MOVT conditional.
31266 (movt splitter): Likewise.
31268 2017-05-12 Richard Biener <rguenther@suse.de>
31270 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31271 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31272 Fold all stmts not inplace.
31274 2017-05-12 Richard Biener <rguenther@suse.de>
31276 PR tree-optimization/80713
31277 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31278 inserted_exprs bit for not removed stmts.
31280 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31282 PR middle-end/69921
31283 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31284 parallelized" attribute for parallelized OpenACC kernels.
31285 * omp-offload.c (execute_oacc_device_lower): Use it.
31287 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31288 Set "oacc kernels" attribute.
31289 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31290 parameter. Adjust all users.
31291 (oacc_fn_attrib_kernels_p): Remove function.
31292 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31293 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31294 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31295 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31296 assert "oacc kernels" attribute is set.
31298 2017-05-11 Carl Love <cel@us.ibm.com>
31300 * config/rs6000/rs6000-c: Add support for built-in functions
31301 vector unsigned char vec_popcnt (vector signed char)
31302 vector unsigned char vec_popcnt (vector unsigned char)
31303 vector unsigned short vec_popcnt (vector signed short)
31304 vector unsigned short vec_popcnt (vector unsigned short)
31305 vector unsigned int vec_popcnt (vector signed int)
31306 vector unsigned int vec_popcnt (vector unsigned int)
31307 vector unsigned long long vec_popcnt (vector signed long long)
31308 vector unsigned long long vec_popcnt (vector unsigned long long)
31309 vector signed long long vec_slo (vector signed long long,
31310 vector signed char)
31311 vector signed long long vec_slo (vector signed long long,
31312 vector unsigned char)
31313 vector unsigned long long vec_slo (vector unsigned long long,
31314 vector signed char)
31315 vector unsigned long long vec_slo (vector unsigned long long,
31316 vector unsigned char)
31317 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31318 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31319 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31320 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31321 * doc/extend.texi: Update the built-in documentation file for the
31322 new built-in functions.
31324 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31326 * attribs.h (sorted_attr_string): Move machine independent
31327 functions for target clone support from the i386 port to common
31328 code. Rename ix86_function_versions to common_function_versions.
31329 Rename make_name to make_unique_name.
31330 (common_function_versions): Likewise.
31331 (make_unique_name): Likewise.
31332 (make_dispatcher_decl): Likewise.
31333 (is_function_default_version): Likewise.
31334 * attribs.c (attr_strcmp): Likewise.
31335 (sorted_attr_string): Likewise.
31336 (common_function_versions): Likewise.
31337 (make_unique_name): Likewise.
31338 (make_dispatcher_decl): Likewise.
31339 (is_function_default_version): Likewise.
31340 * config/i386/i386.c (attr_strcmp): Likewise.
31341 (sorted_attr_string): Likewise.
31342 (ix86_function_versions): Likewise.
31343 (make_name): Likewise.
31344 (make_dispatcher_decl): Likewise.
31345 (is_function_default_version): Likewise.
31346 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31348 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31351 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31352 Account for direct move costs for vec_construct of integer
31355 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31358 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31359 (UNSPEC_STX_ATOMIC): Ditto.
31360 (loaddi_via_sse): New insn.
31361 (storedi_via_sse): Ditto.
31362 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31363 Update corresponding peephole2 patterns.
31364 (atomic_storedi_fpu): Ditto.
31366 2017-05-11 Julia Koval <julia.koval@intel.com>
31368 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31369 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31371 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31372 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31373 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31375 2017-05-11 Nathan Sidwell <nathan@acm.org>
31377 * graphite-poly.c: Include dumpfile.h.
31379 * dumpfle.h (dump_function): Declare here ...
31380 * tree-dump.h (dump_function): ... not here.
31381 * dumpfile.c: #include tree-cfg.h.
31382 (dump_function): Move here from ...
31383 * tree-dump.c (dump_function): ... here.
31384 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31385 * graphite-poly.c: Don't include tree-dump.h.
31386 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31387 * print-tree.c: Likewise.
31388 * stor-layout.c: Likewise.
31389 * tree-nested.c: Likewise.
31391 * dumpfile.c (dump_start): Use TDF_FLAGS.
31392 (dump_enable_all): Fix TDF_KIND check thinko.
31394 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31396 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31397 array entries to represent two legal parameterizations of the
31398 overloaded __builtin_cmpb function, as represented by the
31399 P6_OV_BUILTIN_CMPB constant.
31400 (altivec_resolve_overloaded_builtin): Add special case handling
31401 for the __builtin_cmpb function, as represented by the
31402 P6_OV_BUILTIN_CMPB constant.
31403 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31404 (BU_P6_64BIT_2): New macro.
31405 (BU_P6_OVERLOAD_2): New macro
31406 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31407 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31408 (CMPB): Add overload support to represent both 32-bit and 64-bit
31409 compare-bytes function.
31410 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31411 support for TARGET_CMPB.
31412 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31413 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31414 documentation of the __builtin_cmpb overloaded built-in function.
31416 2017-05-11 Richard Biener <rguenther@suse.de>
31418 PR tree-optimization/80705
31419 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31420 bases are not vectorizable.
31422 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31424 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31425 when counting register pressure.
31427 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31429 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31430 (iv_ca_more_deps): Renamed to ...
31431 (iv_ca_compare_deps): ... this.
31432 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31434 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31436 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31438 (determine_group_iv_costs): ... here.
31439 (find_inv_vars_cb): Record inv var if it's not recorded before.
31441 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31443 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31444 (get_shiftadd_cost): Ditto.
31446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31448 * tree-ssa-address.c: Include header file.
31449 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31451 (add_to_parts): Refactor.
31452 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31453 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31456 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31458 PR tree-optimization/53090
31459 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31461 (extract_cond_operands): Detect condition with IV on both sides
31462 and return COMP_IV_EXPR_2.
31463 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31464 (rewrite_use_compare): Simplify by removing call to function
31465 extract_cond_operands.
31467 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31469 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31470 (extract_cond_operands): Detect condition comparing against non-
31471 invariant bound and return appropriate enum value.
31472 (find_interesting_uses_cond): Update use of extract_cond_operands.
31473 Handle its return value accordingly.
31474 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31476 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31478 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31479 nonlinear iv_use computation in loop invariant sensitive way.
31481 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31483 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31484 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31486 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31488 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31489 (dump_cand): Support iv_cand.inv_exprs.
31490 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31492 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31495 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31497 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31499 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31500 as local function. Include necessary header files.
31501 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31503 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31505 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31507 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31509 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31510 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31511 RSHIFT_EXPR and BIT_NOT_EXPR.
31513 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31515 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31516 (adjust_setup_cost): New parameter supporting round up adjustment.
31517 (struct address_cost_data): Delete.
31518 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31519 (split_address_cost, ptr_difference_cost): Delete.
31520 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31521 (struct ainc_cost_data): New struct.
31522 (get_address_cost_ainc): New function.
31523 (get_address_cost, get_computation_cost): Reimplement.
31524 (determine_group_iv_cost_address): Record inv_expr for all uses of
31526 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31527 (iv_ca_has_deps): Reimplemented to ...
31528 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31530 (iv_ca_extend): Call iv_ca_more_deps.
31532 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31534 * tree-ssa-address.c (struct mem_address): Move to header file.
31535 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31536 * tree-ssa-address.h (struct mem_address): Move from C file.
31537 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31539 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31541 * tree-affine.h (aff_combination_type): New interface.
31542 (aff_combination_zero_p): Remove static.
31543 (aff_combination_const_p): New interface.
31544 (aff_combination_singleton_var_p): New interfaces.
31546 2017-05-11 Richard Biener <rguenther@suse.de>
31548 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31549 Skip unreachable blocks and destinations.
31550 (eliminate): Move stmt removal and fixup ...
31551 (fini_eliminate): ... here. Skip inserted exprs.
31552 (pass_pre::execute): Move fini_pre after fini_eliminate.
31553 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31554 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31555 PRE to get rid of dead code that has invalid SSA form and
31556 split critical edges again.
31558 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31560 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31562 2017-05-11 Richard Biener <rguenther@suse.de>
31564 * passes.c (execute_function_todo): Verify loops if they are
31565 said to be up-to-date.
31566 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31567 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31569 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31572 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31573 handle calling assemble_external ourself.
31576 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31577 modes with zero size. Enhance comment.
31579 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31581 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31582 built-ins for vec_xl and vec_xst with short and char pointer
31585 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31587 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31588 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31589 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31590 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31591 (_mm_maskz_min_round_ss): New intrinsics.
31592 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31593 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31594 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31595 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31596 (__builtin_ia32_minss_mask_round): New builtins.
31597 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31598 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31599 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31601 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31602 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31604 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31607 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31609 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31610 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31611 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31612 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31613 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31614 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31615 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31616 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31617 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31618 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31619 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31620 (__builtin_ia32_mulss_mask_round): New builtins.
31621 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31622 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31623 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31625 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31626 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31628 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31631 2017-05-10 Julia Koval <julia.koval@intel.com>
31633 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31634 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31635 (_mm256_setr_m128i): New intrinsics.
31637 2017-05-10 Julia Koval <julia.koval@intel.com>
31639 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31640 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31641 (_mm_maskz_rcp14_ss): New intrinsics.
31642 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31643 (__builtin_ia32_rcp14ss_mask): New builtins.
31644 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31646 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31648 PR tree-optimization/51513
31649 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31650 (assert_unreachable_fallthru_edge_p): Use it.
31651 (group_case_labels_stmt): Likewise.
31652 * tree-cfg.h: Prototype it.
31653 * stmt.c: Include cfghooks.h and tree-cfg.h.
31654 (emit_case_dispatch_table) <gap_label>: New local variable.
31655 Use it to fill dispatch table gaps.
31656 Test for default_label before updating probabilities.
31657 (expand_case) <default_label>: Remove unneeded initialization.
31658 Test for unreachable default case statement and remove its edge.
31659 Set default_label accordingly.
31660 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31662 2017-05-10 Carl Love <cel@us.ibm.com>
31664 * config/rs6000/rs6000-c: Add support for built-in functions
31665 vector signed char vec_neg (vector signed char)
31666 vector signed short int vec_neg (vector short int)
31667 vector signed int vec_neg (vector signed int)
31668 vector signed long long vec_neg (vector signed long long)
31669 vector float vec_neg (vector float)
31670 vector double vec_neg (vector double)
31671 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31673 * config/rs6000/altivec.h: Add define for vec_neg
31674 * doc/extend.texi: Update the built-in documentation for the
31675 new built-in functions.
31677 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31679 PR tree-optimization/77644
31680 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31682 2017-05-10 Nathan Sidwell <nathan@acm.org>
31684 * dumpfile.h (TDI_lang_all): New.
31685 (TDF_KIND): New. Renumber others
31686 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31688 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31690 (get_dump_file_name): Adjust suffix generation.
31691 (dump_enable_all): Use TDF_KIND.
31692 * doc/invoke.texi (-fdump-lang-all): Document.
31694 * dumpfile.h: Tabify.
31696 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31699 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31700 Move member access before delete.
31702 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31704 * tree-inline.c (expand_call_inline): Split block at stmt
31707 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31710 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31711 are now unused after splitting mov{sf,sd}_hardfloat.
31712 (f32_lr2): Likewise.
31713 (f32_lm): Likewise.
31714 (f32_lm2): Likewise.
31715 (f32_li): Likewise.
31716 (f32_li2): Likewise.
31717 (f32_lv): Likewise.
31718 (f32_sr): Likewise.
31719 (f32_sr2): Likewise.
31720 (f32_sm): Likewise.
31721 (f32_sm2): Likewise.
31722 (f32_si): Likewise.
31723 (f32_si2): Likewise.
31724 (f32_sv): Likewise.
31725 (f32_dm): Likewise.
31726 (f32_vsx): Likewise.
31727 (f32_av): Likewise.
31728 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31729 For movsf, order stores so the VSX stores occur before the GPR
31730 store which encourages the register allocator to use a traditional
31731 FPR instead of a GPR. For movsd, order the stores so that the GPR
31732 store comes before the VSX stores to allow the power6 to work.
31733 This is due to the power6 not having a 32-bit integer store
31734 instruction from a FPR.
31735 (movsf_hardfloat): Likewise.
31736 (movsd_hardfloat): Likewise.
31738 2017-05-09 Martin Sebor <msebor@redhat.com>
31740 PR translation/80280
31741 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31744 PR translation/80280
31745 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31746 data member added in r247778.
31747 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31749 2017-05-09 Nathan Sidwell <nathan@acm.org>
31751 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31753 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31756 2017-05-09 Marek Polacek <polacek@redhat.com>
31758 * doc/invoke.texi: Fix typo.
31760 2017-05-09 Richard Biener <rguenther@suse.de>
31762 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31763 (vrp_val_is_min): Likewise.
31764 (set_value_range_to_value): Likewise.
31765 (set_value_range_to_nonnegative): Likewise.
31766 (gimple_assign_nonzero_p): Likewise.
31767 (gimple_stmt_nonzero_p): Likewise.
31768 (vrp_int_const_binop): Likewise. Remove unreachable case.
31769 (adjust_range_with_scev): Adjust comments.
31770 (compare_range_with_value): Likewise.
31771 (extract_range_from_phi_node): Likewise.
31772 (test_for_singularity): Likewise.
31774 2017-05-09 Richard Biener <rguenther@suse.de>
31776 * tree-vrp.c (get_single_symbol): Add assert that we don't
31777 get overflowed constants as invariant part.
31778 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31779 checks. Use wi::cmp instead of recursing for integer constants.
31780 (compare_values): Just ignore whether we assumed undefined
31781 overflow instead of failing the compare.
31782 (extract_range_for_var_from_comparison_expr): Add comment before the
31783 TREE_NO_WARNING sets.
31784 (test_for_singularity): Likewise.
31785 (extract_range_from_comparison): Do not disable optimization
31786 when we assumed undefined overflow.
31787 (extract_range_basic): Remove init of unused var.
31789 2017-05-09 Richard Biener <rguenther@suse.de>
31791 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31792 (extract_range_from_multiplicative_op_1): Adjust.
31793 (extract_range_from_binary_expr_1): Use int_const_binop.
31795 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31798 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31799 rs6000_store_data_bypass_p in seven define_bypass directives and
31800 in several comments.
31801 * config/rs6000/rs6000-protos.h: Add prototype for
31802 rs6000_store_data_bypass_p function.
31803 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31804 function implements slightly different (rs6000-specific) semantics
31805 than store_data_bypass_p, returning false rather than aborting
31806 with assertion error when arguments do not satisfy the
31807 requirements of store data bypass.
31808 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31809 rs6000_store_data_bypass_p.
31811 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31813 * config/xtensa/xtensa-protos.h
31814 (xtensa_initial_elimination_offset): New declaration.
31815 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31816 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31817 macro definition, add case for FRAME_POINTER_REGNUM when
31818 FRAME_GROWS_DOWNWARD.
31819 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31820 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31821 xtensa_initial_elimination_offset.
31823 2017-05-08 Nathan Sidwell <nathan@acm.org>
31825 * doc/invoke.texi: Alphabetize -fdump options.
31827 2017-05-08 Martin Sebor <msebor@redhat.com>
31829 PR translation/80280
31830 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31832 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31834 * target.def (compute_frame_layout): New optional target hook.
31835 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31836 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31837 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31839 * reload1.c (verify_initial_elim_offsets): Likewise.
31840 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31841 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31842 (arm_get_frame_offsets): Split up into this ...
31843 (arm_compute_frame_layout): ... and this function.
31845 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31847 * config/aarch64/constraints.md (Usa): New constraint.
31848 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31850 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31852 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31853 with_multilib_list after it has been checked.
31855 2017-05-08 Richard Biener <rguenther@suse.de>
31857 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31858 (bitmap_set_subtract_values): Likewise.
31860 2017-05-08 Richard Biener <rguenther@suse.de>
31862 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31863 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31865 (gimple_stmt_nonzero_warnv_p): Rename to ...
31866 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31868 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31869 (extract_range_basic): Adjust, do not disable propagation on
31870 strict overflow sensitive simplification.
31871 (vrp_visit_cond_stmt): Likewise.
31873 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31875 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31876 body size unconditionally.
31878 2017-05-07 Jeff Law <law@redhat.com>
31881 2017-05-06 Jeff Law <law@redhat.com>
31882 PR tree-optimization/78496
31883 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31886 PR tree-optimization/78496
31887 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31888 (simplify_stmt_using_ranges): Call it.
31889 (vrp_dom_walker::before_dom_children): Extract equivalences
31890 from an ASSERT_EXPR with an equality comparison against a
31893 2017-05-06 Jeff Law <law@redhat.com>
31895 PR tree-optimization/78496
31896 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31899 PR tree-optimization/78496
31900 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31901 (simplify_stmt_using_ranges): Call it.
31902 (vrp_dom_walker::before_dom_children): Extract equivalences
31903 from an ASSERT_EXPR with an equality comparison against a
31906 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31908 * lra-constraints.c (lra_copy_reg_equiv): New function.
31909 (split_reg): Use it to copy equivalence information from the
31910 original register to the spill register.
31912 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31914 PR rtl-optimization/75964
31915 * simplify-rtx.c (simplify_const_relational_operation): Remove
31916 invalid handling of comparisons of integer ABS.
31918 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31920 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31921 initialize to zero.
31922 (init_regs): Remove declaration.
31923 (function_arg_advance_32): Initialize error_p as boolean variable.
31925 2017-05-05 Nathan Sidwell <nathan@acm.org>
31927 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31928 lines. Use for (;;).
31930 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31932 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31933 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31934 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31935 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31936 VF=2 that require versioning.
31938 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31940 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31943 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31945 * diagnostic.h (diagnostic_override_option_index): Convert from
31946 macro to inline function.
31948 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31950 * diagnostic.c (last_module_changed_p): New function.
31951 (set_last_module): New function.
31952 (diagnostic_report_current_module): Convert macro usage to
31953 the above functions.
31954 * diagnostic.h (diagnostic_context::last_module): Strengthen
31955 from const line_map * to const line_map_ordinary *.
31956 (diagnostic_last_module_changed): Delete macro.
31957 (diagnostic_set_last_module): Delete macro.
31959 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31961 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31962 with diagnostic_report_diagnostic.
31963 (diagnostic_n_impl_richloc): Likewise.
31964 * diagnostic.h (report_diagnostic): Delete macro.
31965 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31966 with diagnostic_report_diagnostic.
31967 * substring-locations.c (format_warning_va): Likewise.
31969 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31971 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31972 save/restor of format_spec. Move option-printing code to...
31973 (print_option_information): ...this new function, and
31974 reimplement by simply printing to the pretty_printer,
31975 rather than appending to the format string.
31977 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31979 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31980 handling logic into...
31981 (update_effective_level_from_pragmas): ...this new function.
31983 2017-05-04 Andrew Waterman <andrew@sifive.com>
31985 * config/riscv/riscv.opt (mstrict-align): New option.
31986 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31987 (SLOW_UNALIGNED_ACCESS): Define.
31988 (riscv_slow_unaligned_access): Declare.
31989 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31991 (riscv_slow_unaligned_access): New variable.
31992 (rocket_tune_info): Set slow_unaligned_access to true.
31993 (optimize_size_tune_info): Set slow_unaligned_access to false.
31994 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31995 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31996 (riscv_option_override): Set riscv_slow_unaligned_access.
31997 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31999 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32001 * config/riscv/riscv.md: Unify indentation.
32003 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32008 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32010 (extendsi<mode>2): Add support for doing sign extension via
32011 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32012 don't have ISA 3.0 instructions.
32013 (extendsi<mode>2 splitter): Likewise.
32014 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32015 generate the normal insns since SImode can now go in vector
32016 registers. Disallow the special UNSPECs needed for previous
32017 machines to hide SImode being used. Add new insns
32018 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32019 (fix_trunc<mode>si2_stfiwx): Likewise.
32020 (fix_trunc<mode>si2_internal): Likewise.
32021 (fixuns_trunc<mode>si2): Likewise.
32022 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32023 (fctiw<u>z_<mode>_smallint): Likewise.
32024 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32025 of floating point to 32-bit integer from doing a direct move to
32026 the GPR registers to do a store.
32027 (fctiwz_<mode>): Break long line.
32029 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32031 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32032 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32033 (addr_list, addr_offset_valid_p): New.
32034 (split_address_groups): Check offset validity with above function.
32035 (gt-tree-ssa-loop-ivopts.h): Include header file.
32037 2017-05-05 Nathan Sidwell <nathan@acm.org>
32039 * config.gcc (arm*-*-*): Add missing 'fi'.
32041 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32043 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32044 included in -fopt-info does not matter.
32045 * doc/optinfo.texi (-fopt-info): Fix description of default
32046 behavour. Explicitly say order of options included in -fopt-info
32049 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32051 * config.gcc: Allow combinations of aprofile and rmprofile values for
32052 --with-multilib-list.
32053 * config/arm/t-multilib: New file.
32054 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32055 variables. Remove setting of ISA and floating-point ABI in
32056 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32057 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32058 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32059 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32061 * config/arm/t-rmprofile: Likewise except for the matches changes.
32062 * doc/install.texi (--with-multilib-list): Document the combination of
32063 aprofile and rmprofile values and warn about pitfalls in doing that.
32065 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32067 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32068 (movdi_aarch64): Likewise.
32070 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32072 PR tree-optimization/80632
32073 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32075 (build_arrays): Initialize it for virtual phis.
32076 (fix_phi_nodes): Use it for virtual phis.
32078 PR tree-optimization/80558
32079 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32080 [x, y] op z into [x op, y op z] for op & or | if conditions
32083 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32084 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32087 * config/arm/arm.md (use_literal_pool): Remove.
32088 (64-bit immediate split): No longer takes cost into consideration
32089 if arm_disable_literal_pool is enabled.
32090 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32091 used when arm_disable_literal_pool is enabled.
32092 (arm_max_const_double_inline_cost): Remove use of
32093 arm_disable_literal_pool.
32094 (push_minipool_fix): Add assert.
32095 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32096 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32097 (no_literal_pool_sf_immediate): New.
32099 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32101 PR tree-optimization/80613
32102 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32103 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32105 2017-05-05 Richard Biener <rguenther@suse.de>
32107 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32109 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32111 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32112 of this flag from insn conditions due to removal from r247495.
32114 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32116 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32118 (arm_early_store_addr_dep_ptr): Likewise.
32119 * config/arm/aarch-common-protos.h
32120 (arm_early_load_addr_dep_ptr): Add prototype.
32121 (arm_early_store_addr_dep_ptr): Likewise.
32122 * config/arm/cortex-a53.md: Add new bypasses.
32124 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32126 * tree.c (next_type_uid): Change type to unsigned.
32127 (type_hash_canon): Decrement back next_type_uid if
32128 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32129 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32132 2017-05-04 Martin Sebor <msebor@redhat.com>
32134 * builtins.c: Fix a trivial typo in a comment.
32136 PR middle-end/79234
32137 * builtins.c (check_sizes): Adjust to handle reading past the end.
32138 Avoid printing excessive upper bound of ranges. Use %E to print
32139 tree nodes instead of converting them to %wu.
32140 (expand_builtin_memchr): New function.
32141 (compute_dest_size): Rename...
32142 (compute_objsize): ...to this.
32143 (expand_builtin_memcpy): Adjust.
32144 (expand_builtin_mempcpy): Adjust.
32145 (expand_builtin_strcat): Adjust.
32146 (expand_builtin_strcpy): Adjust.
32147 (check_strncat_sizes): Adjust.
32148 (expand_builtin_strncat): Adjust.
32149 (expand_builtin_strncpy): Adjust and simplify.
32150 (expand_builtin_memset): Adjust.
32151 (expand_builtin_bzero): Adjust.
32152 (expand_builtin_memcmp): Adjust.
32153 (expand_builtin): Handle memcmp.
32154 (maybe_emit_chk_warning): Check strncat just once.
32156 2017-05-04 Martin Sebor <msebor@redhat.com>
32158 PR preprocessor/79214
32159 PR middle-end/79222
32160 PR middle-end/79223
32161 * builtins.c (check_sizes): Add inlining context and issue
32162 warnings even when -Wno-system-headers is set.
32163 (check_strncat_sizes): Same.
32164 (expand_builtin_strncat): Same.
32165 (expand_builtin_memmove): New function.
32166 (expand_builtin_stpncpy): Same.
32167 (expand_builtin): Handle memmove and stpncpy.
32169 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32171 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32172 which is not used any more.
32174 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32176 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32178 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32180 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32181 (cortexa53_tunings): Likewise.
32182 (cortexa57_tunings): Likewise.
32183 (cortexa72_tunings): Likewise.
32184 (cortexa73_tunings): Likewise.
32186 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32188 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32189 Set loop alignment to 8.
32191 2017-05-04 Martin Sebor <msebor@redhat.com>
32193 PR translation/80280
32194 * builtins.c (expand_builtin_object_size): Add missing quoting to
32195 %D and like directives.
32196 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32197 (hsa_type_for_tree_type): Same.
32198 (verify_function_arguments): Same.
32199 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32200 * varasm.c (get_section): Same.
32203 2017-05-04 Martin Sebor <msebor@redhat.com>
32205 PR translation/80280
32206 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32208 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32210 * config/aarch64/aarch64.c (generic_addrcost_table):
32211 Change HI/TI mode setting.
32213 2017-05-04 Martin Jambor <mjambor@suse.cz>
32215 PR tree-optimization/80622
32216 * tree-sra.c (comes_initialized_p): New function.
32217 (build_accesses_from_assign): Only set write lazily when
32218 comes_initialized_p is false.
32219 (analyze_access_subtree): Use comes_initialized_p.
32220 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32221 instead of testing for PARM_DECL.
32223 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32225 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32226 constraint on operand 0 to allow more general addressing modes.
32227 Adjust output template.
32228 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32230 * config/aarch64/aarch64-protos.h
32231 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32232 * config/aarch64/constraints.md (Dp): New address constraint.
32233 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32236 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32238 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32239 update use of estimate_ipcp_clone_size_and_time.
32240 (estimate_local_effects): Update use of
32241 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32242 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32243 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32244 Return nonspecialized time.
32246 2017-05-04 Richard Biener <rguenther@suse.de>
32248 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32249 for the last VUSE which def dominates the PHI. Directly call
32251 (get_continuation_for_phi_1): Remove.
32253 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32255 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32256 to explain the use of truncating division. Cap the number of
32257 iterations to the maximum given by nb_iterations_upper_bound,
32260 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32262 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32263 * configure: Regenerate.
32264 * config.in: Regenerate.
32265 * config/i386/driver-mingw32.c: new file.
32266 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32267 * config.host: Link driver-mingw32.o on MinGW host.
32268 * doc/install.texi: Document new --enable-mingw-wildcard configure
32271 2017-05-04 Marek Polacek <polacek@redhat.com>
32273 PR tree-optimization/80612
32274 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32276 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32277 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32279 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32280 (movt splitter): Likewise.
32281 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32282 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32283 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32284 block for Thumb-1 with MOVT.
32285 (thumb2_legitimate_address_p): Move code block ...
32286 (can_avoid_literal_pool_for_label_p): ... into this new function.
32287 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32289 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32290 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32291 "M-profile targets with the MOVT instruction".
32293 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32295 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32296 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32297 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32299 2017-05-04 Martin Liska <mliska@suse.cz>
32301 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32302 variable cond_code.
32304 2017-05-04 Richard Biener <rguenther@suse.de>
32306 * tree.c (array_at_struct_end_p): Handle arrays at struct
32307 end with flexarrays more conservatively. Refactor and treat
32308 arrays of arrays or aggregates more strict. Fix
32309 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32310 * tree.c (array_at_struct_end_p): Adjust prototype.
32311 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32312 * gimple-fold.c (get_range_strlen): Likewise.
32313 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32315 2017-05-04 Richard Biener <rguenther@suse.de>
32317 PR tree-optimization/31130
32318 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32320 (supports_overflow_infinity): Likewise.
32321 (is_negative_overflow_infinity): Likewise.
32322 (is_positive_overflow_infinity): Likewise.
32323 (is_overflow_infinity): Likewise.
32324 (stmt_overflow_infinity): Likewise.
32325 (overflow_infinity_range_p): Likewise.
32326 (usable_range_p): Remove as always returning true.
32327 (make_overflow_infinity): Remove.
32328 (negative_overflow_infinity): Likewise.
32329 (positive_overflow_infinity): Likewise.
32330 (avoid_overflow_infinity): Likewise.
32331 (set_value_range): Adjust accordingly.
32332 (set_value_range_to_nonnegative): Likewise, remove now unused
32333 overflow_infinity arg.
32334 (vrp_operand_equal_p): Adjust.
32335 (update_value_range): Likewise.
32336 (range_int_cst_singleton_p): Likewise.
32337 (operand_less_p): Likewise.
32338 (compare_values_warnv): Likewise.
32339 (extract_range_for_var_from_comparison_expr): Likewise.
32340 (vrp_int_const_binop): Likewise.
32341 (zero_nonzero_bits_from_vr): Likewise.
32342 (extract_range_from_multiplicative_op_1): Likewise.
32343 (extract_range_from_binary_expr_1): Likewise.
32344 (extract_range_from_unary_expr): Likewise.
32345 (extract_range_from_comparison): Likewise.
32346 (extract_range_basic): Likewise.
32347 (adjust_range_with_scev): Likewise.
32348 (compare_ranges): Likewise.
32349 (compare_range_with_value): Likewise.
32350 (dump_value_range): Likewise.
32351 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32353 (simplify_cond_using_ranges): Adjust.
32355 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32357 * brig-builtins.def: Added a builtin for class_f64.
32358 * builtin-types.def: Added a builtin type needed by class_f64.
32360 2017-05-03 Jason Merrill <jason@redhat.com>
32362 * timevar.def: Add TV_CONSTEXPR.
32364 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32366 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32368 2017-05-03 Martin Jambor <mjambor@suse.cz>
32370 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32371 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32372 * ipa-cp.c (ipcp_propagate_stage): Do not call
32373 ipa_update_after_lto_read.
32374 * ipa-inline.c (ipa_inline): Likewise.
32376 2017-05-03 Martin Jambor <mjambor@suse.cz>
32378 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32379 tag. Added a default constructor and a destructor.
32380 (ipa_edge_args_sum_t): New class;
32381 (ipa_edge_args_sum): Declare.
32382 (ipa_edge_args_vector): Remove declaration.
32383 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32384 (ipa_free_edge_args_substructures): Remove declaration.
32385 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32386 (ipa_edge_args_info_available_for_edge_p): Likewise.
32387 * ipa-prop.c (ipa_edge_args_vector): Removed.
32388 (edge_removal_hook_holder): Likewise.
32389 (edge_duplication_hook_holder): Likewise.
32390 (ipa_edge_args_sum): New variable.
32391 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32392 ipa_edge_args_vector.
32393 (ipa_free_edge_args_substructures): Likewise.
32394 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32395 ipa_edge_args_vector.
32396 (ipa_edge_removal_hook): Turned into method
32397 ipa_edge_args_sum_t::remove.
32398 (ipa_edge_duplication_hook): Turned into method
32399 ipa_edge_args_sum_t::duplicate.
32400 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32401 registering edge hooks.
32402 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32403 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32404 ipa_edge_args_sum instead of ipa_edge_args_vector.
32405 * ipa-profile.c (ipa_profile): Likewise.
32407 2017-05-03 Martin Jambor <mjambor@suse.cz>
32409 * symbol-summary.h (function_summary): New method exists.
32410 (function_summary::symtab_removal): Deallocate through release.
32411 (call_summary): New class.
32412 (gt_ggc_mx): New overload.
32413 (gt_pch_nx): Likewise.
32414 (gt_pch_nx): Likewise.
32416 2017-05-03 Jeff Law <law@redhat.com>
32418 PR tree-optimization/78496
32419 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32420 from simplify_cond_using_ranges. Split off code to walk
32421 backwards through casts into ...
32422 (simplify_cond_using_ranges_2): New function.
32423 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32424 (execute_vrp): After identifying jump threads, call
32425 simplify_cond_using_ranges_2.
32427 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32430 * ipa-inline.h (inline_summary): Add ctor.
32431 (create_ggc): Do not use ggc_cleared_alloc.
32433 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32434 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32436 * gcc.c (handle_braces): Support escaping in switch matching
32438 * doc/invoke.texi (Spec Files): Document it.
32439 Remove superfluous @code markup in items.
32441 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32443 * diagnostic-show-locus.c (struct column_range): New struct.
32444 (get_affected_columns): New function.
32445 (get_printed_columns): New function.
32446 (struct correction): New struct.
32447 (correction::ensure_capacity): New function.
32448 (correction::ensure_terminated): New function.
32449 (struct line_corrections): New struct.
32450 (line_corrections::~line_corrections): New dtor.
32451 (line_corrections::add_hint): New function.
32452 (layout::print_trailing_fixits): Reimplement in terms of the new
32454 (selftest::test_overlapped_fixit_printing): New function.
32455 (selftest::diagnostic_show_locus_c_tests): Call it.
32457 2017-05-03 Nathan Sidwell <nathan@acm.org>
32459 Canonicalize canonical type hashing
32460 * tree.h (type_hash_canon_hash): Declare.
32461 * tree.c (type_hash_list, attribute_hash_list): Move into
32462 type_hash_canon_hash.
32463 (build_type_attribute_qual_variant): Break out hash code calc into
32464 type_hash_canon_hash.
32465 (type_hash_canon_hash): New. Generic type hash computation.
32466 (build_range_type_1, build_array_type_1, build_function_type,
32467 build_method_type_directly, build_offset_type, build_complex_type,
32468 make_vector_type): Call it.
32470 2017-05-03 Richard Biener <rguenther@suse.de>
32472 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32473 When all DRs have unknown misaligned do not always peel
32474 when there is a store but apply the same costing model as if
32475 there were only loads.
32477 2017-05-03 Richard Biener <rguenther@suse.de>
32480 PR tree-optimization/80492
32481 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32482 compare_base_decls returning dont-know properly.
32484 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32486 * config/arm/iterators.md (CCSI): New mode iterator.
32487 (arch): New mode attribute.
32488 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32489 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32490 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32491 code iterator for success result mode.
32492 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32493 the corresponding new insn generators.
32495 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32498 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32499 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32501 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32503 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32504 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32505 (DDR_A): Wrap DDR argument in brackets.
32506 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32507 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32508 (DDR_REVERSED_P): Likewise.
32510 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32512 PR tree-optimization/79472
32513 * tree-switch-conversion.c (struct switch_conv_info): Add
32514 contiguous_range and default_case_nonstandard fields.
32515 (collect_switch_conv_info): Compute contiguous_range and
32516 default_case_nonstandard fields, don't clear final_bb if
32517 contiguous_range and only the default case doesn't have the required
32519 (check_all_empty_except_final): Set default_case_nonstandard instead
32520 of failing if contiguous_range and the default case doesn't have empty
32522 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32523 and only the default case doesn't have the required constants. Skip
32525 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32526 if default_case_nonstandard.
32527 (build_constructors): Build constant 1 just once. Assert that default
32528 values aren't inserted in between cases if contiguous_range. Skip
32530 (build_arrays): Skip virtual phis.
32531 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32532 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32533 Handle virtual phis.
32534 (gen_inbound_check): Handle default_case_nonstandard case.
32535 (process_switch): Adjust check_final_bb caller. Call
32536 gather_default_values with the first non-default case instead of
32537 default case if default_case_nonstandard.
32539 2017-05-02 Nathan Sidwell <nathan@acm.org>
32541 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32542 check. Fix formatting.
32544 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32546 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32547 errors when comparing specialized and unspecialized times.
32549 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32551 * diagnostic-show-locus.c
32552 (layout::should_print_annotation_line_p): Make private.
32553 (layout::print_annotation_line): Make private.
32554 (layout::annotation_line_showed_range_p): Make private.
32555 (layout::show_ruler): Make private.
32556 (layout::print_source_line): Make private. Pass in line and
32557 line_width, rather than calling location_get_source_line. Drop
32559 (layout::print_leading_fixits): New method.
32560 (layout::print_any_fixits): Rename to...
32561 (layout::print_trailing_fixits): ...this, and make private.
32562 Don't print newline fixits.
32563 (diagnostic_show_locus): Move logic for printing one row into...
32564 (layout::print_line): ...this new function. Move the
32565 location_get_source_line call and error-handling from
32566 print_source_line to here. Call print_leading_fixits, and rename
32567 print_any_fixits to print_trailing_fixits.
32568 (selftest::test_fixit_insert_containing_newline): Update now that
32569 newlines are partially supported.
32570 (selftest::test_fixit_insert_containing_newline_2): New test.
32571 (selftest::test_fixit_replace_containing_newline): Update comments.
32572 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32573 * edit-context.c (class added_line): New class.
32574 (class edited_line): Describe newline handling in comment.
32575 (edited_line::actually_edited_p): New method.
32576 (edited_line::print_content): Delete redundant decl.
32577 (edited_line::m_predecessors): New field.
32578 (edited_file::print_content): Call edited_line::print_content.
32579 (edited_file::print_diff): Update to support newlines.
32580 (edited_file::print_diff_hunk): Likewise.
32581 (edited_file::print_run_of_changed_lines): New function.
32582 (edited_file::print_diff_line): Convert to...
32583 (print_diff_line): ...this.
32584 (edited_file::get_effective_line_count): New function.
32585 (edited_line::edited_line): Initialize new field m_predecessors.
32586 (edited_line::~edited_line): Clean up m_predecessors.
32587 (edited_line::apply_fixit): Handle newlines.
32588 (edited_line::get_effective_line_count): New function.
32589 (edited_line::print_content): New function.
32590 (edited_line::print_diff_lines): New function.
32591 (selftest::test_applying_fixits_insert_containing_newline): New
32593 (selftest::test_applying_fixits_replace_containing_newline): New
32595 (selftest::insert_line): New function.
32596 (selftest::test_applying_fixits_multiple_lines): Add example of
32598 (selftest::edit_context_c_tests): Call the new tests.
32600 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32602 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32603 parameter cand. Update dump information.
32604 (get_computation_cost): Update uses.
32606 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32608 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32609 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32610 (get_computation_at, rewrite_use_address): Update use of
32611 get_computation_aff.
32613 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32615 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32616 (get_computation): Delete.
32617 (get_computation_cost): Implement like get_computation_cost_at.
32618 Use get_computation_at.
32619 (get_computation_cost_at): Delete.
32620 (rewrite_use_nonlinear_expr): Use get_computation_at.
32621 (rewrite_use_compare, remove_unused_ivs): Ditto.
32623 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32625 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32627 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32629 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32630 (ivopts_global_cost_for_size): Rename parameter and update uses.
32631 (iv_ca_recount_cost): Update uses.
32632 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32633 candidates seperately in n_invs and n_cands.
32634 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32636 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32638 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32639 (find_inv_vars_cb): New.
32640 (find_depends): Renamed to ...
32641 (find_inv_vars): ... this.
32642 (add_candidate_1, force_var_cost): Call find_inv_vars.
32643 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32645 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32647 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32648 inv_vars. Add inv_exprs.
32649 (struct iv_cand): Rename depends_on to inv_vars.
32650 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32651 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32652 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32653 (dump_cand): Dump inv_vars.
32654 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32655 (record_invariant, find_depends, add_candidate_1): Ditto.
32656 (set_group_iv_cost, force_var_cost): Ditto.
32657 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32658 (get_computation_cost_at, get_computation_cost): Ditto.
32659 (determine_group_iv_cost_generic): Ditto.
32660 (determine_group_iv_cost_address): Ditto.
32661 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32662 (determine_group_iv_costs): Ditto.
32663 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32664 (iv_ca_set_remove_invariants): Renamed to ...
32665 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32666 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32667 (iv_ca_set_add_invariants): Renamed to ...
32668 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32669 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32670 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32671 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32672 (create_new_ivs): Remove useless dump.
32674 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32676 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32678 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32679 (iv_ca_set_no_cp, create_new_iv): Ditto.
32681 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32683 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32685 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32687 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32688 function tree_check2.
32690 2017-05-02 Martin Liska <mliska@suse.cz>
32692 * doc/gcov.texi: Add missing preposition.
32693 * gcov.c (function_info::function_info): Properly fill up
32694 all member variables.
32696 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32698 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32700 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32702 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32704 2017-05-02 Martin Liska <mliska@suse.cz>
32707 * lto-streamer-in.c (lto_read_tree_1): Remove
32708 LTO_STREAMER_DEBUG.
32709 * lto-streamer.c (struct tree_hash_entry): Likewise.
32710 (struct tree_entry_hasher): Likewise.
32711 (tree_entry_hasher::hash): Likewise.
32712 (tree_entry_hasher::equal): Likewise.
32713 (lto_streamer_init): Likewise.
32714 (lto_orig_address_map): Likewise.
32715 (lto_orig_address_get): Likewise.
32716 (lto_orig_address_remove): Likewise.
32717 * lto-streamer.h: Likewise.
32718 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32719 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32721 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32723 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32724 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32725 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32726 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32727 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32728 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32729 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32730 (mm_maskz_sub_ss): New intrinsics.
32731 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32732 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32733 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32734 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32735 (__builtin_ia32_subss_mask_round): New builtins.
32736 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32737 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32738 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32740 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32741 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32743 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32746 2017-05-02 Martin Jambor <mjambor@suse.cz>
32748 PR tree-optimization/78687
32749 * tree-sra.c (access): New field parent.
32750 (process_subtree_disqualification): New function.
32751 (disqualify_candidate): Call it.
32752 (build_accesses_from_assign): Reset write flag if creating an
32754 (build_access_subtree): Fill in parent field and also prpagate
32755 down grp_write flag.
32756 (create_artificial_child_access): New parameter set_grp_write, set
32757 grp_write to its value.
32758 (propagate_subaccesses_across_link): Also propagate grp_write flag
32760 (propagate_all_subaccesses): Push the closest parent back to work
32761 queue if add_access_to_work_queue returned true.
32763 2017-05-02 Richard Biener <rguenther@suse.de>
32765 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32766 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32767 -fstrict-overflow documentation.
32768 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32769 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32770 flag_strict_overflow.
32771 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32772 * lto-opts.c (lto_write_options): Do not stream it.
32773 * lto-wrapper.c (merge_and_complain): Do not handle it.
32774 * opts.c (default_options_table): Do not set -fstrict-overflow.
32775 (finish_options): Likewise do not clear it when sanitizing.
32776 * simplify-rtx.c (simplify_const_relational_operation): Do not
32777 test flag_strict_overflow.
32779 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32781 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32782 using enabled attribute.
32783 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32784 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32785 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32786 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32787 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32788 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32789 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32790 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32791 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32792 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32794 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32796 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32798 2017-05-02 Richard Biener <rguenther@suse.de>
32800 PR tree-optimization/80591
32802 2017-04-10 Richard Biener <rguenther@suse.de>
32804 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32807 2017-05-02 Richard Biener <rguenther@suse.de>
32809 PR tree-optimization/80549
32810 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32811 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32812 headers if they do not have a preheader.
32814 2017-05-02 Martin Liska <mliska@suse.cz>
32817 * common.opt: Fix typo.
32818 * doc/invoke.texi: Likewise.
32820 2017-05-01 Jan Beulich <jbeulich@suse.com>
32822 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32823 swapping, add (x,x,m,x,n) alternative.
32825 2017-05-01 Nathan Sidwell <nathan@acm.org>
32827 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32828 unnecessary unadjusted_alignment check.
32830 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32833 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32835 * gimplify.c (gimplify_cilk_detach): New function.
32836 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32837 * tree-core.h: Document EXPR_CILK_SPAWN.
32838 * tree.h (EXPR_CILK_SPAWN): Define.
32840 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32842 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32843 to use new fixit_hint representation, using the "replace" logic.
32844 (get_line_span_for_fixit_hint): Likewise.
32845 (layout::print_any_fixits): Likewise.
32846 (selftest::test_one_liner_many_fixits): Rename to...
32847 (selftest::test_one_liner_many_fixits_1): ...this, and update
32848 comment and expected output to reflect that the multiple fix-it
32849 hints are now consolidated into one insertion.
32850 (selftest::test_one_liner_many_fixits_2): New test.
32851 (selftest::test_diagnostic_show_locus_one_liner): Update for
32853 (selftest::test_fixit_consolidation): Update for fix-it API
32855 * diagnostic.c (print_parseable_fixits): Likewise.
32856 * edit-context.c (edited_line::m_line_events): Convert from
32857 auto_vec <line_event *> to auto_vec <line_event>.
32858 (class line_event): Convert from abstract base class to a concrete
32859 class, taking over the role of replace_event.
32860 (class insert_event): Delete.
32861 (class replace_event): Rename to class line_event. Convert to
32863 (edit_context::add_fixits): Reimplement.
32864 (edit_context::apply_insert): Delete.
32865 (edit_context::apply_replace): Rename to...
32866 (edit_context::apply_fixit): ...this. Convert to half-open range.
32867 (edited_file::apply_insert): Delete.
32868 (edited_file::apply_replace): Rename to...
32869 (edited_file::apply_fixit): ...this.
32870 (edited_line::~edited_line): Drop deletion of events.
32871 (edited_line::apply_insert): Delete.
32872 (edited_line::apply_replace): Rename to...
32873 (edited_line::apply_fixit): ...this. Convert to half-open range.
32874 Update for change to type of m_line_events.
32875 * edit-context.h (edit_context::apply_insert): Delete.
32876 (edit_context::apply_replace): Rename to...
32877 (edit_context::apply_fixit): ...this.
32879 2017-05-01 Martin Sebor <msebor@redhat.com>
32881 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32884 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32887 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32888 __get_cpuid_max returns 0.
32889 (__get_cpuid_count): Ditto.
32891 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32893 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32894 replacement expression is another instance of one of its arguments.
32896 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32899 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32900 check for stack push/pop autoinc.
32901 * config/i386/i386.c (ix86_agi_dependent): Return false
32902 if the only reason why modified_in_p returned true is that
32903 addr is SP based and set_insn is a push or pop.
32905 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32907 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32910 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32913 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32914 (account_size_time): Use two predicates - exec_pred and
32916 (evaluate_conditions_for_known_args): Compute both clause and
32918 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32919 (inline_summary_t::duplicate): Update.
32920 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32922 (compute_inline_parameters): Likewise.
32923 (estimate_edge_size_and_time): Update caluclation of time.
32924 (estimate_node_size_and_time): Compute both time and nonspecialized
32926 (estimate_ipcp_clone_size_and_time): Update.
32927 (inline_merge_summary): Update.
32928 (do_estimate_edge_time): Update.
32929 (do_estimate_edge_size): Update.
32930 (do_estimate_edge_hints): Update.
32931 (inline_read_section, inline_write_summary): Stream both new predicates.
32932 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32934 (compute_inlined_call_time): Cleanup.
32935 (big_speedup_p): Update.
32936 (edge_badness): Update.
32937 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32938 (size_time_entry): Replace predicate by exec_predicate and
32939 nonconst_predicate.
32940 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32941 (estimate_edge_time): Return also nonspec_time.
32942 (reset_edge_growth_cache): Update.
32944 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32946 PR rtl-optimization/80491
32947 * ifcvt.c (noce_process_if_block): When looking for x setter
32948 with missing else_bb, don't check only the insn right before
32949 cond_earliest, but look for the last insn that x is modified in
32950 within the same bb.
32952 PR rtl-optimization/80491
32953 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32955 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32957 PR tree-optimization/80487
32958 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32960 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32962 PR tree-optimization/79697
32963 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32964 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32965 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32967 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32968 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32970 2017-04-28 Martin Sebor <msebor@redhat.com>
32972 PR tree-optimization/80523
32973 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32974 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32976 (maybe_warn, format_directive, parse_directive): Use new functions.
32977 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32979 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32981 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32983 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32985 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32986 target_header_dir): Set correctly.
32987 * configure: Regenerated.
32988 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32989 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32990 instead of SYSTEM_HEADER_DIR.
32992 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32994 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32995 (estimate_local_effects): Likewise.
32996 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32997 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32998 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32999 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33000 do_estimate_edge_time, estimate_edge_time): Likewise.
33001 * ipa-inline-analysis.c (estimate_node_size_and_time,
33002 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33003 (estimate_time_after_inlining): Remove.
33005 2017-04-28 Martin Liska <mliska@suse.cz>
33007 * doc/gcov.texi: Enhance documentation of gcov.
33009 2017-04-28 Martin Liska <mliska@suse.cz>
33011 * doc/gcov.texi: Sort options in alphabetic order.
33012 * doc/gcov-dump.texi: Likewise.
33013 * doc/gcov-tool.texi: Likewise.
33014 * gcov.c (print_usage): Likewise.
33015 * gcov-dump.c (print_usage): Likewise.
33016 * gcov-tool.c (print_merge_usage_message): Likewise.
33017 (print_rewrite_usage_message): Likewise.
33018 (print_overlap_usage_message): Likewise.
33020 2017-04-28 Martin Liska <mliska@suse.cz>
33022 PR gcov-profile/53915
33023 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33025 2017-04-28 Martin Liska <mliska@suse.cz>
33027 PR gcov-profile/79891
33028 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33029 is marked by compiler as living on a line.
33030 (get_cycles_count): Remove usage of the union.
33031 (output_intermediate_file): Likewise.
33032 (find_source): Fix GNU coding style.
33033 (accumulate_line_counts): Remove old non-all block mode.
33034 (output_lines): Remove usage of the union.
33035 * profile.c (output_location): Include all BBs, even if
33036 belonging to a same line (and file) as a previous BB.
33038 2017-04-28 Martin Liska <mliska@suse.cz>
33040 * gcov.c (process_args): Handle new argument 'w'.
33041 (read_graph_file): Assign ID to BBs.
33042 (output_branch_count): Display BB # if verbose flag is set.
33043 (output_lines): Likewise for arcs.
33044 (print_usage): Add '--verbose' option help.
33045 * doc/gcov.texi: Document --verbose (-w) option.
33047 2017-04-28 Martin Liska <mliska@suse.cz>
33049 * gcov.c (struct block_location_info): New struct.
33050 (process_file): Fill up the new structure.
33051 (read_graph_file): Replace usage of encoding by the newly added
33053 (add_line_counts): Likewise.
33054 (accumulate_line_counts): Remove usage of the union.
33055 (function_info::function_info): New function.
33056 (function_info::~function_info): Likewise.
33057 (process_file): Call delete instead of release_function.
33058 (release_function): Release the function.
33059 (release_structures): Call delete instead of release_function.
33060 (solve_flow_graph): Replace usage of num_blocks.
33061 (find_exception_blocks): Likewise.
33062 (output_lines): Fix GNU coding style.
33064 2017-04-28 Martin Liska <mliska@suse.cz>
33067 * coverage.c (coverage_remove_note_file): New function.
33068 * coverage.h: Declare the function.
33069 * toplev.c (finalize): Clean if an error has been seen.
33071 2017-04-28 Martin Liska <mliska@suse.cz>
33073 PR gcov-profile/80031
33074 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33075 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33076 * gcov.c (read_graph_file): Read just number of blocks.
33077 * profile.c (branch_prob): Do not stream 0 flags per a basic
33080 2017-04-28 Martin Liska <mliska@suse.cz>
33082 * gcov-dump.c (tag_*): Add new argument to declarations.
33083 (dump_gcov_file): Likewise.
33084 (tag_blocks): Add and use new argument depth.
33085 (tag_arcs): Likewise.
33086 (tag_lines): Likewise.
33087 (tag_counters): Likewise.
33088 (tag_summary): Likewise.
33089 (dump_working_sets): Use depth to do a proper indentation.
33091 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33094 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33095 * symtab.c (symtab_node::debug_symtab): Move definition here.
33097 2017-04-28 Richard Biener <rguenther@suse.de>
33099 * lto-streamer.h (LTO_major_version): Bump to 7.
33101 2017-04-28 Richard Biener <rguenther@suse.de>
33103 * tree-vrp.c (assert_info): New struct.
33104 (add_assert_info): New helper.
33105 (register_edge_assert_for_2): Refactor to add asserts to a vector
33107 (register_edge_assert_for_1): Likewise.
33108 (register_edge_assert_for): Likewise.
33109 (finish_register_edge_assert_for): New helper actually registering
33110 asserts where live on edge.
33111 (find_conditional_asserts): Adjust.
33112 (find_switch_asserts): Likewise.
33113 (evrp_dom_walker::try_find_new_range): Generalize.
33114 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33116 2017-04-27 Marek Polacek <polacek@redhat.com>
33119 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33120 arg10 and arg11 to itype.
33122 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33124 * doc/extend.texi (Object Size Checking): Improve grammar.
33126 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33129 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33130 that the logic for permitting reciprocal estimates matches that
33133 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33136 * tree.c (type_cache_hasher::equal): Only compare
33137 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33138 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33139 non-aggregate element types.
33140 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33141 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33143 2017-04-27 Richard Biener <rguenther@suse.de>
33145 PR middle-end/80533
33146 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33147 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33148 keeping a reference to a trailing array.
33150 2017-04-27 Richard Biener <rguenther@suse.de>
33152 PR middle-end/80539
33153 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33154 being in loop-closed SSA form conservatively.
33155 (chrec_fold_multiply_poly_poly): Likewise.
33157 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33159 PR middle-end/79665
33160 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33161 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33163 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33166 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33167 (aarch64_function_arg_alignment): Return unsigned int again, but still
33168 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33169 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33170 Don't emit -Wpsabi note.
33171 (aarch64_function_arg_boundary): Likewise.
33172 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33175 2017-04-26 Nathan Sidwell <nathan@acm.org>
33177 * tree.h (crc32_unsigned_n): Declare.
33178 (crc32_unsigned, crc32_unsigned): Make inline.
33179 * tree.c (crc32_unsigned_bits): Replace with ...
33180 (crc32_unsigned_n): ... this.
33181 (crc32_unsigned, crc32_byte): Remove.
33182 (crc32_string): Remove unnecessary braces.
33184 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33186 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33187 * ipa-inline-analysis.c (MAX_TIME): Remove.
33188 (account_size_time): Use sreal for time.
33189 (dump_inline_summary): Update.
33190 (estimate_function_body_sizes): Update.
33191 (estimate_edge_size_and_time): Update.
33192 (estimate_calls_size_and_time): Update.
33193 (estimate_node_size_and_time): Update.
33194 (inline_merge_summary): Update.
33195 (inline_update_overall_summary): Update.
33196 (estimate_time_after_inlining): Update.
33197 (inline_read_section): Update.
33198 (inline_write_summary): Update.
33199 * ipa-inline.c (compute_uninlined_call_time): Update.
33200 (compute_inlined_call_time): Update.
33201 (recursive_inlining): Update.
33202 (inline_small_functions): Update.
33203 (dump_overall_stats): Update.
33204 * ipa-inline.h: Include sreal.h.
33205 (size_time_entry): Turn time to sreal.
33206 (inline_summary): Turn self_time nad time to sreal.
33208 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33210 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33212 (sreal::stream_out, sreal::stream_in): New.
33213 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33215 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33217 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33220 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33223 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33225 (dimode_scalar_chain::compute_convert_gain): Ditto.
33226 (dimode_scalar_chain::make_vector_copies): Ditto.
33227 (dimode_scalar_chain::convert_reg): Ditto.
33228 (dimode_scalar_chain::convert_insn): Ditto.
33229 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33230 (VI248_AVX512BW_1): New mode iterator.
33231 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33232 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33235 2017-04-25 Martin Sebor <msebor@redhat.com>
33237 PR tree-optimization/80497
33238 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33239 constants are representable in HOST_WIDE_INT.
33240 (parse_directive): Ditto.
33242 2017-04-25 Martin Sebor <msebor@redhat.com>
33245 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33246 (new_zero_array): Adjust signature.
33247 (dom_info::dom_init): Used unsigned rather that size_t.
33248 (dom_info::dom_info): Same.
33250 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33251 Jakub Jelinek <jakub@redhat.com>
33254 * config/arm/arm.c: Include gimple.h.
33255 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33256 returns negative, increment ncrn only if it returned positive.
33257 (arm_needs_doubleword_align): Return int instead of bool,
33258 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33259 members, but if there is any such non-FIELD_DECL
33260 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33261 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33262 returns negative, increment nregs only if it returned positive.
33263 (arm_setup_incoming_varargs): Likewise.
33264 (arm_function_arg_boundary): Emit -Wpsabi note if
33265 arm_needs_doubleword_align returns negative, return
33266 DOUBLEWORD_ALIGNMENT only if it returned positive.
33268 2017-04-25 Marek Polacek <polacek@redhat.com>
33271 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33272 first argument to type.
33274 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33277 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33278 type checks to test for compatibility instead of equality.
33280 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33281 Jakub Jelinek <jakub@redhat.com>
33284 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33286 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33287 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33288 the alignment computation, but return their maximum in warn_alignment.
33289 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33290 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33292 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33293 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33296 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33298 * config/arc/simdext.md (dmpyh): Fix typo.
33300 2017-04-25 Richard Biener <rguenther@suse.de>
33302 PR tree-optimization/80492
33303 * alias.c (compare_base_decls): Handle registers with asm
33304 specification conservatively.
33305 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33306 compare_base_decls returning dont-know properly.
33308 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33310 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33311 (legitimate_offset_address_p): New function.
33312 (arc_legitimate_address_p): Use above function.
33314 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33316 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33318 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33320 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33321 ACCH registers whenever they are available.
33323 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33325 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33326 double regs fix when not used.
33328 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33330 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33332 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33333 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33335 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33337 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33338 when emitting short ADD instructions.
33340 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33342 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33344 (cmpsi_cc_c_insn): Likewise.
33345 (cbranchsi4_scratch): Compute proper instruction length using
33346 compact_hreg_operand.
33347 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33349 2017-04-25 Richard Biener <rguenther@suse.de>
33351 PR middle-end/80509
33352 * passes.c (pass_manager::pass_manager): Initialize
33353 m_name_to_pass_map.
33355 2017-04-25 Richard Biener <rguenther@suse.de>
33357 PR tree-optimization/79201
33358 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33360 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33363 * config/s390/vector.md: Split MEM->GPR vector moves for
33364 non-s_operand addresses.
33366 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33369 * config/s390/predicates.md (reload_const_wide_int_operand): New
33371 * config/s390/s390.md ("movti"): Remove d/P alternative.
33372 ("movti_bigconst"): New pattern definition.
33374 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33377 * s390-protos.h (s390_expand_cs_hqi): Removed.
33378 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33379 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33380 modes as well as CCZ1mode and CCZmode.
33381 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33382 signature of s390_emit_compare_and_swap.
33383 (s390_expand_cs_hqi): Likewise, make static.
33384 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33385 compare-and-swap, in some cases.
33386 (s390_expand_cs): Wrapper function.
33387 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33389 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33390 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33391 patterns for small and large integers. Forbid symref memory operands.
33392 Move expander to s390.c. Require cc register.
33393 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33394 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33395 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33396 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33397 symref memory operands. Remove CC mode and call s390_match_ccmode
33399 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33401 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33403 * config/s390/s390.md (define_peephole2): New peephole to help
33404 combining the load-and-test pattern with volatile memory.
33406 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33408 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33409 with CCZmode for TARGET_Z196.
33411 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33413 PR rtl-optimization/80501
33414 * combine.c (make_compound_operation_int): Set subreg_code to SET
33415 even for AND with mask of the sign bit of mode.
33417 PR rtl-optimization/80500
33418 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33419 sum's initial value.
33421 2017-04-25 Julian Brown <julian@codesourcery.com>
33422 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33424 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33426 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33428 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33430 2017-04-25 Julian Brown <julian@codesourcery.com>
33431 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33433 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33434 (thunderx2t99_sha): New Reservation.
33436 2017-04-25 Julian Brown <julian@codesourcery.com>
33437 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33439 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33440 type for 1-element load.
33442 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33444 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33446 2017-04-24 Martin Jambor <mjambor@suse.cz>
33448 PR tree-optimization/80293
33449 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33450 char arrays not totally scalarizable if it is false.
33451 (analyze_all_variable_accesses): Pass correct value in the new
33452 parameter. Add a statistics counter.
33454 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33456 PR middle-end/79931
33457 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33459 2017-04-24 Richard Biener <rguenther@suse.de>
33461 PR tree-optimization/80494
33462 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33463 out for complex types.
33465 2017-04-24 Richard Biener <rguenther@suse.de>
33467 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33468 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33469 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33470 (DFS): Adjust and never fail.
33471 (sccvn_dom_walker::fail): Remove.
33472 (sccvn_dom_walker::before_dom_children): Adjust.
33473 (run_scc_vn): Likewise and never fail.
33474 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33475 (pass_fre::execute): Likewise.
33477 2017-04-24 Richard Biener <rguenther@suse.de>
33479 PR tree-optimization/79725
33480 * tree-ssa-sink.c (statement_sink_location): Return whether
33481 failure reason was zero uses. Move that check later.
33482 (sink_code_in_bb): Deal with zero uses by removing the stmt
33485 2017-04-24 Richard Biener <rguenther@suse.de>
33488 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33489 pointer-based references.
33491 2017-04-24 Richard Biener <rguenther@suse.de>
33494 * pass_manager.h (pass_manager::operator new): Remove.
33495 (pass_manager::operator delete): Likewise.
33496 * passes.c (pass_manager::operator new): Remove.
33497 (pass_manager::operator delete): Likewise.
33498 (pass_manager::pass_manager): Zero individual pass members.
33500 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33503 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33504 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33505 Check "XEXP (src, 1)" operand here.
33506 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33507 Check "XEXP (src, 1)" operand here.
33508 (dimode_scalar_chain::make_vector_copies): Detect count register
33509 of a shift instruction. Zero extend count register from QImode
33510 to DImode to satisfy vector shift pattern count operand predicate.
33511 Substitute vector shift count operand with a DImode copy.
33512 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33515 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33517 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33518 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33519 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33520 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33521 (UNSPEC_NOREX_MEM): Remove definition.
33523 2017-04-21 Richard Biener <rguenther@suse.de>
33525 PR tree-optimization/79547
33526 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33527 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33528 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33529 without any constraints.
33531 2017-04-21 Richard Biener <rguenther@suse.de>
33533 PR tree-optimization/78847
33534 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33536 2017-04-21 Richard Biener <rguenther@suse.de>
33538 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33539 (build_distinct_type_copy): Likewise.
33540 (build_variant_type_copy): Likewise.
33541 * tree.c (build_qualified_type): Pass down mem-stat info.
33542 (build_distinct_type_copy): Likewise.
33543 (build_variant_type_copy): Likewise.
33545 2017-04-21 Richard Biener <rguenther@suse.de>
33547 PR tree-optimization/80237
33548 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33550 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33551 for a simplified result.
33553 2016-04-21 Richard Biener <rguenther@suse.de>
33555 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33556 sth as strict as a simple_iv but a chrec without symbols and an
33557 operand defined in the loop we are peeling (and not some subloop).
33558 (propagate_constants_for_unrolling): Propagate all constants.
33560 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33563 * config/i386/i386.c (print_reg): Remove assert for disalowed
33564 regno values, call output_operand_lossage instead.
33566 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33569 * config/i386/constraints.md (Yc): New register constraint.
33570 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33571 Use Yc constraint for alternative 2 of operand 0. Remove
33572 preferred_for_speed attribute.
33574 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33576 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33577 lastprivate clauses in SIMT case.
33579 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33581 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33583 2017-04-20 Richard Biener <rguenther@suse.de>
33585 PR tree-optimization/57796
33586 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33587 as N scalar stores.
33588 (vect_model_load_cost): Cost gathers as N scalar loads.
33590 2017-04-20 Richard Biener <rguenther@suse.de>
33592 * ggc-page.c (ggc_allocated_p): Rename to ...
33593 (safe_lookup_page_table_entry): ... this and return the lookup
33595 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33597 2017-04-20 Richard Biener <rguenther@suse.de>
33599 PR tree-optimization/80453
33600 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33601 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33602 from the conditions.
33603 (vn_phi_eq): Pass them down.
33604 (vn_phi_lookup): Record them.
33605 (vn_phi_insert): Likewise.
33607 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33609 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33610 uninitialized variable warning to avoid buffer overrun.
33612 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33615 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33616 is suppressed for '{ 0 }' in C.
33618 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33620 * BASE-VER: Set to 8.0.0.
33622 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33624 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33625 priority .init_array and .fini_array section with SECTION_NOTYPE
33628 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33630 PR middle-end/80423
33631 * tree.h (build_array_type): Add typeless_storage default argument.
33632 * tree.c (type_cache_hasher::equal): Also compare
33633 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33634 (build_array_type): Add typeless_storage argument, set
33635 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33637 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33638 (build_array_type): Likewise. Add typeless_storage argument.
33640 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33641 Jakub Jelinek <jakub@redhat.com>
33643 PR tree-optimization/80426
33644 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33645 operation on symbolic operands, also compute the overflow for the
33646 invariant part when the operation degenerates into a negation.
33648 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33651 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33652 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33655 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33657 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33660 * config/avr/avr.c (tree.h): Include it.
33661 (cgraph.h): Include it.
33662 (avr_encode_section_info): Don't warn for uninitialized progmem
33663 variable if it's just an alias.
33665 2017-04-19 Richard Biener <rguenther@suse.de>
33668 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33669 when needed by AutoPGO.
33671 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33674 * doc/lto.texi: Remove an extra 'that'.
33676 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33678 PR rtl-optimization/80429
33679 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33680 are only used in debug insns.
33682 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33683 Vladimir Makarov <vmakarov@redhat.com>
33685 * config/sparc/predicates.md (input_operand): Add comment. Return
33686 true for any memory operand when LRA is in progress.
33687 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33689 2017-04-18 Jeff Law <law@redhat.com>
33692 * mips.md ({return,simple_return}_internal): Do not overwrite
33695 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33697 PR tree-optimization/80443
33698 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33699 instead of adding 1, subtract -1 and similarly instead of subtracting
33702 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33704 PR rtl-optimization/80357
33705 * haifa-sched.c (tmp_bitmap): New variable.
33706 (model_recompute): Handle duplicate use records.
33707 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33708 (free_global_sched_pressure_data): Free it.
33710 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33713 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33714 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33715 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33716 instead of SYSTEM_HEADER_DIR.
33718 2017-04-18 Jeff Law <law@redhat.com>
33720 PR middle-end/80422
33721 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33722 predecessors after walking up the insn chain.
33724 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33727 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33728 sizetype type into debug info.
33730 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33733 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33734 unneeded test for TARGET_UPPER_REGS_SF.
33735 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33737 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33740 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33741 instead of gsi_after_labels.
33743 2017-04-18 Jeff Law <law@redhat.com>
33745 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33749 2017-04-13 Jeff Law <law@redhat.com>
33750 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33751 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33753 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33756 * config/avr/avr.c (intl.h): Include it.
33757 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33759 2017-04-18 Martin Liska <mliska@suse.cz>
33761 PR gcov-profile/78783
33762 * gcov-tool.c (gcov_output_files): Validate that destination
33763 file is either removed by the tool or by a user.
33765 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33766 Guy Benyei <guybe@mellanox.com>
33768 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33769 block, and do not negate it, the stored id is already negative.
33771 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33773 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33775 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33778 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33779 masks of options that should be turned off if the VSX vector
33780 options are turned off.
33781 (OTHER_P8_VECTOR_MASKS): Likewise.
33782 (OTHER_VSX_VECTOR_MASKS): Likewise.
33783 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33784 rs6000_disable_incompatible_switches to validate no type switches
33786 (rs6000_incompatible_switch): New function to disallow turning on
33787 other vector options if -mno-vsx, -mno-power8-vector, or
33788 -mno-power9-vector are specified.
33790 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33792 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33794 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33796 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33797 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33798 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33799 (ARG_POINTER_CFA_OFFSET): Likewise.
33801 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33803 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33804 conditions to take advantage of various optimizations.
33806 2017-04-13 Jeff Law <law@redhat.com>
33808 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33809 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33810 (zero_extendsidi2_dext): Likewise.
33812 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33815 * fold-const.c (fold_ternary_loc): Revert
33816 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33819 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33821 PR rtl-optimization/80343
33822 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33823 new scratch pseudo.
33825 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33828 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33829 to ubsan_encode_value.
33831 2017-04-13 Jeff Law <law@redhat.com>
33833 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33834 appearing in DEBUG_INSNs.
33836 2017-04-13 Martin Liska <mliska@suse.cz>
33838 PR gcov-profile/80413
33839 * gcov-io.c (gcov_write_string): Copy to buffer just when
33840 allocated size is greater than zero.
33842 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33845 * dwarf2out.c (decls_for_scope): Ignore declarations of
33846 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33848 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33851 * ipa-visibility.c (non_local_p): Fix typos.
33852 (localize_node): When localizing symbol in same comdat group,
33853 dissolve the group only when we know external symbols are going
33855 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33857 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33859 PR tree-optimization/79390
33860 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33861 order does not result in usable sequence, retry with reversed operand
33867 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33868 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33869 op0 instead of fold_convert_loc (loc, type, arg0).
33871 2017-04-12 Jeff Law <law@redhat.com>
33873 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33874 has a delay slot in the generated code.
33876 * config/cris/cris.md (cris_preferred_reload_class): Return
33877 GENNONACR_REGS rather than GENERAL_REGS.
33879 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33882 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33883 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33884 signedness of the result type.
33886 2017-04-12 Richard Biener <rguenther@suse.de>
33887 Jeff Law <law@redhat.com>
33889 PR tree-optimization/80359
33890 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33891 trim stores to TARGET_MEM_REFs.
33893 2017-04-12 Richard Biener <rguenther@suse.de>
33895 PR tree-optimization/79390
33896 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33897 threading case even more.
33899 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33902 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33903 for quad_address_p for TImode, instead of just not indexed_address.
33905 2017-04-12 Richard Biener <rguenther@suse.de>
33906 Bernd Edlinger <bernd.edlinger@hotmail.de>
33908 PR middle-end/79671
33909 * alias.c (component_uses_parent_alias_set_from): Handle
33910 TYPE_TYPELESS_STORAGE.
33911 (get_alias_set): Likewise.
33912 * tree-core.h (tree_type_common): Add typeless_storage flag.
33913 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33914 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33915 for types containing members with TYPE_TYPELESS_STORAGE.
33916 (place_field): Likewise.
33917 (layout_type): Likewise for ARRAY_TYPE.
33918 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33919 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33920 TYPE_TYPELESS_STORAGE.
33921 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33923 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33926 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33927 first argument to type.
33929 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33933 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33934 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33935 (rs6000_expand_binop_builtin): Likewise.
33936 (rs6000_expand_ternop_builtin): Likewise; also add missing
33937 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33938 vshasigma built-ins.
33939 * doc/extend.texi: Document that vec_xxpermdi's third argument
33940 must be a constant.
33942 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33944 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33945 Use shift_const cost parameter when calculating gain of STV shifts.
33947 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33949 PR rtl-optimization/70478
33950 * lra-constraints.c (process_alt_operands): Check memory for
33951 disfavoring memory insn operand.
33953 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33955 PR middle-end/80100
33956 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33957 left shift in unsigned HOST_WIDE_INT type.
33959 PR rtl-optimization/80385
33960 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33961 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33964 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33965 if they have any depend clauses.
33967 2017-04-11 Martin Liska <mliska@suse.cz>
33970 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33971 * ipa-split.c (split_function): Create a local comdat symbol
33972 if caller is in a comdat group.
33974 2017-04-11 Martin Liska <mliska@suse.cz>
33977 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33980 2017-04-11 Martin Sebor <msebor@redhat.com>
33982 PR middle-end/80364
33983 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33984 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33986 (directive::set_width, directive::set_precision, format_character):
33988 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33991 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33994 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33995 conflict, set target->arch_name instead of target->cpu_name.
33997 2017-04-11 Richard Biener <rguenther@suse.de>
33999 PR tree-optimization/80374
34000 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34001 build_zero_cst, remove fold_convertible_p check again.
34003 2017-04-11 Martin Liska <mliska@suse.cz>
34006 * ubsan.c (instrument_object_size): Do not instrument register
34009 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34012 * config/i386/i386-builtin-types.def
34013 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34014 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34015 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34016 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34017 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34018 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34019 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34020 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34021 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34022 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34023 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34024 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34025 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34026 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34027 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34028 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34029 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34030 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34031 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34032 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34033 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34034 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34035 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34036 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34037 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34038 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34039 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34040 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34041 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34042 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34043 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34044 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34045 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34046 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34047 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34048 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34049 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34050 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34051 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34052 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34053 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34054 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34055 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34056 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34057 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34058 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34060 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34061 flag to second_arg_count, handle 4 argument function type _COUNT
34062 aliases, handle second_arg_count on second argument rather than last.
34064 2017-04-10 Jeff Law <law@redhat.com>
34066 PR tree-optimization/80374
34067 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34068 record anything if we can not convert integer_zero_node to the
34071 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34074 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34075 Enhance special handling given to the TARGET_P9_MINMAX option in
34076 relation to certain other options.
34078 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34080 PR tree-optimization/80153
34081 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34082 remove POINTER_PLUS_EXPR's base part directly, rather than through
34085 2017-04-10 Richard Biener <rguenther@suse.de>
34086 Bin Cheng <bin.cheng@arm.com>
34088 PR tree-optimization/80153
34089 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34090 the first element of pointer type aff_tree. Build result expr in
34092 (add_elt_to_tree): Convert to type unconditionally. Remove other
34093 fold_convert calls.
34094 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34095 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34097 2017-04-10 Richard Biener <rguenther@suse.de>
34099 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34102 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34104 PR rtl-optimization/70478
34105 * lra-constraints.c (curr_small_class_check): New.
34106 (update_and_check_small_class_inputs): New.
34107 (process_alt_operands): Update curr_small_class_check. Disfavor
34108 alternative insn memory operands. Check available regs for small
34111 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34114 * config/mips/mips.opt (-mvirt): Update description.
34115 * doc/invoke.texi (-mvirt): Likewise.
34117 2017-04-10 Richard Biener <rguenther@suse.de>
34119 PR middle-end/80362
34120 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34121 looking for NEGATE_EXPR in -A / -B to A / B folding.
34123 2017-04-10 Martin Liska <mliska@suse.cz>
34125 PR gcov-profile/80224
34126 * gcov.c (print_usage): Fix usage string.
34127 (get_gcov_intermediate_filename): Remove.
34128 (output_gcov_file): Use both for normal and intermediate format.
34129 (generate_results): Do not initialize special file for
34130 intermediate format.
34132 2017-04-10 Richard Biener <rguenther@suse.de>
34134 PR tree-optimization/80304
34135 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34138 2017-04-10 Nathan Sidwell <nathan@acm.org>
34141 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34142 (rs6000_init_builtins): Use it.
34144 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34146 * config/arm/arm.md (<mrc>): Add mode to SET source.
34147 (<mrrc>): Likewise.
34149 2017-04-10 Richard Biener <rguenther@suse.de>
34151 PR middle-end/80344
34152 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34154 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34157 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34158 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34159 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34160 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34161 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34162 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34163 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34164 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34165 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34166 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34167 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34168 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34169 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34170 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34171 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34172 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34173 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34174 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34175 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34176 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34177 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34178 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34179 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34181 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34183 PR rtl-optimization/70478
34184 * lra-constraints.c: Reverse the last patch.
34186 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34188 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34189 Add comment for WCHAR_T.
34191 2017-04-08 Martin Liska <mliska@suse.cz>
34194 2017-04-07 Martin Liska <mliska@suse.cz>
34197 * ipa-split.c (split_function): Add function part to a same comdat
34200 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34203 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34205 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34207 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34208 (vec_pairing): ...to this.
34209 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34210 (rs6000_sched_init): Adjust for name change.
34211 (struct rs6000_sched_context): Likewise.
34212 (rs6000_init_sched_context): Likewise.
34213 (rs6000_set_sched_context): Likewise.
34215 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34221 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34223 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34224 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34225 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34227 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34229 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34231 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34233 PR rtl-optimization/70703
34234 * ira-color.c (update_conflict_hard_regno_costs): Use
34235 int64_t instead of HOST_WIDE_INT.
34237 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34239 PR rtl-optimization/70478
34240 * lra-constraints.c (process_alt_operands): Disfavor alternative
34241 insn memory operands.
34243 2017-04-07 Jeff Law <law@redhat.com>
34245 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34246 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34248 2017-04-07 Martin Liska <mliska@suse.cz>
34251 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34252 Show error message instead of an ICE.
34254 2017-04-07 Martin Liska <mliska@suse.cz>
34257 * ipa-split.c (split_function): Add function part to a same comdat
34260 2017-04-07 Richard Biener <rguenther@suse.de>
34262 PR middle-end/80341
34263 * tree.c (get_unwidened): Also handle ! for_type case for
34265 * convert.c (do_narrow): Split out from ...
34266 (convert_to_integer_1): ... here. Do not pass final truncation
34267 type to get_unwidened for TRUNC_DIV_EXPR.
34269 2017-04-07 Richard Biener <rguenther@suse.de>
34271 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34273 (aff_combination_const): Adjust.
34274 (aff_combination_scale): Likewise.
34275 (aff_combination_add_elt): Likewise.
34276 (aff_combination_add_cst): Likewise.
34277 (aff_combination_convert): Likewise.
34278 (add_elt_to_tree): Likewise. Remove unused argument.
34279 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34281 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34283 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34285 * config/arm/arm.c (arm_default_short_enums): Use
34286 ARM_DEFAULT_SHORT_ENUMS.
34287 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34289 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34292 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34293 members with redundant out-of-class redeclaration.
34295 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34298 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34299 * config/i386/i386.md (*zero_extendsidi2):
34300 Add (?*x,*x) and (?*v,*v) alternatives.
34302 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34305 * config/i386/i386.c (ix86_expand_builtin)
34306 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34307 mode from insn data. Convert operands to insn operand mode.
34308 Copy operands that don't satisfy insn predicate to a register.
34310 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34312 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34315 2017-04-06 Richard Biener <rguenther@suse.de>
34317 PR tree-optimization/80334
34318 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34319 preserve alignment of accesses.
34321 2017-04-06 Richard Biener <rguenther@suse.de>
34323 PR tree-optimization/80262
34324 * tree-sra.c (build_ref_for_offset): Preserve address-space
34326 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34327 Drop useless address-space information on MEM_REF offsets.
34329 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34331 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34333 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34335 PR rtl-optimization/70703
34336 * ira-color.c (update_conflict_hard_regno_costs): Use
34337 HOST_WIDE_INT instead of long.
34339 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34342 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34343 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34345 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34346 for x86_64 target. Handle -m3dnowa option.
34348 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34350 PR rtl-optimization/70703
34351 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34352 (update_conflict_hard_regno_costs): Use long instead of unsigned
34353 arithmetic for cost calculation.
34355 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34356 Bernd Edlinger <bernd.edlinger@hotmail.de>
34359 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34362 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34365 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34366 ptr_mode with Pmode throughout.
34367 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34368 into probe_stack_range and use DImode.
34370 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34373 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34374 call_eh_return is true.
34376 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34378 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34379 Initialize last_match_fntype_index.
34381 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34384 * tree-nvr.c: Include internal-fn.h.
34385 (pass_return_slot::execute): Ignore internal calls without
34388 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34389 Richard Biener <rguenther@suse.de>
34392 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34393 captures used multiple times, except for the last use.
34394 * generic-match-head.c: Include gimplify.h.
34396 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34398 PR tree-optimization/79390
34399 * target.h (struct noce_if_info): Declare.
34400 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34401 * target.def (noce_conversion_profitable_p): New target hook.
34402 * ifcvt.h (struct noce_if_info): New type, moved from ...
34403 * ifcvt.c (struct noce_if_info): ... here.
34404 (noce_conversion_profitable_p): Renamed to ...
34405 (default_noce_conversion_profitable_p): ... this. No longer
34407 (noce_try_store_flag_constants, noce_try_addcc,
34408 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34409 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34410 instead of noce_conversion_profitable_p.
34411 * config/i386/i386.c: Include ifcvt.h.
34412 (ix86_option_override_internal): Don't override
34413 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34414 (ix86_noce_conversion_profitable_p): New function.
34415 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34416 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34417 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34418 * doc/tm.texi: Regenerated.
34420 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34422 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34425 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34428 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34429 instructions for small multiply cores.
34431 2017-04-04 Jeff Law <law@redhat.com>
34433 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34435 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34436 that are not set by the loop over the elements.
34438 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34441 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34442 int mode, convert_modes it to mode as unsigned, otherwise use
34443 lowpart_subreg to mode rather than SImode.
34444 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34445 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34446 Use DImode instead of SImode for the shift count operand.
34447 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34450 2017-04-04 Richard Biener <rguenther@suse.de>
34452 PR middle-end/80281
34453 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34454 arithmetic done for the negate or the plus. Simplify.
34455 (A - (-B) -> A + B): Likewise.
34456 * fold-const.c (split_tree): Make sure to not negate pointers.
34458 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34460 PR rtl-optimization/60818
34461 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34462 a compare of comparisons with the thing compared if this results
34463 in a different machine mode.
34465 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34467 * alias.c (base_alias_check): Fix typo in comment.
34468 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34469 * cgraphunit.c (symbol_table::compile): Likewise.
34470 * collect2.c (maybe_run_lto_and_relink): Likewise.
34471 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34472 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34473 * config/avr/avr.c (avr_map_op_t): Likewise.
34474 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34475 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34476 * config/epiphany/epiphany.md (movcc): Likewise.
34477 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34478 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34480 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34481 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34482 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34483 * config/sh/sh.c (sh_rtx_costs): Likewise.
34484 * fold-const.c (fold_truth_andor): Likewise.
34485 * genautomata.c (collapse_flag): Likewise.
34486 * gengtype.h (struct type::u::s): Likewise.
34487 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34488 * input.c (FORMAT_AMOUNT): Likewise.
34489 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34490 (known_aggs_to_agg_replacement_list): Likewise.
34491 * ipa-inline-analysis.c: Likewise.
34492 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34493 * ipa-polymorphic-call.c
34494 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34495 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34496 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34498 * modulo-sched.c (apply_reg_moves): Likewise.
34499 * omp-expand.c (build_omp_regions_1): Likewise.
34500 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34501 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34502 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34503 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34504 * value-prof.c: Likewise.
34505 * var-tracking.c (val_reset): Likewise.
34507 2017-04-03 Richard Biener <rguenther@suse.de>
34509 PR tree-optimization/80275
34510 * fold-const.c (split_address_to_core_and_offset): Handle
34513 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34515 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34516 descriptors is at least equal to that of functions.
34518 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34520 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34522 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34525 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34526 (mov<IMOD4:mode>): New expander.
34527 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34529 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34531 PR rtl-optimization/79405
34532 * fwprop.c (propagations_left): New variable.
34533 (forward_propagate_into): Decrement it.
34534 (fwprop_init): Initialize it.
34535 (fw_prop): If the variable has reached zero, stop propagating.
34536 (fwprop_addr): Ditto.
34538 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34541 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34542 a FUNCTION_DECL, pass it as decl instead of origin to
34545 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34547 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34550 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34553 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34554 TARGET_VSX_SMALL_INTEGER.
34556 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34558 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34559 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34561 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34563 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34564 extraction from odd-numbered MSA register.
34566 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34568 PR middle-end/80173
34569 * expmed.c (store_bit_field_1): Don't attempt to create
34570 a word subreg out of hard registers wider than word if they
34571 have HARD_REGNO_NREGS of 1 for their mode.
34573 PR middle-end/80163
34574 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34575 conversions to integer types wider than word and pointer.
34578 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34579 (rtx_equal_for_cselib_p): Pass 0 to it.
34580 * cselib.c (cselib_hasher::equal): Likewise.
34581 (rtx_equal_for_cselib_1): Add depth argument. If depth
34582 is 128, don't look up VALUE locs and punt. Increment
34583 depth in recursive calls when walking VALUE locs.
34585 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34587 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34588 (make_gcov_file_name): Use the canonical path name for generating
34590 (read_line): Fix handling of files with ascii null bytes.
34592 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34594 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34595 to initialise a vector register instead
34596 of using a const_int.
34598 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34600 PR translation/80189
34601 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34602 diagnostic messages.
34604 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34607 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34608 (dfp_diex_<mode>): Update mode of operand 1.
34609 * doc/extend.texi (dxex, dxexq): Document change to return type.
34610 (diex, diexq): Document change to argument type.
34612 2017-03-30 Martin Jambor <mjambor@suse.cz>
34615 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34616 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34617 it reflects the signature changes performed at the callee side.
34618 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34619 to cgraph_build_function_type_skip_args.
34620 (build_function_decl_skip_args): Adjust call to the above function.
34622 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34625 * config/i386/sse.md
34626 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34627 register as dest whenever it is a MEM not rtx_equal_p to the
34628 corresponding dup operand, and when forcing into reg move the
34629 reg into the memory afterwards.
34630 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34631 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34632 for the force_reg mode.
34633 (avx512vl_vextractf128<mode>): Use register as dest either
34634 always when a MEM, or when it is a MEM not rtx_equal_p to the
34635 corresponding dup operand, or even not when it is a CONST_VECTOR
34636 depending on the mode and lo vs. hi.
34637 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34639 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34640 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34641 Likewise. Require that operands[2] is even.
34642 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34643 Remove extraneous parens. Require that operands[2] is a multiple
34645 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34646 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34647 disallow memory then.
34649 2017-03-30 Richard Biener <rguenther@suse.de>
34651 PR tree-optimization/77498
34652 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34653 to non-constants over backedges.
34655 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34657 PR rtl-optimization/80233
34658 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34659 as last_combined_insn. Do not test for BARRIER_P separately.
34661 2017-03-29 Andreas Schwab <schwab@suse.de>
34664 * calls.c (prepare_call_address): Convert funexp to Pmode before
34665 copying to temp reg.
34667 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34669 PR tree-optimization/80158
34670 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34671 Handle possible future case of more than one alternate
34673 (replace_rhs_if_not_dup): Likewise.
34674 (replace_one_candidate): Likewise.
34676 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34678 PR rtl-optimization/80193
34679 * ira.c (ira): Do not check allocation for LRA.
34681 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34683 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34684 (nvptx_output_simt_exit): Declare.
34685 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34686 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34687 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34688 (nvptx_declare_function_name): ...here. Emit declaration of local
34689 memory space buffer for omp_simt_enter insn.
34690 (nvptx_output_unisimt_switch): New.
34691 (nvptx_output_softstack_switch): New.
34692 (nvptx_output_simt_enter): New.
34693 (nvptx_output_simt_exit): New.
34694 * config/nvptx/nvptx.h (struct machine_function): New fields
34695 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34696 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34697 (UNSPECV_SIMT_EXIT): Ditto.
34698 (omp_simt_enter_insn): New insn.
34699 (omp_simt_enter): New expansion.
34700 (omp_simt_exit): New insn.
34701 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34703 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34704 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34705 (expand_GOMP_SIMT_EXIT): New.
34706 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34707 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34708 (GOMP_SIMT_EXIT): Ditto.
34709 * target-insns.def (omp_simt_enter): New insn.
34710 (omp_simt_exit): Ditto.
34711 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34713 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34714 (lower_rec_input_clauses): Likewise.
34715 (lower_lastprivate_clauses): Handle SIMT privatization.
34717 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34718 (ompdevlow_adjust_simt_enter): New.
34719 (find_simtpriv_var_op): New.
34720 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34721 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34723 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34724 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34725 (copy_decl_for_dup_finish): Ditto.
34727 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34729 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34732 * config/i386/i386.c (ix86_option_override_internal): Always
34733 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34735 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34737 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34739 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34741 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34742 mark new edge's irreducible flag accordign to it.
34743 (vect_do_peeling): Check loop preheader edge's irreducible flag
34744 and pass it to function slpeel_add_loop_guard.
34746 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34748 PR tree-optimization/80218
34749 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34750 Update block frequencies and counts.
34752 2017-03-28 Richard Biener <rguenther@suse.de>
34754 PR tree-optimization/78644
34755 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34756 of a simplification result we may not use it at all.
34758 2017-03-28 Richard Biener <rguenther@suse.de>
34761 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34762 without arguments, generate default definition of a SSA name.
34764 2017-03-28 Richard Biener <rguenther@suse.de>
34766 PR middle-end/80222
34767 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34768 TYPE_REF_CAN_ALIAS_ALL references.
34769 * fold-const.c (fold_indirect_ref_1): Likewise.
34771 2017-03-28 Martin Liska <mliska@suse.cz>
34774 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34775 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34777 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34778 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34780 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34781 (EXTRA_SPECS): Define.
34782 (SUBTARGET_EXTRA_SPECS): Likewise.
34783 (SUBTARGET_CPP_SPEC): Likewise.
34784 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34785 SUBTARGET_EXTRA_SPECS.
34786 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34788 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34790 * config/arc/simdext.md (vst64_insn): Update pattern.
34791 (vld32wh_insn): Likewise.
34792 (vld32wl_insn): Likewise.
34793 (vld64_insn): Likewise.
34794 (vld32_insn): Likewise.
34796 2017-03-28 Marek Polacek <polacek@redhat.com>
34799 * fold-const.c (fold_comparison): Use protected_set_expr_location
34800 instead of SET_EXPR_LOCATION.
34802 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34804 * tree.c (add_expr): Avoid name lookup warning.
34806 2017-03-27 Jeff Law <law@redhat.com>
34808 PR tree-optimization/80216
34809 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34810 function name. Limit recursion depth.
34811 (record_temporary_equivalences): Corresponding changes.
34813 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34815 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34818 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34821 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34823 * cfgcleanup.c (reg_note_cfa_p): New array.
34824 (insns_have_identical_cfa_notes): New function.
34825 (old_insns_match_p): Don't cross-jump in between /f
34826 and non-/f instructions. If both i1 and i2 are frame related,
34827 verify all CFA notes, their order and content.
34829 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34832 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34833 HImode and SImode with zero extend to DImode to one insn.
34834 (bswap<mode>2_extenddi): Likewise.
34835 (bswapsi2_extenddi): Likewise.
34836 (bswaphi2_extendsi): Likewise.
34837 (bswaphi2): Combine bswap HImode and SImode into one insn.
34838 Separate memory insns from swapping register.
34839 (bswapsi2): Likewise.
34840 (bswap<mode>2): Likewise.
34841 (bswaphi2_internal): Delete, no longer used.
34842 (bswapsi2_internal): Likewise.
34843 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34844 store, and gpr<-gpr swap insns.
34845 (bswap<mode>2_store): Likewise.
34846 (bswaphi2_reg): Register only splitter, combine with the splitter.
34847 (bswaphi2 splitter): Likewise.
34848 (bswapsi2_reg): Likewise.
34849 (bswapsi2 splitter): Likewise.
34850 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34851 the insns into load, store, and register/register insns.
34852 (bswapdi2_ldbrx): Likewise.
34853 (bswapdi2_load): Likewise.
34854 (bswapdi2_store): Likewise.
34855 (bswapdi2_reg): Likewise.
34857 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34859 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34860 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34862 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34865 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34867 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34868 special handling for target option conflicts between dform
34869 options (-mpower9-dform, -mpower9-dform-vector,
34870 -mpower9-dform-scalar) and -mno-direct-move.
34872 2017-03-27 Richard Biener <rguenther@suse.de>
34874 PR tree-optimization/80181
34875 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34877 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34879 * config/arc/predicates.md (move_double_src_operand): Replace the
34880 call to move_double_src_operand with a call to address_operand.
34882 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34884 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34885 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34886 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34888 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34890 * config/arc/predicates.md (long_immediate_loadstore_operand):
34891 Consider scaled addresses cases.
34893 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34895 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34896 restored when in interrupt.
34897 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34898 doesn't have delay slot.
34900 2017-03-27 Richard Biener <rguenther@suse.de>
34903 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34904 inlined thunk clones.
34906 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34909 * asan.c (instrument_derefs): Copy over last operand from
34910 original COMPONENT_REF to the new COMPONENT_REF with
34911 DECL_BIT_FIELD_REPRESENTATIVE.
34912 * ubsan.c (instrument_object_size): Likewise.
34914 2017-03-27 Richard Biener <rguenther@suse.de>
34916 PR tree-optimization/80170
34917 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34918 sure DR/SCEV didnt fold in constants we do not see when looking
34919 at the reference base alignment.
34921 2017-03-27 Richard Biener <rguenther@suse.de>
34923 PR middle-end/80171
34924 * gimple-fold.c (fold_ctor_reference): Properly guard against
34925 NULL return value from canonicalize_constructor_val.
34927 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34930 * config/i386/i386.c (ix86_expand_builtin)
34931 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34932 flags reg setting and flags reg using instructions.
34933 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34934 clobbering instructions to zero extend op2.
34936 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34938 * doc/install.texi (Configuration) <--with-aix-soname>:
34939 Update link to AIX ld.
34941 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34943 PR rtl-optimization/80160
34944 PR rtl-optimization/80159
34945 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34946 reg_alternate_class into account.
34948 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34951 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34952 to consider in curr_insn_transform.
34954 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34956 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34957 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34958 and emit_mode_inner.
34960 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34962 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34963 argument to the overloaded builtin variants. Use the new flag to
34964 deprecate certain builtin variants.
34965 * config/s390/s390-builtin-types.def: Add new builtin types.
34966 * config/s390/s390-builtins.h: Support new flags field for
34967 overloaded builtins.
34968 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34969 (s390_macro_to_expand): Enable vector float data type.
34970 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34971 builtins by incrementing the __VEC__ version number.
34972 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34974 (s390_resolve_overloaded_builtin): Emit error messages depending
34975 on the builtin flags.
34976 * config/s390/s390.c (s390_expand_builtin): Support additional
34977 flags argument. Change error message to match the messages
34978 emitted in s390-c.c.
34979 * config/s390/s390.md: New UNSPEC_* constants.
34980 (op_type): Add new instruction types.
34981 * config/s390/vecintrin.h: Add new builtins and test data class
34983 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34984 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34985 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34986 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34987 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34988 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34990 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34991 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34992 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34993 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34995 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34996 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34997 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34998 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35000 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35001 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35002 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35003 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35004 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35005 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35006 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35008 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35009 ("vec_scatter_element<V_HW_4:mode>_DI")
35010 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35011 ("vec_fpint<mode>", "vflls")
35012 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35013 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35014 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35015 ("*vec_cmphe<mode>_cc"): ... these.
35017 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35018 mode constant instead of magic value.
35020 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35022 * config/s390/s390.c (s390_expand_vec_compare): Support other
35023 vector floating point modes than just V2DF.
35024 (s390_expand_vcond): Likewise.
35025 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35026 (s390_cannot_change_mode_class): Prevent mode changes between TF
35027 and V1TF in vector registers.
35028 * config/s390/s390.md (DF, SF): New mode attributes.
35029 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35030 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35031 SFmode support for VRs.
35032 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35034 (VFT, VF_HW): New mode iterators.
35035 (vw, sdx): New mode attributes.
35036 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35037 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35038 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35039 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35040 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35041 also the new vector floating point modes. Renaming to ...
35043 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35044 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35045 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35046 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35047 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35048 ("vec_unordered<mode>"): ... these.
35050 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35051 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35052 ("*vec_extendv2df"): New insn definitions.
35054 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35056 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35057 ("mulditi3_2", "*muldi3_sign"): New patterns.
35058 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35059 rename the pattern definition.
35061 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35063 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35065 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35067 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35069 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35070 instruction if possible.
35071 * config/s390/vector.md (vec_halfnumelts): New mode
35073 ("*vec_vllezlf<mode>"): New pattern.
35075 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35077 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35078 ("popcountv4si2", "popcountv2di2"): Rename to ...
35079 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35080 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35082 ("popcount<mode>2_vxe"): New pattern.
35084 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35086 * common/config/s390/s390-common.c (processor_flags_table): Add
35088 * config.gcc: Add arch12.
35089 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35090 Default to arch12 for unknown CPU model numbers.
35091 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35092 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35093 PROCESSOR_max sanity check.
35094 * config/s390/s390-opts.h (enum processor_type): Add
35096 * config/s390/s390.c (processor_table): Add arch12.
35097 (s390_expand_builtin): Add check for B_VXE flag.
35098 (s390_issue_rate): Add PROCESSOR_ARCH12.
35099 (s390_get_sched_attrmask): Likewise.
35100 (s390_get_unit_mask): Likewise.
35101 (s390_sched_score): Enable z13 scheduling for arch12.
35102 (s390_sched_reorder): Likewise.
35103 (s390_sched_variable_issue): Likewise.
35104 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35106 (s390_tune_attr): Use z13 scheduling also for arch12.
35107 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35108 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35109 (TARGET_VXE_P): New macros.
35110 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35111 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35112 * config/s390/s390.opt: Add arch12 as processor_type.
35114 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35116 * config/s390/s390.md
35117 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35118 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35119 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35121 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35122 Rename expanders to ...
35124 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35125 ("fixuns_truncdddi2_emu"): ... these.
35127 ("fixuns_trunc<mode>si2_emu"): New expander.
35129 ("*fixuns_truncdfdi2_z13"): Rename to ...
35130 ("*fixuns_truncdfdi2_vx"): ... this.
35132 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35134 * config/s390/2964.md: Remove the single element vector compare
35135 instructions which are no longer used.
35136 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35138 (s390_canonicalize_comparison): Remove handling of DFmode
35140 (s390_expand_vec_compare_scalar): Remove function.
35141 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35142 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35144 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35146 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35148 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35150 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35151 will anyway by matched by mov<mode>_64dfp.
35153 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35155 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35156 vlef/vstef. Add missing operand to vleif.
35158 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35160 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35161 pair for all vector types with 64 bit elements.
35162 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35163 * config/s390/vector.md (V_HW_64): ... here.
35164 (V_128_NOSINGLE): New mode iterator.
35165 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35166 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35167 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35168 ("*vec_load_pairv2di"): Change to ...
35169 ("*vec_load_pair<mode>"): ... this one.
35171 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35173 * config/s390/constraints.md: Add comments.
35174 (jKK): Reject element sizes > 8 bytes.
35175 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35177 * config/s390/s390.md: Add the s_operand checks formerly in
35178 s390_split_ok_p to various splitters where they are still
35180 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35181 for 128 bit vectors. Plus two splitters.
35183 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35185 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35188 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35191 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35192 error if the boundary argument is not constant.
35194 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35196 PR rtl-optimization/80112
35197 * loop-doloop.c (doloop_condition_get): Don't check condition
35198 if cmp isn't SET with IF_THEN_ELSE src.
35200 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35202 PR tree-optimization/80158
35203 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35204 replacing a candidate statement, also replace it for the
35205 candidate's alternate interpretation.
35206 (replace_rhs_if_not_dup): Likewise.
35207 (replace_one_candidate): Likewise.
35209 2017-03-24 Richard Biener <rguenther@suse.de>
35211 PR tree-optimization/80167
35212 * graphite-isl-ast-to-gimple.c
35213 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35215 (translate_isl_ast_to_gimple::get_rename): Likewise.
35217 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35219 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35220 handling of certain combinations of target options, including the
35221 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35222 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35224 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35227 * config/arm/arm.md (*load_multiple): Add reload_completed to
35228 matching condition.
35230 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35231 Richard Biener <rguenth@suse.de>
35233 PR tree-optimization/79908
35234 PR tree-optimization/80136
35235 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35236 been cast away, gimplify_and_add suffices.
35238 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35240 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35242 2017-03-23 Richard Biener <rguenther@suse.de>
35244 PR tree-optimization/80032
35245 * gimplify.c (gimple_push_cleanup): Forced unconditional
35246 cleanups still have to go to the conditional_cleanups
35249 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35251 PR tree-optimization/80072
35252 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35254 (next_operand_entry_id): Change type to unsigned int.
35255 (sort_by_operand_rank): Make sure to return the right return value
35256 even if unsigned fields are bigger than INT_MAX.
35257 (struct oecount): Change cnt and id type to unsigned int.
35258 (oecount_hasher::equal): Formatting fix.
35259 (oecount_cmp): Make sure to return the right return value
35260 even if unsigned fields are bigger than INT_MAX.
35261 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35264 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35265 TREE_READONLY on result if writing it more than once.
35268 * doc/invoke.texi (-fsanitize=thread): Document that with
35269 -fnon-call-exceptions atomics are not able to throw
35273 * tsan.c: Include tree-eh.h.
35274 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35275 maybe_clean_or_replace_eh_stmt where needed.
35276 (instrument_memory_accesses): Add cfg_changed argument.
35277 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35278 if it returned true.
35279 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35281 PR rtl-optimization/63191
35282 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35283 wrapper function, moved the whole old content into ...
35284 (ix86_delegitimize_address_1): ... this. New inline function.
35285 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35286 true as last argument instead of ix86_delegitimize_address.
35288 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35290 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35291 cortexa57_branch_cost.
35293 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35295 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35297 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35300 * doc/md.texi (Constraints): Document wA constraint.
35301 * config/rs6000/constraints.md (wA): New.
35302 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35303 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35304 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35305 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35307 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35310 * gimplify.c (is_oacc_declared): New function.
35311 (oacc_default_clause): Use it to set default flags for acc declared
35312 variables inside parallel regions.
35313 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35314 declared variables.
35315 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35316 declare attribute to any decl as necessary.
35318 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35321 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35322 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35323 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35324 (arm_arch_lpae): This.
35325 * config/arm/arm.c (arm_arch7ve): Rename into ...
35326 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35327 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35330 2017-03-22 Martin Liska <mliska@suse.cz>
35333 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35334 error message instead of an ICE.
35336 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35338 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35340 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35342 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35344 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35347 2017-03-21 Martin Sebor <msebor@redhat.com>
35349 * doc/extend.texi: Use "cannot" instead of "can't."
35350 * doc/hostconfig.texi: Same.
35351 * doc/install.texi: Same.
35352 * doc/invoke.texi: Same.
35353 * doc/loop.texi: Same.
35354 * doc/md.texi: Same.
35355 * doc/objc.texi: Same.
35356 * doc/rtl.texi: Same.
35357 * doc/tm.texi: Same.
35358 * doc/tm.texi.in: Same.
35359 * doc/trouble.texi: Same.
35361 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35364 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35365 (collect_checksum_attributes): Set it.
35366 (die_checksum_ordered): Use it.
35368 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35370 PR tree-optimization/79908
35371 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35372 change: For a VA_ARG whose LHS has been cast away, use
35373 force_gimple_operand to construct the side effects.
35375 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35377 PR translation/80001
35378 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35379 more amenable to translation.
35380 (oacc_loop_auto_partitions): Likewise.
35382 2017-03-21 Marek Polacek <polacek@redhat.com>
35383 Martin Sebor <msebor@redhat.com>
35385 PR tree-optimization/80109
35386 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35387 on INTEGRAL_TYPE_P.
35389 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35390 Segher Boessenkool <segher@kernel.crashing.org>
35393 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35394 check reg_used_between_p between insn and one of succ or succ2
35395 depending on if succ is artificial insn not inserted into insn
35398 2017-03-21 Martin Liska <mliska@suse.cz>
35400 PR gcov-profile/80081
35401 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35402 * doc/gcc.texi: Include gcov-dump stuff.
35403 * doc/gcov-dump.texi: New file.
35405 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35407 PR rtl-optimization/79150
35408 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35409 conditional jump, if the jump is the last insn of the loop.
35411 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35412 Richard Biener <rguenth@suse.de>
35414 PR tree-optimization/79908
35415 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35416 been cast away, use force_gimple_operand to construct the side
35419 2017-03-21 Martin Liska <mliska@suse.cz>
35421 PR libfortran/79956
35422 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35425 2017-03-21 Brad Spengler <spender@grsecurity.net>
35428 * plugin.c (htab_hash_plugin): New function.
35429 (add_new_plugin): Use it and adjust.
35430 (parse_plugin_arg_opt): Adjust.
35431 (init_one_plugin): Likewise.
35433 2017-03-21 Richard Biener <rguenther@suse.de>
35435 PR tree-optimization/80032
35436 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35437 if set force the cleanup to happen unconditionally.
35438 (gimplify_target_expr): Push inserted clobbers with force_uncond
35439 to avoid them being removed by control-dependent DCE.
35441 2017-03-21 Richard Biener <rguenther@suse.de>
35443 PR tree-optimization/80122
35444 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35445 va_arg_pack_len when the inlined call stmt requires pack
35447 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35449 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35452 * tsan.c (instrument_builtin_call): If the memory model argument
35453 is not a constant, assume it is valid.
35456 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35459 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35461 PR rtl-optimization/79910
35462 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35463 if its dest is used by an insn before I2 (other than the combined
35464 insns themselves, which are properly handled already).
35466 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35469 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35471 * combine.c (record_used_regs): New static function.
35472 (try_combine): Handle situations where there is an additional
35473 instruction between I2 and I3 which needs to have a LOG_LINK
35477 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35479 * combine.c (try_combine): Delete redundant i1 test. Call
35480 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35482 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35485 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35486 alternatives 13/14.
35488 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35490 PR tree-optimization/80054
35491 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35492 the optimization if a PHI or any of its arguments is not dominated
35493 by the candidate's basis. Use gphi* rather than gimple* as
35495 (replace_profitable_candidates): Clean up a gimple* variable that
35496 should be a gphi* variable.
35498 2017-03-20 Martin Sebor <msebor@redhat.com>
35501 * doc/extend.texi (attribute constructor): Document present limitation.
35503 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35506 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35507 __POWER9_VECTOR__ #ifdef control, change template definition to
35508 use Power9-specific built-in function.
35509 (vec_any_eq): Likewise.
35510 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35511 to control outcomes from this test.
35512 (vector_ae_<mode>p): For VEC_F modes, likewise.
35514 2017-03-20 Ian Lance Taylor <iant@google.com>
35516 * config/i386/i386.c (ix86_function_regparm): Save an extra
35517 register for -fsplit-stack with DECL_STATIC_CHAIN.
35519 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35522 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35523 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35525 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35527 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35529 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35532 2017-03-20 Marek Polacek <polacek@redhat.com>
35535 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35537 2017-03-20 Richard Biener <rguenther@suse.de>
35539 PR tree-optimization/80113
35540 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35541 allocate extra SSA name for PHI def.
35542 (add_close_phis_to_outer_loops): Likewise.
35543 (add_close_phis_to_merge_points): Likewise.
35544 (copy_loop_close_phi_args): Likewise.
35545 (copy_cond_phi_nodes): Likewise.
35547 2017-03-20 Martin Liska <mliska@suse.cz>
35549 PR middle-end/79753
35550 * tree-chkp.c (chkp_build_returned_bound): Do not build
35551 returned bounds for a LHS that's not a BOUNDED_P type.
35553 2017-03-20 Martin Liska <mliska@suse.cz>
35557 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35558 COMPLEX_CST and VECTOR_CST.
35560 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35563 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35564 target operand. A new splitter adds the clobber statement in case
35565 the target operand is dead anyway.
35567 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35569 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35570 to age-old versions of binutils and glibc.
35572 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35574 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35576 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35578 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35580 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35582 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35583 requirement for binutils 2.13.
35585 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35587 * combine.c (try_combine): Delete redundant i1 test. Call
35588 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35590 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35592 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35593 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35595 <riscv64-*-elf>: Re-arrange section
35596 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35597 <riscv32-*-linux>: Likewise.
35598 <riscv64-*-elf>: Likewise
35599 <riscv64-*-linux>: Likewise.
35601 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35604 * aarch64.opt(verbose-cost-dump): Fix typo.
35606 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35609 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35610 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35612 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35614 * reload.c (find_reloads): When reloading a nonoffsettable address,
35615 use RELOAD_OTHER for it and its address reloads.
35617 PR rtl-optimization/79910
35618 * combine.c (record_used_regs): New static function.
35619 (try_combine): Handle situations where there is an additional
35620 instruction between I2 and I3 which needs to have a LOG_LINK
35623 2017-03-17 Jeff Law <law@redhat.com>
35625 PR tree-optimization/71437
35626 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35627 conditional in the hash table first.
35628 (vrp_dom_walker::before_dom_children): Extract condition from
35629 ASSERT_EXPR. Record condition, its inverion and any implied
35630 conditions as well.
35632 2017-03-17 Marek Polacek <polacek@redhat.com>
35633 Markus Trippelsdorf <markus@trippelsdorf.de>
35635 PR tree-optimization/80079
35636 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35639 2017-03-17 Richard Biener <rguenther@suse.de>
35641 PR middle-end/80075
35642 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35643 Properly verify the LHS before the RHS possibly claims to be
35645 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35648 2017-03-17 Martin Jambor <mjambor@suse.cz>
35650 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35651 (List of -O2 options): Likewise.
35652 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35655 2017-03-17 Tom de Vries <tom@codesourcery.com>
35657 * gcov-dump.c (print_usage): Print bug_report_url.
35659 2017-03-17 Richard Biener <rguenther@suse.de>
35661 PR middle-end/80050
35662 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35663 (parser::peek): Likewise.
35665 2017-03-17 Richard Biener <rguenther@suse.de>
35667 PR tree-optimization/80048
35668 * sese.c (free_sese_info): Properly release rename_map and
35669 copied_bb_map elements.
35671 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35673 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35674 Add linked-list forward and backlinks. Insert on
35675 construction, remove on destruction.
35676 (class pass_store_merging): Add m_stores_head field.
35677 (pass_store_merging::terminate_and_process_all_chains):
35678 Iterate over m_stores_head list.
35679 (pass_store_merging::terminate_all_aliasing_chains):
35681 (pass_store_merging::execute): Check for debug stmts first.
35682 Push new chains onto the m_stores_head stack.
35684 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35687 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35688 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35689 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35691 2017-03-16 Jeff Law <law@redhat.com>
35693 PR tree-optimization/71437
35694 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35695 member function. Implementation moved into after_dom_children
35696 member function and into the threader's thread_outgoing_edges
35698 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35699 some code into new thread_outgoing_edges.
35700 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35701 definition. Simplify marker handling (do it here). Assume we always
35702 have the available expression and the const/copies tables.
35703 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35705 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35706 * tree-vrp.c (equiv_stack): No longer file scoped.
35707 (vrp_dom_walker): New class.
35708 (vrp_dom_walker::before_dom_children): New member function.
35709 (vrp_dom_walker::after_dom_children): Likewise.
35710 (identify_jump_threads): Setup domwalker. Use it rather than
35711 walking edges in a random order by hand. Simplify setup/finalization.
35712 (finalize_jump_threads): Remove.
35713 (vrp_finalize): Do not call identify_jump_threads here.
35714 (execute_vrp): Do it here instead and call thread_through_all_blocks
35717 PR tree-optimization/71437
35718 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35720 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35722 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35723 (dom_opt_dom_walker::thread_across_edge): Remove
35724 handle_dominating_asserts argument. All callers changed.
35725 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35726 changes. Remove calls to lhs_of_dominating_assert. Other
35727 uses of handle_dominating_asserts turn into unconditional code
35728 (simplify_control_stmt_condition_1): Likewise.
35729 (simplify_control_stmt_condition): Likewise.
35730 (thread_through_normal_block, thread_across_edge): Likewise.
35731 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35732 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35733 object if it is not an SSA_NAME.
35734 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35735 before calling into the VRP specific simplifiers.
35736 (identify_jump_threads): Remove handle_dominating_asserts
35739 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35742 * tree-diagnostic.c (default_tree_printer): No longer static.
35743 * tree-diagnostic.h (default_tree_printer): New prototype.
35745 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35747 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35748 Change ins into fmov.
35750 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35752 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35753 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35754 Use h_con constraint for operand 1.
35755 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35756 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35758 2017-03-15 Jeff Law <law@redhat.com>
35760 PR tree-optimization/71437
35761 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35762 (record_temporary_equivalences): Use it.
35764 PR tree-optimization/71437
35765 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35766 tree-ssa-scopedtables.
35767 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35768 (record_conditions, record_cond, vuse_eq): Likewise.
35769 (record_edge_info): Adjust to API tweak of record_conditions.
35770 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35771 (record_temporary_equivalences, optimize_stmt): Likewise.
35772 (eliminate_redundant_computations): Likewise.
35773 (record_equivalences_from_stmt): Likewise.
35774 * tree-ssa-scopedtables.c: Include options.h and params.h.
35775 (vuse_eq): New function, moved from tree-ssa-dom.c
35776 (build_and_record_new_cond): Likewise.
35777 (record_conditions): Likewise. Accept vector of conditions rather
35778 than edge_equivalence structure for first argument.
35779 for the first argument.
35780 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35781 from tree-ssa-dom.c.
35782 (avail_exprs_stack::record_cond): Likewise.
35783 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35784 from tree-ssa-dom.c.
35785 (avail_exprs_stack): Add new member functions lookup_avail_expr
35787 (record_conditions): Declare.
35789 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35792 * lra-constraints.c (process_alt_operands): Increase reject for
35793 reloading an input/output operand.
35795 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35798 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35799 insns to convert from signed/unsigned char/short to IEEE 128-bit
35801 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35803 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35806 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35807 subreg of inner mode for values already in registers.
35809 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35811 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35812 iteration reg is used after the loop.
35814 2017-03-14 Martin Sebor <msebor@redhat.com>
35816 PR tree-optimization/79800
35817 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35818 precision in negative-positive range.
35819 (format_floating): Call non-const overload with adjusted precision.
35821 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35824 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35827 2017-03-14 Martin Sebor <msebor@redhat.com>
35829 PR middle-end/80020
35830 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35831 * builtins.def (aligned_alloc): Use it.
35834 * Makefile.in (GTFILES): Add calls.c.
35835 * calls.c: Include "gt-calls.h".
35837 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35839 PR rtl-optimization/79728
35840 * regs.h (struct target_regs): New field
35841 x_contains_allocatable_regs_of_mode.
35842 (contains_allocatable_regs_of_mode): New macro.
35843 * reginfo.c (init_reg_sets_1): Initialize it, and change
35844 contains_reg_of_mode so it includes global regs as well.
35845 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35846 rather than contains_regs_of_mode.
35848 2017-03-14 Martin Liska <mliska@suse.cz>
35850 * doc/invoke.texi: Document options that can't be combined with
35851 -fcheck-pointer-bounds.
35853 2017-03-14 Martin Liska <mliska@suse.cz>
35855 PR middle-end/79831
35856 * doc/invoke.texi (-Wchkp): Document the option.
35858 2017-03-14 Martin Liska <mliska@suse.cz>
35860 * Makefile.in: Install gcov-dump.
35862 2017-03-14 Martin Liska <mliska@suse.cz>
35864 * multiple_target.c (expand_target_clones): Bail out for
35865 an invalid attribute.
35867 2017-03-14 Richard Biener <rguenther@suse.de>
35869 * alias.c (struct alias_set_entry): Pack properly.
35870 * cfgloop.h (struct loop): Likewise.
35871 * cse.c (struct set): Likewise.
35872 * ipa-utils.c (struct searchc_env): Likewise.
35873 * loop-invariant.c (struct invariant): Likewise.
35874 * lra-remat.c (struct cand): Likewise.
35875 * recog.c (struct change_t): Likewise.
35876 * rtl.h (struct address_info): Likewise.
35877 * symbol-summary.h (function_summary): Likewise.
35878 * tree-loop-distribution.c (struct partition): Likewise.
35879 * tree-object-size.c (struct object_size_info): Likewise.
35880 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35881 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35882 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35883 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35884 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35885 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35886 (struct _stmt_vec_info): Likewise.
35888 2017-03-14 Martin Liska <mliska@suse.cz>
35891 * multiple_target.c (create_dispatcher_calls): Check that
35892 a target can create a function dispatcher.
35894 2017-03-14 Martin Liska <mliska@suse.cz>
35897 * multiple_target.c (expand_target_clones): Drop local.local
35898 flag for default implementation.
35900 2017-03-14 Richard Biener <rguenther@suse.de>
35902 PR tree-optimization/80030
35903 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35905 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35907 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35908 gcc_fallthrough() instead of __attribute__((fallthrough));
35910 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35912 * doc/gcc.texi: Remove "up" link to (DIR).
35913 * doc/gccint.texi: Ditto.
35915 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35917 * doc/install.texi (Specific) <avr>: Remove reference to
35920 2017-03-13 Jeff Law <law@redhat.com>
35922 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35923 attribute rather than comments.
35925 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35926 match_scratch operand is highest.
35928 2017-03-13 Martin Liska <mliska@suse.cz>
35930 PR middle-end/78339
35931 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35932 is a CHKP clone, use original declaration.
35934 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35936 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35937 (arc_conditional_register_usage): Use a different allocation order
35938 when optimizing for size.
35939 * common/config/arc/arc-common.c (arc_option_optimization_table):
35940 Section anchors default on when optimizing for size.
35942 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35944 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35946 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35948 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35949 * config/arc/arc.md (cpu_facility): Add cd variant.
35950 (*movqi_insn): Add code density variant.
35951 (*movhi_insn): Likewise.
35952 (*movqi_insn): Likewise.
35953 (*addsi3_mixed): Likewise.
35954 (subsi3_insn): Likewise.
35956 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35958 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35960 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35962 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35963 expressions with MINUS and UNARY ops.
35965 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35968 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35970 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35971 between vec_select and vector argument.
35972 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35973 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35974 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35975 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35976 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35977 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35979 2017-03-13 Richard Biener <rguenther@suse.de>
35982 * params.def (vect-max-peeling-for-alignment): Fix typo.
35984 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35986 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35987 issue that only occurred with binutils below 2.18.
35989 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35991 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35992 refer to binutils 2.11/2.12 minimum.
35994 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35996 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35997 ftp.kernel.org and simplify binutils requirement.
35999 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36001 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36003 (Optimize Options): Ditto. Also remove redundancy.
36005 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36007 PR translation/79848
36008 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36010 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36011 to G_ to avoid double translation.
36013 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36015 PR translation/79923
36016 * auto-profile.c (get_combined_location): Convert leading
36017 character of diagnostics to lower case and remove trailing period.
36018 (read_profile): Likewise for various diagnostics.
36019 * config/arm/arm.c (arm_option_override): Remove trailing period
36020 from various diagnostics.
36021 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36022 (msp430_expand_delay_cycles): Likewise.
36024 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36027 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36028 full command-line argument, rather than just "str".
36029 (aarch64_validate_march): Likewise.
36030 (aarch64_validate_mtune): Likewise.
36032 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36034 PR rtl-optimization/78911
36035 * lra-assigns.c (must_not_spill_p): New function.
36036 (spill_for): Use it.
36038 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36040 PR tree-optimization/79981
36041 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36042 ATOMIC_COMPARE_EXCHANGE ifn result.
36043 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36044 IFN_ATOMIC_COMPARE_EXCHANGE.
36046 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36049 * opts.c (parse_sanitizer_options): Add missing question mark to
36050 "did you mean" message.
36052 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36054 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36056 (VMULEUH_UNS): Likewise.
36057 (VMULOUB_UNS): Likewise.
36058 (VMULOUH_UNS): Likewise.
36059 * config/rs6000/rs6000.c (builtin_function_type): Remove
36060 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36062 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36065 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36066 x with result of extra_parsing_for_operand_code_0.
36067 (function_reader::extra_parsing_for_operand_code_0): Convert
36068 return type from void to rtx, returning x. When reading
36069 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36070 larger size containing struct block_symbol.
36072 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36074 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36075 -mfloat128-hardware without -m64.
36077 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36080 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36081 entries to the case statement that marks unsigned arguments to
36082 overloaded functions.
36084 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36086 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36087 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36089 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36092 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36093 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36095 2017-03-10 Martin Liska <mliska@suse.cz>
36099 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36100 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36103 2017-03-10 Olivier Hainque <hainque@adacore.com>
36105 * tree-switch-conversion (array_value_type): Start by resetting
36106 candidate type to it's main variant.
36108 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36110 PR rtl-optimization/79909
36111 * combine.c (try_combine): Use simplify_replace_rtx on individual
36112 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36113 of the whole CALL_INSN_FUNCTION_USAGE.
36115 PR tree-optimization/79972
36116 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36117 get_range_info on SSA_NAMEs. Formatting fixes.
36119 2017-03-10 Richard Biener <rguenther@suse.de>
36120 Jakub Jelinek <jakub@redhat.com>
36122 PR tree-optimization/77975
36123 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36124 edge to be constant.
36125 (get_val_for): For constant x return it. Formatting fix.
36126 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36127 would use the same bases as the current one.
36129 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36131 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36132 instead of vec_select for V1TImode.
36133 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36135 (VSX_LE_128): Add V1TI to this mode iterator.
36136 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36137 (*vsx_le_perm_store_<mode>): Likewise.
36138 (pre-reload splitter for VSX stores): Likewise.
36139 (post-reload splitter for VSX stores): Likewise.
36140 (*vsx_xxpermdi2_le_<mode>): Likewise.
36141 (*vsx_lxvd2x2_le_<mode>): Likewise.
36142 (*vsx_stxvd2x2_le_<mode>): Likewise.
36144 2017-03-09 Michael Eager <eager@eagercon.com>
36146 Correct failures with --enable-checking=yes,rtl.
36148 * config/microblaze/microblaze.c (microblaze_expand_shift):
36149 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36150 test for const0_rtx.
36151 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36152 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36154 2017-03-09 Richard Biener <rguenther@suse.de>
36156 PR tree-optimization/79977
36157 * graphite-scop-detection.c (scop_detection::merge_sese):
36158 Handle the case of extra exits to blocks dominating the entry.
36160 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36162 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36165 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36167 PR rtl-optimization/79949
36168 * lra-constraints.c (process_alt_operands): Check memory when
36169 trying to predict a cycle. Print about the overall increase.
36171 2017-03-09 Richard Biener <rguenther@suse.de>
36173 PR middle-end/79971
36174 * gimple-expr.c (useless_type_conversion_p): Preserve
36175 TYPE_SATURATING for fixed-point types.
36177 2017-03-09 Richard Biener <rguenther@suse.de>
36180 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36181 alignment of BLKmode params.
36183 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36186 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36187 (VALL_NO_V2Q): Likewise.
36189 * config/aarch64/aarch64-simd.md
36190 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36192 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36193 VALL_NO_V2Q mode iterator.
36194 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36196 2017-03-09 Martin Liska <mliska@suse.cz>
36198 PR tree-optimization/79631
36199 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36200 tree_int_cst_sign_bit just for INTEGER constants.
36202 2017-03-09 Martin Liska <mliska@suse.cz>
36206 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36209 2017-03-09 Marek Polacek <polacek@redhat.com>
36212 * tree.c (inchash::add_expr): Handle TREE_VEC.
36214 2017-03-09 Martin Liska <mliska@suse.cz>
36217 (chkp_narrow_size_and_offset): New function.
36218 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36219 (void chkp_parse_bit_field_ref): New function.
36220 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36221 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36223 2017-03-09 Martin Liska <mliska@suse.cz>
36226 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36227 (chkp_find_bounds_1): Remove gcc_unreachable.
36229 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36232 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36233 BUILT_IN_SYNC*, determine the access type from the size suffix and
36234 always build a MEM_REF with that type. Handle forgotten
36235 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36238 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36239 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36240 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36241 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36242 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36243 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36244 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36245 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36246 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36247 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36248 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36249 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36250 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36251 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36252 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36253 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36254 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36255 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36256 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36257 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36258 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36259 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36260 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36261 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36262 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36263 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36264 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36265 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36266 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36267 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36268 definitions outside of __OPTIMIZE__ guarded section.
36271 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36272 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36273 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36274 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36277 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36279 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36280 ("vfenez<mode>"): Add missing constraints.
36282 2017-03-08 Martin Sebor <msebor@redhat.com>
36285 * config/nds32/nds32.c (nds32_option_override):
36286 Fix misspelled diagnostic.
36288 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36291 * gimplify.c (gimplify_omp_for): Replace index var in outer
36292 taskloop statement with an artificial variable and add
36293 OMP_CLAUSE_PRIVATE clause for it.
36295 2017-03-08 Richard Biener <rguenther@suse.de>
36297 PR tree-optimization/79955
36298 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36299 for accesses that are completely outside of the variable.
36301 2017-03-08 Andrew Haley <aph@redhat.com>
36303 PR tree-optimization/79943
36304 * tree-ssa-loop-split.c (compute_new_first_bound): When
36305 calculating the new upper bound, (END-BEG) should be added, not
36308 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36310 * config/avr/avr.md (setmemhi): Make sure match_dup
36311 operand number comes before match_scratch.
36313 2017-03-08 Richard Biener <rguenther@suse.de>
36315 PR tree-optimization/79920
36316 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36317 with ncopies == 1 to ...
36318 (vect_transform_slp_perm_load): ... here. Properly compute
36319 all element loads by iterating VF times over the group. Do
36320 not handle ncopies (computed in a broken way) in
36321 vect_create_mask_and_perm.
36323 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36326 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36327 is a uniform vector, use uniform_vector_p return value instead of
36328 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36330 2017-03-07 Marek Polacek <polacek@redhat.com>
36332 PR middle-end/79809
36333 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36334 (alloca_call_type): Likewise.
36336 2017-03-07 Martin Liska <mliska@suse.cz>
36338 * gcov.c (process_args): Put comment to correct location.
36340 2017-03-07 Martin Liska <mliska@suse.cz>
36342 PR middle-end/68270
36343 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36344 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36345 (chkp_narrow_bounds_for_field): Likewise.
36346 (chkp_parse_array_and_component_ref): Pass one more argument to
36349 2017-03-07 Richard Biener <rguenther@suse.de>
36351 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36354 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36356 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36357 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36359 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36362 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36363 to end of description.
36364 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36366 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36368 PR rtl-optimization/79901
36369 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36371 (*avx512f_<code><mode>3<mask_name>): ... this.
36372 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36373 iterator instead of VI8_AVX2_AVX512BW.
36375 PR rtl-optimization/79901
36376 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36377 min/max expander, expand it using expand_vec_cond_expr.
36380 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36383 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36386 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36387 to void * for PCH reasons.
36388 * dwarf2out.c (output_loc_operands, output_die): Cast
36389 v.val_vec.array to unsigned char *.
36391 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36394 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36397 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36399 PR rtl-optimization/79571
36400 * lra-constraints.c (process_alt_operands): Calculate static
36401 reject and subtract it from overall when only addresses will be
36404 2017-03-06 Julia Koval <julia.koval@intel.com>
36407 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36408 incoming stack boundary to 128 for 64-bit targets.
36410 2017-03-06 Richard Biener <rguenther@suse.de>
36412 PR tree-optimization/79894
36413 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36414 to NULL after folding it.
36416 2017-03-06 Richard Biener <rguenther@suse.de>
36418 PR tree-optimization/79824
36419 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36420 check disabling peeling for gaps.
36422 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36424 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36425 attributes): Document gettimeofday.
36427 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36429 * config/s390/s390.c (s390_option_override_internal): Set
36430 PARAM_MIN_VECT_LOOP_BOUND
36432 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36434 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36435 * config/s390/s390.md: Likewise.
36437 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36440 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36441 (<avx2_avx512>_perm<mode>): Rename to ...
36442 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36444 (<avx512>_perm<mode>_mask): Rename to ...
36445 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36447 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36448 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36449 instead of VI8F_256_512.
36450 (avx512f_perm<mode>): New define_expand.
36451 (avx512f_perm<mode>_mask): Likewise.
36452 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36453 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36455 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36457 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36458 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36460 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36462 2017-03-06 Martin Liska <mliska@suse.cz>
36465 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36466 when having a SSA NAME w/o VAR_DECL assigned to it.
36468 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36470 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36471 msa_dpsub_<su>_d): Fix MODE for vec_select.
36473 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36475 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36477 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36479 2017-03-06 Richard Biener <rguenther@suse.de>
36481 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36482 * plugin.c (register_plugin_info): Likewise.
36483 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36485 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36487 * config/i386/sse.md (sse_storehps, sse_storelps,
36488 avx_<castmode><avxsizesuffix>_<castmode>,
36489 avx512f_<castmode><avxsizesuffix>_<castmode>,
36490 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36491 in condition that at least one operand is not a MEM.
36493 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36495 PR middle-end/79805
36496 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36497 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36499 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36500 gimple_call_nothrow_p flag based on whether original builtin can throw.
36501 If it can, emit following stmts on the fallthrough edge.
36502 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36503 don't create new bb if inserting just debug stmts on the edge, try to
36504 insert them on the fallthru bb or just reset debug stmts.
36506 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36509 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36510 restore recog_data (including the operand rtxes inside it) around
36511 the call to get_insn_template.
36513 2017-03-03 Martin Sebor <msebor@redhat.com>
36515 PR tree-optimization/79699
36516 * context.c (context::~context): Free MPFR caches to avoid
36517 a memory leak on program exit.
36519 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36521 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36522 Use wide_int::ulow () instead of .elt (0).
36524 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36526 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36527 (*pushxf): Limit oF constraint to 32bit targets and add oC
36528 constraint for 64bit targets.
36529 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36530 (*pushdf): Change rmF constraint to rmC.
36532 2017-03-03 Martin Liska <mliska@suse.cz>
36534 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36535 Remove unused variable.
36537 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36540 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36541 is a memory operand, increase num_memory.
36542 (ix86_expand_args_builtin): Likewise.
36544 2017-03-03 Jan Hubicka <jh@suse.cz>
36547 * ipa-devirt.c (maybe_record_node): Properly handle
36548 __cxa_pure_virtual visibility.
36550 2017-03-03 Martin Liska <mliska@suse.cz>
36552 PR tree-optimization/79803
36553 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36555 (pass_loop_prefetch::execute): Disabled optimization if an
36556 assumption about L1 cache size is not met.
36558 2017-03-03 Martin Liska <mliska@suse.cz>
36560 PR rtl-optimization/79574
36561 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36562 (hash_scan_set): Likewise.
36563 (dump_hash_table): Likewise.
36564 (hoist_code): Likewise.
36566 2017-03-03 Richard Biener <rguenther@suse.de>
36568 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36569 in place of uhigh ().
36570 (fixed_convert_from_real): Likewise.
36572 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36575 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36577 2017-03-03 Richard Biener <rguenther@suse.de>
36579 PR middle-end/79818
36580 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36581 TYPE_OVERFLOW_UNDEFINED check.
36583 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36585 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36587 (vector_ae_<mode>_p): Likewise.
36588 (vector_nez_<mode>_p): Likewise.
36589 (vector_ne_v2di_p): Likewise.
36590 (vector_ae_v2di_p): Likewise.
36591 (vector_ne_<mode>_p): Likewise.
36592 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36594 (vsx_tsqrt<mode>2_fe): Likewise.
36596 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36599 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36601 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36603 PR rtl-optimization/79780
36604 * cprop.c (one_cprop_pass): When second and further conditional trap
36605 in a single basic block is turned into an unconditional trap, turn it
36606 into a deleted note to avoid RTL verification failures.
36608 2017-03-02 Richard Biener <rguenther@suse.de>
36610 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36612 2017-03-02 Richard Biener <rguenther@suse.de>
36614 PR tree-optimization/79345
36616 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36617 param and abort the walk, returning -1 if it is hit.
36618 (walk_aliased_vdefs): Take a limit param and pass it on.
36619 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36620 defaulting to 0 and return a signed int.
36621 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36622 (check_defs): New helper.
36623 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36624 about uninitialized memory.
36625 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36626 bogus uninitialized warning.
36627 (fixed_convert_from_real): Likewise.
36629 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36631 PR tree-optimization/66768
36632 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36633 iv_use if base object can't be determined.
36635 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36637 PR tree-optimization/79345
36638 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36639 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36640 (get_pattern_stats): Initialize it.
36641 * genemit.c (gen_expand): Verify match_scratch numbers come after
36642 match_operand/match_dup numbers.
36643 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36644 match_scratch numbers.
36645 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36647 * config/s390/s390.md (trunctdsd2): Likewise.
36649 2017-03-02 Richard Biener <rguenther@suse.de>
36651 * wide-int.h (wide_int_storage::operator=): Implement in terms
36654 2017-03-02 Richard Biener <rguenther@suse.de>
36656 PR tree-optimization/79777
36657 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36658 the to insert expression to sth existing.
36660 2017-03-01 Martin Sebor <msebor@redhat.com>
36662 PR middle-end/79692
36663 * gimple-ssa-sprintf.c
36664 (directive::known_width_and_precision): New function.
36665 (format_integer): Use it.
36666 (get_mpfr_format_length): Consider the full range of precision
36667 when computing %g output with the # flag. Set the likely byte
36668 count to 3 rather than 1 when precision is indeterminate.
36669 (format_floating): Correct the lower bound of precision.
36671 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36673 * doc/invoke.texi: Document default code model for 64-bit Linux.
36675 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36678 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36679 udiv rather than div since input pattern is unsigned.
36681 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36683 * config/i386/i386.c (print_reg): Warn for values of
36684 unsupported size in integer register.
36686 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36689 * config/rs6000/predicates.md (current_file_function_operand): Do
36690 not allow self calls to be local if the function is replaceable.
36692 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36695 * config/rs6000/altivec.h (vec_ctz and others): Change the
36696 preprocessor macro that controls conditional compilation from
36697 _ARCH_PWR9 to __POWER9_VECTOR__.
36698 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36699 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36700 control (instead of _ARCH_PWR9 control) so that template
36701 definition uses power9-specific function.
36702 (vec_any_eq): Likewise.
36703 (vec_all_ne): Change macro definition to use a power9-specific
36704 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36705 _ARCH_PWR9 control).
36706 (vec_any_eq) Likewise.
36707 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36708 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36709 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36710 support for xvcmpnedp instruction.
36711 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36712 macro expansion so that Power9 implementation of vec_all_ne does
36713 not use the AltiVec predicate framework.
36714 (VCMPNEH_P): Likewise.
36715 (VCMPNEW_P): Likewise.
36716 (VCMPNED_P): Likewise.
36717 (VCMPNEFP_P): Likewise.
36718 (VCMPNEDP_P): Likewise.
36719 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36720 implementation of vec_any_eq to not use AltiVec predicate
36722 (VCMPAEH_P): Likewise.
36723 (VCMPAEW_P): Likewise.
36724 (VCMPAED_P): Likewise.
36725 (VCMPAEFP_P): Likewise.
36726 (VCMPAEDP_P): Likewise.
36727 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36728 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36729 not use the AltiVec predicate framework.
36730 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36731 of vec_any_eq to not use AltiVec predicate framework.
36732 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36733 support for predefined __POWER9_VECTOR__ macro to indicate that
36734 Power9 instruction selection is enabled.
36735 (altivec_overloaded_builtins): Remove extraneous
36736 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36737 function argument types RS6000_BTI_bool_V16QI and
36738 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36739 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36740 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36741 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36742 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36743 Power9 for implementations of vec_cmpne. Change the signature for
36744 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36745 (representing vec_all_ne) to remove the previously described first
36746 argument of type RS6000_BTI_INTSI, as this was an artifact of
36747 reliance on the AltiVec predicate framework, which is no longer
36748 used in the implementation of these functions. Add
36749 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36750 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36751 since, unlike the AltiVec predicate framework implementation, we
36752 do not share function descriptors between vec_alle and vec_anyeq.
36753 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36754 set of modes that receive special treatment even when
36755 TARGET_P9_VECTOR is true. The special treatment emits code that
36756 does not depend on Power9 instructions.
36757 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36758 define_expand to not rely on AltiVec predicate framework.
36759 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36761 (vector_ne_v2di_p): Change this define_expand to not rely on
36762 AltiVec predicate framework.
36763 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36765 (vector_ne_<mode>_p): Change this define_expand to not rely on
36766 AltiVec predicate framework.
36767 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36769 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36770 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36771 define_insn pattern.
36772 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36773 define_insn pattern because the xvcmpne<VSs>. instruction is not
36775 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36776 instruction is not supported.
36778 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36780 * config/nvptx/nvptx.c: Include intl.h.
36782 2017-03-01 Martin Jambor <mjambor@suse.cz>
36785 * ipa-prop.h (ipa_bits): Removed field known.
36786 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36787 to pointers. Adjusted their comments to warn about their sharing.
36788 (ipcp_transformation_summary): Change bits to a vector of pointers.
36789 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36790 (ipa_get_ipa_bits_for_value): Declare.
36791 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36792 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36793 (ipa_bits_hash_table): Likewise.
36794 (ipa_vr_ggc_hash_traits): Likewise.
36795 (ipa_vr_hash_table): Likewise.
36796 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36797 being pointers and vr_known being removed.
36798 (ipa_set_jf_unknown): Likewise.
36799 (ipa_get_ipa_bits_for_value): New function.
36800 (ipa_set_jfunc_bits): Likewise.
36801 (ipa_get_value_range): New overloaded functions.
36802 (ipa_set_jfunc_vr): Likewise.
36803 (ipa_compute_jump_functions_for_edge): Use the above functions to
36804 construct bits and vr parts of jump functions.
36805 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36806 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36808 (ipcp_grow_transformations_if_necessary): Also allocate
36809 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36811 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36812 them. Fix too long lines.
36813 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36814 vr_known being removed.
36815 (ipa_read_jump_function): Use new setter functions to construct bits
36816 and vr parts of jump functions or set them to NULL.
36817 (write_ipcp_transformation_info): Adjust for bits being pointers.
36818 (read_ipcp_transformation_info): Likewise.
36819 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36821 Include gt-ipa-prop.h.
36822 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36824 (ipcp_store_bits_results): Likewise.
36825 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36826 Do not write to existing jump functions but use a temporary instead.
36828 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36831 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36832 attempt to use its first operand as BIT_FIELD_REF base.
36834 2017-03-01 Richard Biener <rguenther@suse.de>
36836 PR middle-end/79721
36837 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36838 interpolating formula in wrapping arithmetic.
36839 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36841 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36843 PR tree-optimization/79734
36844 * tree-vect-generic.c (expand_vector_condition): Optimize
36845 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36846 Handle VEC_COND_EXPR where comparison has different inner width from
36847 type's inner width.
36849 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36851 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36852 markup, and similar issues. Remove @opindex entries for things
36853 that aren't options. Add missing -mmpy-option entries.
36855 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36857 PR tree-optimization/79737
36858 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36859 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36860 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36861 instead of byte_size. Formatting fix.
36862 (shift_bytes_in_array_right): Formatting fix.
36864 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36867 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36868 condition on optimize for the leaf function test.
36870 2017-02-28 Martin Liska <mliska@suse.cz>
36873 * read-rtl-function.c (function_reader::handle_unknown_directive):
36874 Bail out when one uses -flto.
36876 2017-02-28 Martin Liska <mliska@suse.cz>
36878 * common.opt: Replace space with tabular for options of <number>
36880 * config/i386/i386.opt: Show <number> value for
36881 -mlarge-data-threshold.
36882 * opts.c (print_filtered_help): Do not display number in hexadecimal
36885 2017-02-28 Martin Liska <mliska@suse.cz>
36887 * common.opt: Fix --help=option -Q for options which are of
36890 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36892 * config/i386/i386.c (print_reg): Error out for values
36893 of 8-bit size in invalid integer register.
36895 2017-02-28 Martin Sebor <msebor@redhat.com>
36897 PR tree-optimization/79691
36898 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36900 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36903 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36904 gcc_unreachable with output_operand_lossage.
36906 2017-02-28 Richard Biener <rguenther@suse.de>
36908 PR tree-optimization/79740
36909 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36911 (visit_nary_op): Insert the nary into the hashtable if we
36912 pattern-matched sth.
36913 * tree-ssa-pre.c (eliminate_insert): Robustify.
36915 2017-02-28 Richard Biener <rguenther@suse.de>
36917 PR middle-end/79731
36918 * fold-const.c (decode_field_reference): Reject out-of-bound
36921 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36923 * config/i386/i386.c: Include intl.h.
36924 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36925 instead of just cond ? "..." : "...".
36926 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36927 * coverage.c (read_counts_file): Likewise.
36928 * omp-offload.c: Include intl.h.
36929 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36930 of just cond ? "..." : "...".
36931 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36932 of just cond ? "..." : "...".
36934 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36937 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36939 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36940 'tune for' CPU name.
36941 * config/arm/arm-cpu-data.h: Regenerated.
36943 2017-02-28 Richard Biener <rguenther@suse.de>
36945 PR tree-optimization/79732
36946 * tree-inline.c (expand_call_inline): Do not shadow var.
36948 2017-02-28 Richard Biener <rguenther@suse.de>
36950 PR tree-optimization/79723
36951 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36952 address-space properly.
36954 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36956 * doc/optinfo.texi (Optimization groups): Fix option used for
36958 * doc/invoke.texi (-fopt-info): Document "omp".
36959 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36960 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36961 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36962 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36963 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36965 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36967 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36968 OPTGROUP_OMP with "omp".
36970 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36973 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36974 for arithmetic shift of unsigned V2DI.
36976 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36978 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36979 arc/linux.h headers.
36980 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36981 (LINK_SPEC): Likewise.
36982 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36983 (EXTRA_SPECS): Likewise.
36984 (STARTFILE_SPEC): Likewise.
36985 (ENDFILE_SPEC): Likewise.
36986 (LIB_SPEC): Likewise.
36987 (TARGET_SDATA_DEFAULT): Likewise.
36988 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36989 (MULTILIB_DEFAULTS): Likewise.
36990 (DWARF2_UNWIND_INFO): Likewise.
36991 * config/arc/big.h: New file.
36992 * config/arc/elf.h: Likewise.
36993 * config/arc/linux.h: Likewise.
36994 * config/arc/t-uClibc: Remove.
36996 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36998 PR tree-optimization/77536
36999 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37000 (tree_transform_and_unroll_loop): Use above function to compute the
37001 estimated niter of unrolled loop and use it when scaling profile.
37002 Also use count info rather than frequency if it's non-zero.
37003 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37004 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37005 (vect_transform_loop): Call above function.
37007 2017-02-27 Richard Biener <rguenther@suse.de>
37009 PR tree-optimization/45397
37010 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37011 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37012 (visit_nary_op): Add pattern matching for CSEing sign-changed
37013 or truncated operations with wider ones.
37015 2017-02-27 Richard Biener <rguenther@suse.de>
37017 PR tree-optimization/79690
37018 * tree-vect-stmts.c (vectorizable_store): Use vector type
37019 built from the DR with address-space.
37021 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37023 * doc/invoke.texi (Optimize Options): Refine the description
37024 of asan-use-after-return.
37026 2017-02-25 Alan Modra <amodra@gmail.com>
37028 PR rtl-optimization/79584
37029 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37030 base, not ad->base_term, the reg within base. Remove assertion
37031 that ad->base == ad->base_term. Replace gen_int_mode using
37032 bogus mode with const0_rtx.
37034 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37036 PR middle-end/79396
37037 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37038 FMA_EXPR like tcc_binary or tcc_unary.
37040 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37043 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37045 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37046 (output_loc_operands): Handle DW_OP_call_ref and
37047 DW_OP_GNU_variable_value.
37048 (struct variable_value_struct): New type.
37049 (struct variable_value_hasher): Likewise.
37050 (variable_value_hash): New variable.
37051 (string_types): Remove.
37052 (copy_loc_descr): New function.
37053 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37054 (prepend_loc_descr_to_each): New function.
37055 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37056 instead of add_loc_descr_to_each if the first argument is single
37057 location list and the second has multiple.
37058 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37059 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37060 when looking for variable value which doesn't have other location info.
37061 (loc_list_from_tree): Formatting fix.
37062 (gen_array_type_die): Simplify DW_AT_string_length handling.
37063 (adjust_string_types): Remove.
37064 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37065 string_types. Call resolve_variable_values.
37066 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37067 (resolve_addr_in_expr): Likewise. Add A argument.
37068 (copy_deref_exprloc): Remove deref argument. Adjust for the
37069 original expression being DW_OP_GNU_variable_value with optionally
37070 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37071 optionally after it.
37072 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37073 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37074 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37075 (variable_value_hasher::hash, variable_value_hasher::equal): New
37077 (resolve_variable_value_in_expr, resolve_variable_value,
37078 resolve_variable_values, note_variable_value_in_expr,
37079 note_variable_value): New functions.
37080 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37083 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37086 * opts.h (handle_generated_option): Add GENERATED_P argument.
37087 * opts-common.c (handle_option): Adjust function comment.
37088 (handle_generated_option): Add GENERATED_P argument, pass it to
37090 (control_warning_option): Pass false to handle_generated_option
37092 * opts.c (maybe_default_option): Pass true to handle_generated_option
37094 * optc-gen.awk: Likewise.
37096 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37098 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37099 a REG, look at the REG it is a SUBREG of.
37100 (splitter for cmpeqsi_t): Ditto.
37102 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37104 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37105 the special USEs with the pattern of the insn, not the insn itself.
37107 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37110 * doc/invoke.texi: Document -mload-store-pairs.
37112 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37113 Sandra Loosemore <sandra@codesourcery.com>
37115 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37116 argument isn't a CONST_INT.
37117 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37118 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37119 (nios2_validate_compare): Bypass alternate compare logic if *op2
37120 is not a CONST_INT.
37121 (ldstwm_operation_p): Return false if first_base is not a REG or
37122 if first_offset is not a CONST_INT.
37124 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37126 * config/cris/cris.md: Use correct operand in a define_peephole2.
37128 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37130 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37132 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37134 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37135 this_insn if it is an INSN or JUMP_INSN.
37136 (force_offsettable): Look at base, not at addr.
37137 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37138 on things that aren't necessarily CONST_INTs.
37140 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37142 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37143 -mfpmath=sse is the default also for x86-32 targets with SSE2
37144 instruction set when @option{-ffast-math} is enabled
37146 2017-02-24 Jeff Law <law@redhat.com>
37148 PR rtl-optimizatoin/79286
37149 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37152 2017-02-24 Richard Biener <rguenther@suse.de>
37154 PR tree-optimization/79389
37155 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37158 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37160 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37161 function comment to reflect reality.
37162 (loop_exits_before_overflow): Fix typo in function description.
37164 2017-02-24 Richard Biener <rguenther@suse.de>
37166 PR tree-optimization/79389
37167 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37168 properly that a threading opportunity exists. Detect conditional
37169 copy/constant propagation opportunities.
37171 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37173 * config/visium/visium.md (type): Add trap.
37174 (b): New mode attribute.
37175 (*btst): Rename into...
37176 (*btst<mode>): ...this and adjust.
37177 (*cbranchsi4_btst_insn): Rename into...
37178 (*cbranch<mode>4_btst_insn): ...this and adjust.
37179 (trap): New define_insn.
37181 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37183 PR tree-optimization/79389
37184 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37185 (noce_reversed_cond_code): New function.
37186 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37187 reversed_comparison_code. Formatting fix.
37188 (noce_try_store_flag): Test rev_cond != NULL in addition to
37189 reversed_comparison_code.
37190 (noce_try_store_flag_constants): Likewise.
37191 (noce_try_store_flag_mask): Likewise.
37192 (noce_try_addcc): Use rev_cond if non-NULL instead of
37193 reversed_comparison_code.
37194 (noce_try_cmove_arith): Likewise. Formatting fixes.
37195 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37196 (noce_find_if_block): Initialize rev_cond.
37197 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37198 instead of false as last argument never attempt to reverse it
37201 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37203 PR tree-optimization/79663
37204 * tree-predcom.c (combine_chains): Process refs in reverse order
37205 only for ZERO length chains, and add explaining comment.
37207 2017-02-23 Jeff Law <law@redhat.com>
37209 PR tree-optimization/79578
37210 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37211 in call to operand_equal_p.
37213 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37216 * config/i386/cpuid.h: Fix another undefined behavior.
37218 2017-02-23 Richard Biener <rguenther@suse.de>
37220 PR tree-optimization/79683
37221 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37222 vector types for data-refs.
37224 2017-02-23 Martin Liska <mliska@suse.cz>
37226 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37228 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37230 PR middle-end/79665
37231 * internal-fn.c (get_range_pos_neg): Moved to ...
37232 * tree.c (get_range_pos_neg): ... here. No longer static.
37233 * tree.h (get_range_pos_neg): New prototype.
37234 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37235 are known to be in between 0 and signed maximum inclusive, try to
37236 expand both unsigned and signed divmod and use the cheaper one from
37239 2017-02-22 Jeff Law <law@redhat.com>
37241 PR tree-optimization/79578
37242 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37243 to compare base operands.
37245 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37248 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37249 gpc_reg_operand instead of fpr_reg_operand.
37251 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37253 * config/mips/mips.c (mips_return_in_memory): Force FP
37254 vector types to be returned in memory for o32 ABI.
37256 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37258 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37259 instead of DW_TAG_member for static data member declarations and don't
37260 set no_linkage_name for static inline data members.
37261 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37264 2017-02-22 Martin Liska <mliska@suse.cz>
37266 * doc/invoke.texi: Replace inequality signs with square brackets
37269 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37272 * lra-constraints.c (simplify_operand_subreg): Handle
37273 WORD_REGISTER_OPERATIONS targets.
37275 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37278 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37279 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37280 elimination by swapping fld*.
37282 2017-02-22 Richard Biener <rguenther@suse.de>
37284 PR tree-optimization/79673
37285 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37286 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37287 irrelevant address-space qualifiers and avoiding a
37288 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37290 2017-02-22 Richard Biener <rguenther@suse.de>
37292 PR tree-optimization/79666
37293 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37294 to not symbolically negate if that may introduce undefined
37297 2017-02-22 Martin Liska <mliska@suse.cz>
37300 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37301 * data-streamer-out.c (streamer_write_gcov_count_stream):
37303 * value-prof.c (stream_out_histogram_value): Make assert more
37304 precise based on type of counter.
37306 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37309 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37310 Use nonimmediate_operand instead of memory_operand for operand 1.
37311 (float-extend standard_x87sse_constant_load splitter): Ditto.
37313 2017-02-21 Jeff Law <law@redhat.com>
37315 PR tree-optimization/79621
37316 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37317 blocks with edges to themselves.
37319 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37322 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37323 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37324 Use gimple_call_builtin_p.
37327 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37328 on temporarily removed DEBUG_INSNs.
37330 PR tree-optimization/79649
37331 * tree-loop-distribution.c (classify_partition): Give up on
37332 non-generic address space loads/stores.
37334 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37336 * doc/loop.texi (Loop manipulation): Remove nonexistent
37337 tree_ssa_loop_version from the documentation.
37338 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37340 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37343 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37344 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37345 * config/rs6000/rs6000.c: Include except.h.
37346 (rs6000_expand_split_stack_prologue): Call
37347 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37349 2017-02-21 Martin Jambor <mjambor@suse.cz>
37352 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37353 have been analyzed.
37355 2017-02-21 Martin Jambor <mjambor@suse.cz>
37357 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37358 for backward compatibility only.
37359 * doc/invoke.texi (Option Summary): Remove all references to
37360 -fipa-cp-alignment.
37362 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37366 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37368 * lra-constraints.c (curr_insn_transform): Handle
37369 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37371 2017-02-21 Martin Liska <mliska@suse.cz>
37373 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37375 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37377 PR translation/79638
37378 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37380 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37383 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37384 (arm_function_ok_for_sibcall): Return false for an indirect call by
37385 descriptor if all the argument registers are used.
37386 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37387 alignment of the function.
37389 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37391 PR tree-optimization/61441
37392 * simplify-rtx.c (simplify_const_unary_operation): For
37393 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37394 the sNaN unmodified.
37396 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37398 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37399 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37400 instead of SYSTEM_HEADER_DIR.
37402 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37403 Martin Liška <mliska@suse.cz>
37405 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37406 Fix typos and grammar, use active voice, and clarify.
37408 2017-02-20 Marek Polacek <polacek@redhat.com>
37410 PR middle-end/79537
37411 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37414 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37416 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37419 * config/i386/i386.c (ix86_expand_builtin): Handle
37420 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37421 ix86_builtins_isa[fcode].isa as a requirement of those
37422 flags and any other flag in the bitmask.
37423 (ix86_init_mmx_sse_builtins): Use 0 instead of
37424 ~OPTION_MASK_ISA_64BIT as mask.
37425 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37426 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37427 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37428 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37430 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37433 * lra-constraints.c (split_reg): Check requested split mode
37434 is supported by the register.
37436 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37438 * lra-constraints.c (simplify_operand_subreg): Remove early
37441 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37444 * lra-constraints.c (curr_insn_transform): Tighten condition
37445 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37447 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37450 * lra-constraints.c (curr_insn_transform): Handle
37451 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37453 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37456 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37458 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37460 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37463 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37466 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37468 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37471 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37473 * final.c (last_columnnum, override_columnnum): New variables.
37474 (final_start_function): Set last_columnnum, pass it to begin_prologue
37475 hook and pass 0 to dwarf2out_begin_prologue.
37476 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37477 to source_line debug hook.
37478 (notice_source_line): Compute last_columnnum and for debug_column_info
37479 return true on column changes.
37480 * debug.h (struct gcc_debug_hooks): Add column argument to
37481 source_line and begin_prologue hooks.
37482 (debug_nothing_int_charstar_int_bool): Remove prototype.
37483 (debug_nothing_int_int_charstar,
37484 debug_nothing_int_int_charstar_int_bool): New prototypes.
37485 (dwarf2out_begin_prologue): Add column argument.
37486 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37487 begin_prologue hooks.
37488 (debug_nothing_int_charstar_int_bool): Remove.
37489 (debug_nothing_int_int_charstar,
37490 debug_nothing_int_int_charstar_int_bool): New functions.
37491 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37492 through to dwarf2out_source_line.
37493 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37494 (dwarf2out_source_line): Add column argument, emit it if requested.
37495 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37497 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37498 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37499 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37500 through to dwarf2out_begin_prologue.
37501 (vmsdbgout_source_line): Add column argument, pass it through to
37502 dwarf2out_source_line.
37503 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37504 dbxout_source_line caller.
37505 (dbxout_source_line): Add column argument.
37507 * common.opt (gno-column-info, gcolumn-info): New options.
37508 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37509 (check_die): Also test for multiple DW_AT_decl_column attributes.
37510 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37511 DW_AT_decl_column if requested.
37512 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37514 (gen_variable_die): Likewise.
37515 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37516 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37519 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37520 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37521 (ix86_handle_option): Handle OPT_m3dnowa.
37522 * doc/invoke.texi (-m3dnowa): Document.
37523 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37524 -m3dnowa instead of -m3dnow -march=athlon.
37527 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37528 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37530 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37533 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37534 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37535 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37536 generator for vsx_xxpermdi_<mode>_be.
37537 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37538 force big-endian semantics.
37539 (vsx_xxpermdi_<mode>_be): New define_expand with same
37540 implementation as previous version of vsx_xxpermdi_<mode>.
37542 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37544 PR tree-optimization/79327
37545 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37546 variable, its initialization and use.
37548 2017-02-17 Julia Koval <julia.koval@intel.com>
37550 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37551 (OPTION_MASK_ISA_PKU_UNSET): New.
37552 (ix86_handle_option): Handle -mrdpid.
37553 * config/i386/cpuid.h (bit_RDPID): New.
37554 * config/i386/driver-i386.c (host_detect_local_cpu):
37555 Detect RDPID feature.
37556 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37557 * config/i386/i386-c.c (ix86_target_macros_internal):
37559 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37560 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37561 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37562 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37563 * config/i386/i386.md (define_insn "rdpid"): New.
37564 * config/i386/i386.opt Add -mrdpid.
37565 * config/i386/immintrin.h (_rdpid_u32): New.
37567 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37569 PR rtl-optimization/79541
37570 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37571 instead of transforming it into USE.
37573 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37575 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37576 If HONOR_SNANS (SFmode) force the input to a register.
37577 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37578 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37579 an frsp or similar insn.
37581 2017-02-17 Martin Liska <mliska@suse.cz>
37583 PR rtl-optimization/79577
37584 * params.def (selsched-max-sched-times): Increase minimum to 1.
37586 2017-02-17 Martin Liska <mliska@suse.cz>
37588 PR rtl-optimization/79574
37589 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37591 2017-02-17 Martin Liska <mliska@suse.cz>
37593 PR tree-optimization/79529
37594 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37595 ssa_defined_default_def_p to handle cases which are implicitly
37597 * tree-ssa.c (ssa_defined_default_def_p): New function.
37598 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37599 which are implicitly defined.
37600 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37602 2017-02-17 Richard Biener <rguenther@suse.de>
37604 PR middle-end/79576
37605 * params.def (max-ssa-name-query-depth): Limit to 10.
37607 2017-02-17 Richard Biener <rguenther@suse.de>
37609 PR tree-optimization/79552
37610 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37613 2017-02-17 Richard Biener <rguenther@suse.de>
37616 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37618 2017-02-17 Marek Polacek <polacek@redhat.com>
37620 PR middle-end/79536
37621 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37622 (fold_negate_expr): New wrapper.
37624 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37626 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37627 Correct terminology and de-emphasize pre-standard behavior.
37629 2017-02-16 Alan Modra <amodra@gmail.com>
37631 PR rtl-optimization/79286
37632 * ira.c (def_dominates_uses): New function.
37633 (update_equiv_regs): Don't create an equivalence for insns that
37634 may trap where the register def does not dominate the use.
37636 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37638 PR rtl-optimization/78127
37639 * lra.c (lra): Call lra_eliminate before finish the loop after
37642 2017-02-16 Richard Biener <rguenther@suse.de>
37644 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37646 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37647 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37648 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37649 (isl_val_int_from_wi): New function.
37650 (extract_affine_gmp): Rename to ...
37651 (extract_affine_wi): ... this, take a widest_int.
37652 (extract_affine_int): Just wrap extract_affine_wi.
37653 (add_param_constraints): Use isl_val_int_from_wi.
37654 (add_loop_constraints): Likewise, and extract_affine_wi.
37656 2017-02-15 Jeff Law <law@redhat.com>
37658 PR middle-end/79521
37659 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37660 ira_init_register_move_cost_if_necessary.
37662 2017-02-15 Martin Sebor <msebor@redhat.com>
37664 PR middle-end/32003
37665 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37666 removed in a prior commit.
37668 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37670 PR tree-optimization/79347
37671 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37672 counters during peeling.
37674 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37676 * Makefile.in (site.exp): Remove "set ISLVER".
37678 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37681 * real.c (real_from_integer): Call real_convert even for decimal.
37683 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37686 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37688 2017-02-14 Andrew Pinski <apinski@cavium.com>
37690 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37691 cores and change the partno/implementer to be correct.
37692 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37693 the 'B" as the implementer.
37694 * config/aarch64/aarch64-tune.md: Regenerate.
37696 2017-02-14 Carl Love <cel@us.ibm.com>
37698 * config/rs6000/rs6000.c: Add case statement entry to make the
37699 xvcvuxdsp built-in argument unsigned.
37700 * config/rs6000/vsx.md: Fix the source and return operand types so they
37701 match the instruction definitions from the ISA document. Fix typo
37702 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37705 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37708 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37709 member early_clobber_alts.
37710 * lra-lives.c (reg_early_clobber_p): New.
37711 (process_bb_lives): Use it.
37712 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37713 (debug_operand_data): Initialize early_clobber_alts.
37714 (setup_operand_alternative): Set up early_clobber_alts.
37715 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37716 alternatives to new_insn_reg.
37717 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37719 (lra_update_insn_regno_info): Pass the new arg.
37721 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37723 PR middle-end/79505
37724 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37725 (new_oacc_loop_raw): Don't clear already cleared fields.
37728 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37729 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37730 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37732 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37735 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37737 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37740 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37741 the extra instruction to the right place to store 128-bit constant
37744 2017-02-14 Martin Sebor <msebor@redhat.com>
37746 PR middle-end/79448
37747 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37748 warning for strings of unknown length.
37750 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37752 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37754 2017-02-14 Jeff Law <law@redhat.com>
37757 * ira-costs.c (scan_one_insn): Initialize register move costs
37758 for pseudos seen in USE/CLOBBER insns.
37760 PR tree-optimization/79095
37761 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37762 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37763 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37764 if the operands are known to be not equal, then the resulting range
37766 (intersect_ranges): If the new range is ~[0,0] and the old range is
37767 wide, then prefer ~[0,0].
37768 * tree-vrp.c (overflow_comparison_p_1): New function.
37769 (overflow_comparison_p): New function.
37770 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37771 if NAME is used in an overflow test.
37772 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37773 overflow check that can be expressed as an equality test, then adjust
37774 ops to be that equality test.
37776 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37778 * config/s390/s390-builtin-types.def: Remove flags argument.
37779 * config/s390/s390.c (s390_init_builtins): Likewise.
37781 2017-02-14 Martin Liska <mliska@suse.cz>
37783 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37784 vector. Fix trailing white spaces.
37786 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37788 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37791 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37793 PR rtl-optimization/68664
37794 * config/arm/arm.c (arm_sched_can_speculate_insn):
37795 New function. Declare prototype.
37796 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37798 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37800 PR rtl-optimization/68664
37801 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37803 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37805 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37807 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37808 max skip bytes for function, loop and jump.
37810 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37812 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37813 ABS_EXPR for gimple dump.
37815 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37818 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37820 PR tree-optimization/79408
37821 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37822 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37823 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37824 also if rhs1 is INTEGER_CST.
37826 2017-02-14 Richard Biener <rguenther@suse.de>
37828 PR middle-end/79432
37829 * tree-into-ssa.c (insert_phi_nodes): When the function can
37830 have abnormal edges rewrite SSA names with broken use-def
37831 dominance out of SSA and register them for PHI insertion.
37833 2017-02-13 Martin Sebor <msebor@redhat.com>
37835 PR middle-end/79496
37836 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37837 clearing info.nowrite flag when snprintf size argument is a range.
37839 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37841 * cprop.c (cprop_jump): Add missing space in string literal.
37842 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37843 (get_constraint_for_component_ref): Likewise.
37844 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37845 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37846 * lra-constraints.c (process_alt_operands): Likewise.
37847 * ipa-inline.c (inline_small_functions): Likewise.
37848 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37849 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37850 * trans-mem.c (diagnose_tm_1_op): Likewise.
37851 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37852 (grid_parallel_clauses_gridifiable): Likewise.
37854 * config/nvptx/mkoffload.c (process): Add space in between
37857 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37858 "MOD4_SSE_REGS" and "ALL_REGS".
37860 * spellcheck.c (test_data): Add , in between "foo" and "food".
37862 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37865 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37866 boundary crossing check and subsequent code generation agree.
37868 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37870 * config/aarch64/aarch64.c (has_memory_op): Delete.
37871 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37874 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37876 PR rtl-optimization/79388
37877 PR rtl-optimization/79450
37878 * combine.c (distribute_notes): When removing TEM_INSN for which
37879 corresponding dest has last value recorded, invalidate that last
37882 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37884 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37885 of explicit '@'. Add missing assembly comment marker on branch costs
37888 2017-02-13 Nathan Sidwell <nathan@acm.org>
37890 * gengtype-lex.l (<in_struct>): Add '/'.
37892 2017-02-13 Martin Liska <mliska@suse.cz>
37895 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37897 2017-02-13 Richard Biener <rguenther@suse.de>
37899 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37901 * configure: Re-generate.
37902 * config.in: Likewise.
37903 * graphite-dependences.c: Simplify as if
37904 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37905 * graphite-isl-ast-to-gimple.c: Likewise.
37906 * graphite-optimize-isl.c: Likewise.
37907 * graphite-poly.c: Likewise.
37908 * graphite-sese-to-poly.c: Likewise.
37909 * graphite.h: Likewise.
37910 * toplev.c: Include isl/version.h and use isl_version () for
37911 printing the ISL version.
37912 * doc/install.texi: Update ISL requirement.
37914 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37916 * doc/standards.texi (Standards): Update reference to
37919 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37921 * doc/extend.texi (Named Address Spaces): sourceware.org now
37923 * doc/install.texi (Binaries): Ditto.
37926 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37928 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37929 standard terminology "stringize"/"stringizing" throughout.
37930 * doc/cppinternals.texi: Likewise.
37932 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37934 * doc/extend.texi: Fix some spelling mistakes and typos.
37935 * doc/invoke.texi: Likewise.
37937 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37940 * params.def (inline-min-speedup) Change from 10 to 8.
37942 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37944 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37947 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37950 * ipa-inline-analysis.c (get_minimal_bb): New function.
37951 (record_modified): Use it.
37952 (remap_edge_change_prob): Handle also ancestor functions.
37954 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37956 * doc/contrib.texi (Contributors): Remove broken link into
37957 the Mauve CVS repository.
37959 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37961 PR middle-end/79454
37962 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37963 result computation whenever lhs doesn't have vector mode, not
37964 just when it has BLKmode.
37966 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37968 * doc/makefile.texi (profiledbootstrap): Refer to the
37969 installation instructions only in textual form.
37971 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37974 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37976 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37978 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37979 (Specific): Update mingw-w64 reference.
37981 (Specific): Remove broken link to Renesas RX processor.
37983 2017-02-10 Richard Biener <rguenther@suse.de>
37985 * toplev.c (process_options): Do not mention obsolete graphite
37986 options when printing sorry message about missing graphite support.
37987 Mention -floop-nest-optimize.
37989 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37991 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37992 (vtst_p16): Likewise.
37993 (vtstq_p8): Likewise.
37994 (vtstq_p16): Likewise.
37996 (vtstq_p64): Likewise.
37997 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37998 (vset_lane_p64): New.
37999 (vsetq_lane_p64): New.
38001 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38003 PR tree-optimization/79411
38004 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38005 stmt operands are SSA_NAMEs used in abnormal phis.
38006 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38009 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38012 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38015 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38017 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38019 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38021 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38025 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38026 not arbitrary TREE_CONSTANT.
38029 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38030 "omp declare target link" attribute unless is_global_var.
38031 * omp-offload.c (find_link_var_op): Likewise.
38033 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38034 Chung-Lin Tang <cltang@codesourcery.com>
38036 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38038 (gimplify_adjust_omp_clauses): Don't delete TILE.
38039 (gimplify_omp_for): Deal with TILE.
38040 * internal-fn.c (expand_GOACC_TILE): New function.
38041 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38043 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38044 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38046 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38047 avoid DIV for outermost collapse var.
38048 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38049 Remove out of date comments, fix whitespace.
38050 * omp-general.c (omp_extract_for_data): Deal with tiling.
38051 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38052 adjust OLF_DIM_BASE value.
38053 (struct omp_for_data): Add tiling field.
38054 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38055 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38056 for auto loops. Remove default auto determining, moved to
38057 oacc_loop_fixed_partitions.
38058 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38059 stmts, add e_mask field.
38060 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38061 (oacc_thread_numbers): Use oacc_dim_call.
38062 (oacc_xform_tile): New.
38063 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38064 (finish_oacc_loop): Adjust for ifns vector.
38065 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38066 add case for GOACC_TILE fns.
38067 (oacc_loop_xform_loop): Delete.
38068 (oacc_loop_process): Iterate over call list directly, and add
38069 handling for GOACC_TILE fns.
38070 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38072 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38073 vector partitioning to outer loops. Assign 2 partitions to loops
38074 when available. Add TILE handling.
38075 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38076 (execite_oacc_device_lower): Process GOACC_TILE fns,
38077 ignore unknown specs.
38078 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38079 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38080 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38082 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38084 * configure.ac (ACX_BUGURL): Update.
38085 * configure: Regenerate.
38087 2017-02-09 Richard Biener <rguenther@suse.de>
38089 PR tree-optimization/69823
38090 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38091 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38093 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38095 * config/arc/arc-c.def: Add __NPS400__ definition.
38096 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38097 (TARGET_NPS400): Define.
38099 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38101 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38103 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38104 pointer, arch_info.
38105 (arc_cpu_types): Fill the arch_info field with a pointer into the
38106 arc_arch_types table.
38107 (arc_selected_cpu): Declare.
38108 * config/arc/arc.c (arc_selected_cpu): Make global.
38109 (arc_selected_arch): Delete.
38110 (arc_base_cpu): Delete.
38111 (arc_override_options): Remove references to deleted variables,
38112 update access to arch information.
38113 (ARC_OPT): Update access to arch information.
38114 (ARC_OPTX): Likewise.
38115 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38116 (TARGET_ARC600): Update access to arch information.
38117 (TARGET_ARC601): Likewise.
38118 (TARGET_ARC700): Likewise.
38119 (TARGET_EM): Likewise.
38120 (TARGET_HS): Likewise.
38121 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38124 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38127 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38128 condition/operands for integer GE/LE/GEU/LEU operations.
38130 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38132 PR translation/79397
38133 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38136 2017-02-08 Martin Jambor <mjambor@suse.cz>
38139 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38140 whether allocation happened.
38141 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38142 nothing was allocated.
38144 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38146 PR tree-optimization/79408
38147 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38148 constant, but SSA_NAME with a known integer range, use the minimum
38149 of that range instead of op1 to determine if modulo can be replaced
38150 with its first operand.
38152 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38154 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38156 2017-02-08 Richard Biener <rguenther@suse.de>
38158 PR tree-optimization/71824
38159 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38160 Check all loops contained in the merged region.
38162 2017-02-07 Andrew Pinski <apinski@cavium.com>
38164 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38166 2017-02-07 Andrew Pinski <apinski@cavium.com>
38168 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38169 (thunderxt88): Likewise.
38170 (thunderxt81): Disable LSE and change v8.1 to v8.
38171 (thunderxt83): Likewise.
38173 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38174 Richard Biener <rguenther@suse.de>
38176 PR middle-end/79399
38177 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38178 type from int to size_t.
38179 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38181 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38183 PR rtl-optimization/79386
38184 * cprop.c (bypass_conditional_jumps): Initialize
38185 bypass_last_basic_block already before splitting bbs after
38186 unconditional traps...
38187 (bypass_conditional_jumps): ... rather than here.
38190 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38191 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38192 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38193 fix -masm=intel patterns.
38195 2017-02-07 Richard Biener <rguenther@suse.de>
38197 PR tree-optimization/79256
38198 PR middle-end/79278
38199 * builtins.c (get_object_alignment_2): Use min_align_of_type
38200 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38201 and ADJUST_FIELD_ALIGN.
38203 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38205 * doc/tm.texi: Regenerate.
38206 * stor-layout.c (layout_decl): Adjust.
38207 (update_alignment_for_field): Likewise.
38208 (place_field): Likewise.
38209 (min_align_of_type): Likewise.
38210 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38211 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38212 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38213 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38214 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38215 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38216 * config/i386/i386.c (x86_field_alignment): Likewise.
38217 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38218 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38219 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38220 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38221 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38222 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38226 2017-01-30 Richard Biener <rguenther@suse.de>
38228 PR tree-optimization/79256
38229 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38230 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38233 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38235 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38236 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38237 builtins to SImode and emit a zero-extend, if necessary.
38239 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38241 * docs/invoke.texi (RISC-V Options): Alphabetize.
38243 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38245 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38248 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38250 * config/riscv/riscv.c: New file.
38251 * common/config/riscv/riscv-common.c: Likewise.
38252 * config.gcc: Likewise.
38253 * config/riscv/constraints.md: Likewise.
38254 * config/riscv/elf.h: Likewise.
38255 * config/riscv/generic.md: Likewise.
38256 * config/riscv/linux.h: Likewise.
38257 * config/riscv/multilib-generator: Likewise.
38258 * config/riscv/peephole.md: Likewise.
38259 * config/riscv/pic.md: Likewise.
38260 * config/riscv/predicates.md: Likewise.
38261 * config/riscv/riscv-builtins.c: Likewise.
38262 * config/riscv/riscv-c.c: Likewise.
38263 * config/riscv/riscv-ftypes.def: Likewise.
38264 * config/riscv/riscv-modes.def: Likewise.
38265 * config/riscv/riscv-opts.h: Likewise.
38266 * config/riscv/riscv-protos.h: Likewise.
38267 * config/riscv/riscv.h: Likewise.
38268 * config/riscv/riscv.md: Likewise.
38269 * config/riscv/riscv.opt: Likewise.
38270 * config/riscv/sync.md: Likewise.
38271 * config/riscv/t-elf-multilib: Likewise.
38272 * config/riscv/t-linux: Likewise.
38273 * config/riscv/t-linux-multilib: Likewise.
38274 * config/riscv/t-riscv: Likewise.
38275 * configure.ac: Likewise.
38276 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38277 Waterman as RISC-V maintainers.
38278 * doc/install.texi: Add RISC-V entries.
38279 * doc/invoke.texi: Add RISC-V options section.
38280 * doc/md.texi: Add RISC-V constraints section.
38281 * configure: Regenerated.
38283 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38286 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38287 false values to be constant vectors with all 0 or all 1 bits set.
38288 (vcondu<mode><mode>): Likewise.
38289 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38291 (fpmask_comparison_operator): Update comment.
38292 (vecint_comparison_operator): New predicate.
38293 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38294 vector conditionals when the true and false values are constant
38295 vectors with all 0 bits or all 1 bits set.
38297 2017-02-06 Martin Sebor <msebor@redhat.com>
38299 PR tree-optimization/79376
38300 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38302 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38304 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38305 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38306 to simplify split condition.
38308 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38310 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38311 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38314 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38316 PR rtl-optimization/68664
38317 * target.def (can_speculate_insn): New hook.
38318 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38319 * doc/tm.texi: Regenerate.
38320 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38321 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38322 (rs6000_sched_can_speculate_insn): New function.
38324 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38326 PR tree-optimization/79284
38327 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38328 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38329 vectorizable_mask_load_store, vectorizable_operation,
38330 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38331 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38332 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38333 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38334 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38335 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38336 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38337 is_gimple_assign (stmt). Replace another such test with
38338 is_gimple_assign (stmt).
38340 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38343 * config/avr/avr.c (rtl-iter.h): Include it.
38344 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38345 (avr_legitimate_combined_insn): ...and implementation.
38347 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38349 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38350 * config/s390/s390.c (s390_const_operand_ok)
38351 (s390_canonicalize_comparison, s390_extract_part)
38352 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38353 (s390_contiguous_bitmask_p, s390_rtx_costs)
38354 (legitimize_pic_address): Likewise.
38355 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38356 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38357 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38358 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38359 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38361 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38363 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38364 REGNO($0) == REGNO($1).
38366 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38368 * config/s390/linux.h(SIZE_TYPE): Add comment.
38370 2017-02-06 Julian Brown <julian@codesourcery.com>
38371 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38372 Virendra Pathak <virendra.pathak@broadcom.com>
38374 * config/aarch64/aarch64-cores.def: Change the scheduler
38376 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38377 * config/aarch64/thunderx2t99.md: New file.
38379 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38381 * doc/standards.texi (Go Language): Update link to language
38384 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38386 * tree-eh.c (lower_resx): Sanitize profile.
38387 (cleanup_empty_eh_move_lp): Likewise.
38389 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38392 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38394 * cfgloopmanip.h (loop_version): Update prototype.
38395 * modulo-sched.c (sms_schedule): Update call of loop_version.
38396 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38397 * tree-parloops.c (gen_parallel_loop): Likewise.
38398 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38399 * tree-ssa-loop-split.c (split_loop): Likewise.
38400 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38401 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38403 2017-02-05 Martin Liska <mliska@suse.cz>
38406 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38408 (print_operand_address): Initialize a struct to zero.
38410 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38412 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38413 garbage collector only in textual form.
38415 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38417 * doc/extend.texi (x86 specific memory model extensions for
38418 transactional memory): Simplify a phrase.
38420 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38423 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38424 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38425 (atomic_storedi_1): Likewise.
38427 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38429 PR tree-optimization/79338
38430 * tree-parloops.c (gather_scalar_reductions): Don't call
38431 vect_analyze_loop_form for loop->inner before destroying loop's
38434 2017-02-03 Martin Sebor <msebor@redhat.com>
38436 PR tree-optimization/79327
38437 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38438 when precision has resulted in leading zeros.
38439 (format_integer): Adjust the likely counter to assume an unknown
38440 argument that may be zero is non-zero.
38442 2017-02-03 Jason Merrill <jason@redhat.com>
38445 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38446 avoid copying non-taken branch.
38448 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38450 PR tree-optimization/79340
38451 * tree-vect-loop.c (vectorizable_reduction): Release
38452 vec_defs elements after safe_splicing them into other vectors.
38455 PR tree-optimization/79327
38456 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38457 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38459 (format_integer): Use wide_int_to_tree instead of build_int_cst
38460 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38461 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38462 of shortest and longest sequence.
38464 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38466 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38467 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38469 2017-02-03 Walter Lee <walt@tilera.com>
38472 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38473 after initial stackframe link reg save.
38474 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38476 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38479 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38480 wu for stxssp alternative.
38482 2017-02-03 Martin Sebor <msebor@redhat.com>
38484 PR tree-optimization/79352
38485 * gimple-fold.c (get_range_strlen): Add argument.
38486 (get_range_strlen): Change return type to bool.
38487 (get_maxval_strlen): Pass in a dummy argument.
38488 * gimple-fold.h (get_range_strlen): Change return type to bool.
38489 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38490 * tree.h (array_at_struct_end_p): Add argument.
38491 * tree.c (array_at_struct_end_p): Handle it.
38493 2017-02-03 Martin Liska <mliska@suse.cz>
38496 * multiple_target.c (create_dispatcher_calls): Redirect edge
38497 from a caller of a dispatcher.
38498 (expand_target_clones): Make the clones local.
38499 (ipa_target_clone): Do both target clones and resolvers.
38500 (ipa_dispatcher_calls): Remove the pass.
38501 (pass_dispatcher_calls::gate): Likewise.
38502 (make_pass_dispatcher_calls): Likewise.
38503 * passes.def (pass_target_clone): Put as very first IPA early
38506 2017-02-03 Martin Liska <mliska@suse.cz>
38508 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38509 in case of a function with ifunc attribute.
38511 2017-02-03 Martin Liska <mliska@suse.cz>
38513 * cgraph.c (cgraph_node::dump): Dump function version info.
38514 * symtab.c (symtab_node::dump_base): Add missing new line.
38516 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38518 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38519 (ifcombine_ifandif): Use it.
38521 2017-02-03 Martin Liska <mliska@suse.cz>
38523 * doc/invoke.texi: Document default value for
38524 use-after-scope-direct-emission-threshold.
38526 2017-02-03 Martin Liska <mliska@suse.cz>
38528 PR tree-optimization/79339
38529 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38530 (format_floating): Likewise.
38532 2017-02-03 Martin Liska <mliska@suse.cz>
38535 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38537 (ipa_node_params_t::remove): Likewise.
38538 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38539 initialization from removed ipa_node_params_t::insert.
38540 (ipa_node_params::~ipa_node_params): Move from removed
38541 ipa_node_params_t::release.
38542 * symbol-summary.h (symbol_summary::m_released): New member.
38543 Do not release a summary twice. Do not allow to call finalizer
38544 for types of a summary that live in GGC memory.
38546 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38548 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38551 2017-02-02 Martin Sebor <msebor@redhat.com>
38553 PR middle-end/79275
38554 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38555 (format_string): Tighten up the range of output for non-constant
38556 strings and correct the expected range for wide non-constant strings.
38558 2017-02-02 Martin Sebor <msebor@redhat.com>
38560 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38562 PR middle-end/32003
38563 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38565 (-fdump-tree-@var): Add to index and document how to come up
38566 with pass-specific option and dump file names.
38567 (-fdump-passes): Clarify where to look for output.
38569 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38571 PR middle-end/77445
38572 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38573 statistics of the analyzed path; allow threading for speed when
38574 any of BBs along the path are optimized for speed.
38576 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38578 PR middle-end/78468
38579 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38580 settings of the virtual registers.
38583 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38585 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38586 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38589 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38591 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38592 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38594 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38596 * config/s390/s390.md: Add missing comments with the expanded
38598 * config/s390/vector.md: Likewise.
38599 * config/s390/vx-builtins.md: Likewise.
38601 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38604 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38605 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38606 conditions on a single line.
38608 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38610 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38611 __S390_VX__ to __VX__.
38613 2017-02-01 Andrew Pinski <apinski@cavium.com>
38615 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38616 stmt_info to record_stmt_cost.
38617 (vect_get_known_peeling_cost): Pass stmt_info if known to
38619 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38620 cpu_vector_cost field into
38621 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38622 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38623 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38624 splitting of scalar_stmt_cost and vec_stmt_cost.
38625 (thunderx_vector_cost): Likewise.
38626 (cortexa57_vector_cost): LIkewise.
38627 (exynosm1_vector_cost): Likewise.
38628 (xgene1_vector_cost): Likewise.
38629 (thunderx2t99_vector_cost): Improve after the splitting of the two
38631 (aarch64_builtin_vectorization_cost): Update for the splitting of
38632 scalar_stmt_cost and vec_stmt_cost.
38634 2017-02-01 Torvald Riegel <triegel@redhat.com>
38635 Richard Henderson <rth@redhat.com>
38637 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38638 conditional on existance of a fast atomic load.
38639 * optabs-query.c (can_atomic_load_p): New function.
38640 * optabs-query.h (can_atomic_load_p): Declare it.
38641 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38642 no fast atomic load is available for the particular size of access.
38643 (expand_atomic_compare_and_swap): Likewise.
38644 (expand_atomic_load): Likewise.
38645 (expand_atomic_store): Likewise.
38646 (expand_atomic_fetch_op): Likewise.
38647 * testsuite/lib/target-supports.exp
38648 (check_effective_target_sync_int_128): Remove x86 because it provides
38649 no fast atomic load.
38650 (check_effective_target_sync_int_128_runtime): Likewise.
38652 2017-02-01 Richard Biener <rguenther@suse.de>
38654 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38655 (graphite_transform_loops): Provide opt-info for optimized nests.
38656 * tree-parloop.c (parallelize_loops): Provide opt-info for
38657 parallelized loops.
38659 2017-02-01 Richard Biener <rguenther@suse.de>
38661 PR middle-end/79315
38662 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38663 was not set before.
38665 2017-02-01 Richard Biener <rguenther@suse.de>
38667 PR tree-optimization/71824
38668 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38669 Verify the loops are valid in the merged SESE region.
38670 (scop_detection::can_represent_loop_1): Check analyzing the
38671 evolution of the number of iterations in the region succeeds.
38673 2017-01-31 Ian Lance Taylor <iant@golang.org>
38675 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38676 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38678 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38680 PR preprocessor/79210
38681 * input.c (get_substring_ranges_for_loc): Replace line_width
38682 assertion with error-handling.
38684 2017-01-31 Richard Biener <rguenther@suse.de>
38686 PR tree-optimization/77318
38687 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38688 (create_pw_aff_from_tree): Take loop parameter.
38689 (add_condition_to_pbb): Pass loop of the condition to
38690 create_pw_aff_from_tree.
38692 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38694 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38695 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38697 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38701 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38703 (convert_int_to_float128): Likewise.
38704 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38705 (convert_int_to_float128): Likewise.
38706 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38707 (UNSPEC_IEEE128_CONVERT): Likewise.
38708 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38709 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38710 Use local variables for IBM extended format.
38711 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38712 (fix_trunc<mode>si2_fprs): Likewise.
38713 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38714 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38715 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38716 to know that we can now have integers of all sizes in vector
38718 (fix<uns>_<mode>di2_hw): Likewise.
38719 (float<uns>_<mode>si2_hw): Likewise.
38720 (fix_<mode>si2_hw): Likewise.
38721 (fixuns_<mode>si2_hw): Likewise.
38722 (float<uns>_<mode>di2_hw): Likewise.
38723 (float_<mode>di2_hw): Likewise.
38724 (float_<mode>si2_hw): Likewise.
38725 (floatuns_<mode>di2_hw): Likewise.
38726 (floatuns_<mode>si2_hw): Likewise.
38727 (xscvqp<su>wz_<mode>): Delete, no longer used.
38728 (xscvqp<su>dz_<mode>): Likewise.
38729 (xscv<su>dqp_<mode>): Likewise.
38730 (ieee128_mfvsrd_64bit): Likewise.
38731 (ieee128_mfvsrd_32bit): Likewise.
38732 (ieee128_mfvsrwz): Likewise.
38733 (ieee128_mtvsrw): Likewise.
38734 (ieee128_mtvsrd_64bit): Likewise.
38735 (ieee128_mtvsrd_32bit): Likewise.
38737 2017-01-31 Martin Liska <mliska@suse.cz>
38740 * ipa-prop.c (ipa_free_all_node_params): Call release method
38741 instead of ~sumbol_summary to not to trigger double times
38744 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38746 PR tree-optimization/71691
38747 * bitmap.h (class auto_bitmap): New.
38748 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38749 is_maybe_undefined instead of ssa_undefined_value_p.
38751 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38753 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38754 __S390_ARCH_LEVEL__ to __ARCH__.
38756 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38758 PR tree-optimization/79267
38759 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38760 if should_remove_lhs_p is true.
38762 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38765 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38766 (add_alignment_attribute): New.
38767 (base_type_die): Add alignment attribute.
38768 (subrange_type_die): Likewise.
38769 (modified_type_die): Likewise.
38770 (gen_array_type_die): Likewise.
38771 (gen_descr_array_type_die: Likewise.
38772 (gen_enumeration_type_die): Likewise.
38773 (gen_subprogram_die): Likewise.
38774 (gen_variable_die): Likewise.
38775 (gen_field_die): Likewise.
38776 (gen_ptr_to_mbr_type_die): Likewise.
38777 (gen_struct_or_union_type_die): Likewise.
38778 (gen_subroutine_type_die): Likewise.
38779 (gen_typedef_die): Likewise.
38780 (base_type_cmp): Compare alignment attribute.
38782 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38785 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38786 (setb_unsigned) New pattern for setb with CCUNS.
38787 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38788 subfc./subfe sequence to avoid overflow problems. Generate a
38789 shorter sequence with cmpld/setb for power9.
38790 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38791 for generating subfc. instruction.
38792 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38793 now uses this instruction.
38795 2017-01-30 Ian Lance Taylor <iant@google.com>
38798 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38799 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38801 2017-01-30 Martin Sebor <msebor@redhat.com>
38803 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38804 Move constant to the right of a relational operator.
38805 (get_mpfr_format_length, format_character, format_string): Ditto.
38806 (should_warn_p, maybe_warn): Same.
38808 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38810 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38813 * asan.c (get_translation_unit_decl): Remove function.
38814 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38816 2017-01-30 Martin Liska <mliska@suse.cz>
38818 PR gcov-profile/79259
38819 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38820 -fprofile-generate.
38822 2017-01-30 Martin Liska <mliska@suse.cz>
38825 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38826 Initialize variables with NULL value.
38828 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38831 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38833 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38835 2017-01-30 Richard Biener <rguenther@suse.de>
38837 PR tree-optimization/79276
38838 * tree-vrp.c (process_assert_insertions): Properly adjust common
38839 when removing a duplicate.
38841 2017-01-30 Richard Biener <rguenther@suse.de>
38843 PR tree-optimization/79256
38844 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38845 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38847 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38849 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38852 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38853 ("*r<noxa>sbg_<mode>_sll_bitmask")
38854 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38855 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38856 Use contiguous_bitmask_nowrap_operand.
38858 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38861 * config/rs6000/altivec.h (vec_xl): Revise #define.
38862 (vec_xst): Likewise.
38864 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38866 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38868 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38870 PR rtl-optimization/79194
38871 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38872 traps before call to bypass_conditional_jumps.
38874 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38876 PR tree-optimization/71374
38877 * lra-constraints.c (check_conflict_input_operands): New.
38878 (match_reload): Use it.
38880 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38883 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38884 account to calculate conflict_set.
38886 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38888 PR rtl-optimization/78559
38889 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38890 other_insn in combine.
38892 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38894 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38895 uint16_type_node for BT_UINT16.
38897 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38899 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38900 "RTL Tests" to menu.
38901 (GIMPLE Tests): New node.
38902 (RTL Tests): New node.
38904 2017-01-27 Richard Biener <rguenther@suse.de>
38906 PR tree-optimization/79245
38907 * tree-loop-distribution.c (distribute_loop): Apply cost
38908 modeling also to detected patterns.
38910 2017-01-27 Richard Biener <rguenther@suse.de>
38912 PR tree-optimization/71433
38913 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38914 (compare_assert_loc): New function.
38915 (process_assert_insertions): Sort and optimize assert locations
38916 to remove duplicates and push down identical assertions on
38917 edges to their destination block.
38919 2017-01-27 Richard Biener <rguenther@suse.de>
38921 PR tree-optimization/79244
38922 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38923 out SSA names even if abnormal.
38925 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38927 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38928 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38929 instead of MPFR_RNDN.
38931 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38934 * arm.c (arm_option_override): Don't call build_target_option_node
38935 until after doing all option overrides.
38936 (arm_valid_target_attribute_tree): Likewise.
38938 2017-01-27 Martin Liska <mliska@suse.cz>
38940 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38941 for {cd}tors and C++ {cd}tors.
38943 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38945 * config/s390/s390.md ("*setmem_long_and")
38946 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38948 2017-01-26 Martin Sebor <msebor@redhat.com>
38950 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38953 2017-01-26 Martin Sebor <msebor@redhat.com>
38955 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38956 HAVE_DFmode before using XFmode or DFmode.
38957 (parse_directive): Avoid using the z length modifier to avoid
38958 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38960 PR middle-end/78703
38961 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38962 to accept adjustment as an array.
38963 (get_int_range): New function.
38964 (struct directive): Make width and prec arrays.
38965 (directive::set_width, directive::set_precision): Call get_int_range.
38966 (format_integer, format_floating): Handle width and precision ranges.
38967 (format_string, parse_directive): Same.
38969 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38972 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38973 comdat_type_p set, just clone them, but keep the children in the
38977 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38978 which have direct callers with -fvar-tracking-assignments enabled
38980 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38981 inside of type units.
38983 2017-01-26 Martin Sebor <msebor@redhat.com>
38985 PR middle-end/78703
38986 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38988 (struct format_result): Replace number_chars, number_chars_min,
38989 and number_chars_max with a single member of struct result_range.
38991 (format_result::operator+=): Adjust.
38992 (struct fmtresult): Remove bounded. Handle likely and unlikely
38994 (fmtresult::adjust_for_width_or_precision): New function.
38995 (fmtresult:type_max_digits): New function.
38996 (bytes_remaining): Handle likely and unlikely counters.
38997 (min_bytes_remaining): Remove.
38998 (format_percent): Simplify.
38999 (format_integer, format_floating): Set likely and unlikely counters.
39000 (get_string_length, format_character, format_string): Same.
39001 (format_plain, should_warn_p): New function.
39002 (maybe_warn): Call should_warn_p. Update diagnostic messages
39003 and handle those for all directives, including plain strings.
39004 (format_directive): Handle likely and unlikely counters.
39005 Remove unnecessary quoting from diagnostics. Add an informational
39007 (add_bytes): Remove.
39008 (pass_sprintf_length::compute_format_length): Simplify.
39009 (try_substitute_return_value): Handle likely and unlikely counters.
39011 2017-01-26 Carl Love <cel@us.ibm.com>
39013 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39014 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39016 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39019 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39020 endianess for subregs into account.
39021 * lra-constraints.c (lra_constraints): Do risky transformations
39022 always on the first iteration.
39023 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39024 last_call_used_reg_set.
39025 (process_bb_lives): Define and use last_call_used_reg_set.
39026 * lra.c (lra): Always continue after lra_constraints on the first
39029 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39031 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39033 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39035 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39037 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39038 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39039 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39040 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39041 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39042 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39043 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39044 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39045 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39047 2017-01-26 Marek Polacek <polacek@redhat.com>
39050 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39051 for the third operand.
39053 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39055 PR middle-end/79236
39056 * omp-low.c (struct omp_context): Add simt_stmt field.
39057 (scan_omp_for): Return omp_context *.
39058 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39059 context to the _simt_ SIMD stmt.
39060 (lower_omp_for): For combined SIMD with sibling _simt_
39061 SIMD, make sure to use the same decls in _looptemp_
39062 clauses as in the sibling.
39064 2017-01-26 David Sherwood <david.sherwood@arm.com>
39066 PR middle-end/79212
39067 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39070 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39073 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39074 emit fld b; fld a; if possible.
39076 * brig-builtins.def: Update copyright years.
39077 * config/arm/arm_acle_builtins.def: Update copyright years.
39079 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39082 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39083 constraint instead of o for the stxsd instruction.
39085 2017-01-25 Carl Love <cel@us.ibm.com>
39087 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39088 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39090 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39092 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39094 2017-01-25 Richard Biener <rguenther@suse.de>
39096 PR tree-optimization/69264
39097 * target.def (vector_alignment_reachable): Improve documentation.
39098 * doc/tm.texi: Regenerate.
39099 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39101 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39102 earlier changes with respect to TYPE_USER_ALIGN.
39103 (vector_alignment_reachable_p): Likewise. Improve dumping.
39105 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39108 * config/arm/arm.md (xordi3): Force constant operand into a register
39111 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39113 * doc/invoke.texi (-fstore-merging): Correct default optimization
39114 levels at which it is enabled.
39115 (-O): Move -fstore-merging from list to...
39118 2017-01-25 Richard Biener <rguenther@suse.de>
39121 * omp-expand.c: Include debug.h.
39122 (expand_omp_taskreg): Make sure to generate early debug before
39123 outlining anything from a function.
39124 (expand_omp_target): Likewise.
39125 (grid_expand_target_grid_body): Likewise.
39127 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39130 * asan.c (get_translation_unit_decl): New function.
39131 (asan_add_global): Extract modules file name from globals
39132 TRANSLATION_UNIT_DECL name.
39134 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39137 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39138 for long calls with APCS frame and VFP.
39140 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39142 * cfg.c (original_copy_tables_initialized_p): New function.
39143 * cfg.h (original_copy_tables_initialized_p): New decl.
39144 * cfgrtl.c (relink_block_chain): Guard the call to
39145 free_original_copy_tables with a call to
39146 original_copy_tables_initialized_p.
39147 * cgraph.h (symtab_node::native_rtl_p): New decl.
39148 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39149 (symtab_node::needed_p): Don't assert for early assembly output
39150 for __RTL functions.
39151 (cgraph_node::finalize_function): Set "force_output" for __RTL
39153 (cgraph_node::analyze): Bail out early for __RTL functions.
39154 (analyze_functions): Update assertion to support __RTL functions.
39155 (cgraph_node::expand): Bail out early for __RTL functions.
39156 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39158 * function.h (struct function): Update comment for field
39160 * gimple-expr.c: Include "tree-pass.h".
39161 (gimple_has_body_p): Return false for __RTL functions.
39162 * Makefile.in (OBJS): Add run-rtl-passes.o.
39163 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39165 (gcc::pass_manager::get_clean_slate): New accessor.
39166 * passes.c: Include "insn-addr.h".
39167 (should_skip_pass_p): Add logging. Update logic for running
39168 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39169 property-provider override so it is only done for gimple passes.
39171 (skip_pass): New function.
39172 (execute_one_pass): Call skip_pass when skipping passes.
39173 * read-md.c (md_reader::read_char): Support filtering
39174 the input to a subset of line numbers.
39175 (md_reader::md_reader): Initialize fields
39176 m_first_line and m_last_line.
39177 (md_reader::read_file_fragment): New function.
39178 * read-md.h (md_reader::read_file_fragment): New decl.
39179 (md_reader::m_first_line): New field.
39180 (md_reader::m_last_line): New field.
39181 * read-rtl-function.c (function_reader::create_function): Only
39182 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39183 curr_properties. Set DECL_INITIAL to a dummy block.
39184 (read_rtl_function_body_from_file_range): New function.
39185 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39187 * run-rtl-passes.c: New file.
39188 * run-rtl-passes.h: New file.
39190 2017-01-24 Jeff Law <law@redhat.com>
39192 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39195 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39197 PR tree-optimization/79159
39198 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39199 (record_nonwrapping_iv): Improve boundary using above function if no
39200 value range information.
39202 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39203 Martin Jambor <mjambor@suse.cz>
39205 * brig-builtins.def: New file.
39206 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39207 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39208 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39209 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39210 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39211 * builtin-types.def (BT_INT8): New.
39212 (BT_INT16): Likewise.
39213 (BT_UINT8): Likewise.
39214 (BT_UINT16): Likewise.
39215 (BT_FN_ULONG): Likewise.
39216 (BT_FN_UINT_INT): Likewise.
39217 (BT_FN_UINT_ULONG): Likewise.
39218 (BT_FN_UINT_LONG): Likewise.
39219 (BT_FN_UINT_PTR): Likewise.
39220 (BT_FN_ULONG_PTR): Likewise.
39221 (BT_FN_INT8_FLOAT): Likewise.
39222 (BT_FN_INT16_FLOAT): Likewise.
39223 (BT_FN_UINT32_FLOAT): Likewise.
39224 (BT_FN_UINT16_FLOAT): Likewise.
39225 (BT_FN_UINT8_FLOAT): Likewise.
39226 (BT_FN_UINT64_FLOAT): Likewise.
39227 (BT_FN_UINT16_UINT32): Likewise.
39228 (BT_FN_UINT32_UINT16): Likewise.
39229 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39230 (BT_FN_INT_PTR_INT): Likewise.
39231 (BT_FN_UINT_PTR_UINT): Likewise.
39232 (BT_FN_LONG_PTR_LONG): Likewise.
39233 (BT_FN_ULONG_PTR_ULONG): Likewise.
39234 (BT_FN_VOID_UINT64_UINT64): Likewise.
39235 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39236 (BT_FN_INT8_INT8_INT8): Likewise.
39237 (BT_FN_INT16_INT16_INT16): Likewise.
39238 (BT_FN_INT_INT_INT): Likewise.
39239 (BT_FN_UINT_FLOAT_UINT): Likewise.
39240 (BT_FN_FLOAT_UINT_UINT): Likewise.
39241 (BT_FN_ULONG_UINT_UINT): Likewise.
39242 (BT_FN_ULONG_UINT_PTR): Likewise.
39243 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39244 (BT_FN_UINT_UINT_UINT): Likewise.
39245 (BT_FN_VOID_UINT_PTR): Likewise.
39246 (BT_FN_UINT_UINT_PTR: Likewise.
39247 (BT_FN_UINT32_UINT64_PTR): Likewise.
39248 (BT_FN_INT_INT_UINT_UINT): Likewise.
39249 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39250 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39251 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39252 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39253 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39254 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39255 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39256 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39257 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39258 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39259 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39260 * doc/frontends.texi: List BRIG FE.
39261 * doc/install.texi (Testing): Add BRIG tesring requirements.
39262 * doc/invoke.texi (Overall Options): Mention BRIG.
39263 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39265 2017-01-24 Richard Biener <rguenther@suse.de>
39267 PR translation/79208
39268 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39270 2017-01-24 Martin Jambor <mjambor@suse.cz>
39273 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39274 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39275 and known_contexts.
39277 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39279 PR middle-end/79123
39280 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39281 casts from signed to unsigned really don't have a range.
39283 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39285 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39286 GMP_RNDx for compatiblity.
39288 2017-01-24 Martin Liska <mliska@suse.cz>
39291 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39292 that would prevent us to call alloca with -1 as argument.
39294 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39296 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39297 Avoid -Wformat-security warning.
39299 2017-01-23 Andrew Pinski <apinski@cavium.com>
39301 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39304 2017-01-23 Martin Sebor <msebor@redhat.com>
39306 PR middle-end/78703
39307 * gimple-ssa-sprintf.c (warn_level): New global.
39308 (format_integer): Use it here and throughout the rest of the file.
39309 Use the same switch to compute sign as base.
39310 (maybe_warn): New function.
39311 (format_directive): Factor out warnings into maybe_warn.
39312 Add debugging output. Use warn_level.
39313 (add_bytes): Use warn_level.
39314 (pass_sprintf_length::compute_format_length): Add debugging output.
39315 (try_substitute_return_value): Same.
39316 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39318 PR middle-end/78703
39319 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39320 (struct fmtresult, format_integer, format_floating): Adjust.
39321 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39322 (get_string_length, format_string,format_directive): Same.
39323 (pass_sprintf_length::compute_format_length): Same.
39324 (try_substitute_return_value): Simplify slightly.
39326 PR middle-end/78703
39327 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39328 (fmtresult::operator+=): Outlined.
39329 (struct fmtresult): Add ctors.
39330 (struct conversion_spec): Rename...
39331 (struct directive): ...to this. Add and remove data members.
39332 (directive::set_width, directive::set_precision): New functions.
39333 (format_percent): Use fmtresult ctor.
39334 (get_width_and_precision): Remove.
39335 (format_integer): Make naming changes. Avoid computing width and
39337 (format_floating): Same. Adjust indentation.
39338 (format_character, format_none): New functions.
39339 (format_string): Moved character handling to format_character.
39340 (format_directive): Remove arguments, change return type.
39341 (parse_directive): New function.
39342 (pass_sprintf_length::compute_format_length): Move directive
39343 parsing to parse_directive.
39345 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39347 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39348 (assign_assembler_name_if_needed): ... this.
39349 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39350 (assign_assembler_name_if_needed): ... this.
39351 (free_lang_data_in_cgraph): Adjust callers.
39352 * cgraphunit.c (cgraph_node::analyze): Likewise.
39353 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39356 2017-01-23 Richard Biener <rguenther@suse.de>
39358 PR tree-optimization/79088
39359 PR tree-optimization/79188
39360 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39361 resetting loop bounds after last path deletion. Reset loop
39362 bounds of the target loop, make code match the comments.
39363 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39364 Make sure loops need no fixups.
39366 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39368 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39369 exponent support with double type for first argument.
39370 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39371 type returned by __builtin_vec_extract_sig,
39372 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39373 functions from "vector int" to "vector unsigned int" or from
39374 "vector long long int" to "vector unsigned long long int".
39375 Changed type returned by __builtin_vec_extract_exp,
39376 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39377 functions from "vector int" to "vector unsigned int" or from
39378 "vector long long int" to "vector unsigned long long int".
39379 Changed return type of __builtin_vec_test_data_class,
39380 __builtin_vec_test_data_class_sp, and
39381 __builtin_vec_test_data_class_dp from "vector int" to
39382 "vector bool int" or from "vector long long int" to "vector bool
39383 long long int" and changed second argument type from "unsigned
39384 int" to "int". Added new overloaded function forms "vector float
39385 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39386 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39387 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39388 double, vector unsigned long long int)" and "vector double
39389 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39390 long int)". Changed return type of
39391 __builtin_scalar_test_data_class and
39392 __builtin_scalar_test_data_class_sp and
39393 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39394 int" and changed second argument from "unsigned int" to "int".
39395 Changed type returned by __builtin_scalar_test_neg,
39396 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39397 from "int" to "bool int". Added new overloaded function form
39398 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39399 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39400 exponent double-precision with floating point first argument.
39401 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39402 documentation of scalar_test_data_class, scalar_test_neg,
39403 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39404 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39405 vec_test_data_class built-in functions to reflect refinements in
39406 their type signatures.
39408 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39410 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39412 (aarch64_elf_asm_destructor): Likewise.
39414 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39416 PR rtl-optimization/78634
39417 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39418 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39419 * ifcvt.c (noce_try_cmove): Add missing cost check.
39421 PR rtl-optimization/71724
39422 * combine.c (if_then_else_cond): Look for situations where it is
39423 beneficial to undo the work of one of the recursive calls.
39425 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39427 PR tree-optimization/70754
39428 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39429 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39430 combined stmt before it if not NULL.
39431 (combine_chains): Process refs reversely and compute dominance point
39434 2017-01-23 Martin Liska <mliska@suse.cz>
39436 PR tree-optimization/79196
39437 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39438 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39440 (strlen_optimize_stmt): Call the renamed function.
39442 2017-01-23 Michael Matz <matz@suse.de>
39444 PR tree-optimization/78384
39445 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39447 2017-01-23 Richard Biener <rguenther@suse.de>
39449 PR tree-optimization/79186
39450 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39451 both incoming edges before moving an assert.
39453 2017-01-23 Martin Jambor <mjambor@suse.cz>
39455 * ipa-prop.c (load_from_param_1): Removed.
39456 (load_from_unmodified_param): Bits from load_from_param_1 put back
39458 (load_from_param): Removed.
39459 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39460 with stmt. Reverted back to use of load_from_unmodified_param.
39462 2017-01-23 Martin Jambor <mjambor@suse.cz>
39465 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39466 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39467 field a pointer to garbage collected vector, mark lattices and
39468 ipcp_orig_node with GTY((skip)).
39469 (ipa_get_param_count): Adjust to descriptors being a pointer.
39470 (ipa_get_param): Likewise.
39471 (ipa_get_type): Likewise.
39472 (ipa_get_param_move_cost): Likewise.
39473 (ipa_set_param_used): Likewise.
39474 (ipa_get_controlled_uses): Likewise.
39475 (ipa_set_controlled_uses): Likewise.
39476 (ipa_is_param_used): Likewise.
39477 (ipa_node_params_t): Move into garbage collector. New methods insert
39479 (ipa_node_params_sum): Annotate wth GTY(()).
39480 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39482 (ipa_load_from_parm_agg): Adjust declaration.
39483 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39484 * ipa-profile.c (ipa_profile): Likewise.
39485 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39486 (ipa_populate_param_decls): Make descriptors parameter garbage
39488 (ipa_dump_param): Adjust to descriptors being a pointer.
39489 (ipa_alloc_node_params): Likewise.
39490 (ipa_initialize_node_params): Likewise.
39491 (load_from_param_1): Make descriptors parameter garbage collected.
39492 (load_from_unmodified_param): Likewise.
39493 (load_from_param): Likewise.
39494 (ipa_load_from_parm_agg): Likewise.
39495 (ipa_node_params::~ipa_node_params): Removed.
39496 (ipa_free_all_node_params): Remove call to delete operator.
39497 (ipa_node_params_t::insert): New.
39498 (ipa_node_params_t::remove): Likewise.
39499 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39500 copy known_csts and known_contexts vectors.
39501 (ipa_read_node_info): Adjust to descriptors being a pointer.
39502 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39504 (ipcp_transform_function): Make descriptors variable garbage
39507 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39509 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39510 * config/i386/avx512dqintrin.h: Ditto.
39511 * config/i386/avx512fintrin.h: Ditto.
39512 * config/i386/i386.c: Handle new builtins.
39513 * config/i386/i386-builtin.def: Add new builtins.
39514 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39515 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39517 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39518 Martin Liska <mliska@suse.cz>
39520 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39521 * asan.c (asan_expand_poison_ifn): Support stores and use
39522 appropriate ASAN report function.
39523 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39524 * internal-fn.def (ASAN_POISON_USE): Declare.
39525 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39526 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39527 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39528 ASAN_POISON calls w/o LHS.
39529 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39530 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39531 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39532 * gimplify.c (asan_poison_variables): Add attribute
39533 use_after_scope_memory to variables that really needs to live
39535 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39536 having the attribute.
39538 2017-01-23 Martin Liska <mliska@suse.cz>
39540 * asan.c (create_asan_shadow_var): New function.
39541 (asan_expand_poison_ifn): Likewise.
39542 * asan.h (asan_expand_poison_ifn): New declaration.
39543 * internal-fn.c (expand_ASAN_POISON): Likewise.
39544 * internal-fn.def (ASAN_POISON): New builtin.
39545 * sanopt.c (pass_sanopt::execute): Expand
39546 asan_expand_poison_ifn.
39547 * tree-inline.c (copy_decl_for_dup_finish): Make function
39549 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39550 * tree-ssa.c (is_asan_mark_p): New function.
39551 (execute_update_addresses_taken): Rewrite local variables
39552 (identified just by use-after-scope as addressable) into SSA.
39554 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39556 * doc/install.texi (Specific): opensource.apple.com uses https
39557 now. Remove trailing slash.
39559 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39561 * README.Portability: Remove note on an Irix compatibility issue.
39563 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39565 * gcov.c (INCLUDE_ALGORITHM): Define.
39566 (INCLUDE_VECTOR): Define.
39567 No longer include <vector> and <algorithm> directly.
39569 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39571 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39573 * doc/invoke.texi (Code Gen Options): Ditto.
39575 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39578 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39580 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39582 rtl-optimization/79125
39583 * cprop.c (local_cprop_pass): Handle cases where we make an
39584 unconditional trap.
39586 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39590 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39591 read from, for big endian.
39593 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39595 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39596 register pauth builtins for LP64 only.
39598 2017-01-20 Marek Polacek <polacek@redhat.com>
39601 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39604 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39606 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39608 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39609 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39610 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39612 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39615 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39616 in big-endian mode when they are not a single duplicated value.
39618 2017-01-20 Richard Biener <rguenther@suse.de>
39620 * BASE-VER: Bump to 7.0.1.
39622 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39624 * omp-low.c (omplow_simd_context): New struct. Use it...
39625 (lower_rec_simd_input_clauses): ...here and...
39626 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39627 references to idx, lane, max_vf, is_simt.
39629 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39631 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39634 2017-01-20 Martin Jambor <mjambor@suse.cz>
39636 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39637 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39639 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39640 (GTFILES): Rename hsa.c to hsa-common.c.
39641 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39642 * hsa-dump.c: Likewise.
39643 * hsa-gen.c: Likewise.
39644 * hsa-regalloc.c: Likewise.
39645 * ipa-hsa.c: Likewise.
39646 * omp-expand.c: Likewise.
39647 * omp-low.c: Likewise.
39648 * toplev.c: Likewise.
39650 2017-01-20 Marek Polacek <polacek@redhat.com>
39653 * doc/invoke.texi: Document -Wduplicated-branches.
39654 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39655 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39656 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39657 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39658 return 0 only when not OEP_LEXICOGRAPHIC.
39659 (fold_build_cleanup_point_expr): Use the expression
39660 location when building CLEANUP_POINT_EXPR.
39661 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39662 * tree.c (add_expr): Handle error_mark_node.
39664 2017-01-20 Martin Liska <mliska@suse.cz>
39667 * tree-profile.c (init_ic_make_global_vars): Do not call
39669 (gimple_init_gcov_profiler): Likewise.
39671 2017-01-20 Martin Liska <mliska@suse.cz>
39674 * cgraph.h (maybe_create_reference): Remove argument and
39676 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39678 * ipa-cp.c (create_specialized_node): Likewise.
39679 * symtab.c (symtab_node::maybe_create_reference): Handle
39680 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39682 2017-01-20 Martin Liska <mliska@suse.cz>
39684 * read-rtl-function.c (function_reader::create_function): Use
39685 build_decl instread of build_decl_stat.
39687 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39689 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39690 * config/i386/avx512dqintrin.h: Ditto.
39691 * config/i386/avx512fintrin.h: Ditto.
39692 * config/i386/i386-builtin-types.def: Add new types.
39693 * config/i386/i386.c: Handle new types.
39694 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39695 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39696 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39697 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39698 (__builtin_ia32_kshiftridi): New.
39699 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39701 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39705 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39706 define to rs6000_init_stack_protect_guard.
39707 (rs6000_init_stack_protect_guard): New function.
39709 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39710 Yunqiang Su <yunqiang.su@imgtec.com>
39712 * config.gcc (supported_defaults): Add madd4.
39713 (with_madd4): Add validation.
39714 (all_defaults): Add madd4.
39715 * config/mips/mips.opt (mmadd4): New option.
39716 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39718 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39720 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39721 (ISA_HAS_FUSED_MADD4): Likewise.
39722 * doc/invoke.texi (-mmadd4): Document the new option.
39723 * doc/install.texi (--with-madd4): Document the new option.
39725 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39727 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39728 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39729 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39730 (aarch64_init_pauth_hint_builtins): New.
39731 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39732 (aarch64_expand_builtin): Expand new builtins.
39734 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39736 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39737 * combine-stack-adj.c (no_unhandled_cfa): Handle
39738 REG_CFA_TOGGLE_RA_MANGLE.
39739 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39740 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39741 info for return address signing.
39742 (aarch64_expand_epilogue): Likewise.
39744 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39746 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39747 * config/aarch64/aarch64-protos.h
39748 (aarch64_return_address_signing_enabled): New declaration.
39749 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39751 (aarch64_expand_prologue): Sign return address before it's pushed onto
39753 (aarch64_expand_epilogue): Authenticate return address fetched from
39755 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39756 (aarch64_attributes): New function attributes for "sign-return-address".
39757 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39758 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39759 ("*do_return"): Generate combined instructions according to key index.
39760 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39761 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39763 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39764 * config/aarch64/aarch64.opt (msign-return-address=): New.
39765 * doc/extend.texi (AArch64 Function Attributes): Documents
39766 "sign-return-address=".
39767 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39769 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39771 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39772 overall option summary.
39774 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39776 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39777 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39778 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39779 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39781 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39783 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39784 -mpower9-minmax by default for -mcpu=power9.
39785 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39786 128-bit floating point.
39788 2017-01-20 Alan Modra <amodra@gmail.com>
39790 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39791 optimizing for size.
39793 2017-01-20 Alan Modra <amodra@gmail.com>
39796 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39797 for strcmp and strncmp from corresponding builtin decl.
39799 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39801 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39802 instead of i386/rtems-64.h.
39803 * config/i386/rtems-64.h: Remove.
39805 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39809 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39811 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39813 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39815 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39816 Change int to HOST_WIDE_INT.
39817 * config/aarch64/aarch64-protos.h
39818 (aarch64_simd_gen_const_vector_dup): Likewise.
39819 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39821 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39823 * langhooks-def.h (lhd_type_for_size): New decl.
39824 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39825 * langhooks.c (lhd_type_for_size): New function, taken from
39828 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39830 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39831 define_bypass for CR latency.
39832 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39833 (power9-alu2): Add define_bypass for CR latency.
39835 (power9-mul): Update insn latency.
39836 (power9-mul-compare): Update insn latency, bypass latency and remove
39839 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39841 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39843 * config/aarch64/aarch64.md
39844 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39845 aarch64_nopcrelative_literal_loads.
39846 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39848 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39850 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39851 TARGET_LOONGSON_3A.
39852 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39854 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39857 * config.gcc (supported_defaults): Add lxc1-sxc1.
39858 (with_lxc1_sxc1): Add validation.
39859 (all_defaults): Add lxc1-sxc1.
39860 * config/mips/mips.opt (mlxc1-sxc1): New option.
39861 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39863 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39864 __mips_no_lxc1_sxc1.
39865 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39866 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39867 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39869 2017-01-19 Richard Biener <rguenther@suse.de>
39871 PR tree-optimization/72488
39872 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39873 sure to restore SSA info.
39874 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39876 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39878 PR rtl-optimization/79121
39879 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39880 of the inner type when shifting an extended value.
39882 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39885 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39886 interposable aliases.
39888 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39891 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39892 Use the evmergelohi instruction.
39893 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39894 (mov_sitf_e500_subreg8_2_be): Likewise.
39895 (mov_sitf_e500_subreg12_2_le): Likewise.
39896 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39897 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39898 (mov_sitf_e500_subreg8_2_le): Likewise.
39899 (mov_sitf_e500_subreg12_2_be): Likewise.
39901 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39903 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39904 attribute from vecsimple to vecperm.
39905 (altivec_vbpermq2): Likewise.
39907 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39910 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39912 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39913 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39914 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39915 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39916 case where N arg is SIZE_MAX.
39917 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39918 (cmpstrsi): Add pattern.
39920 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39922 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39923 __builtin_vec_revb builtins.
39924 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39925 built-in functions to support generation of the ISA 3.0 XXBR<x>
39926 vector byte reverse instructions.
39927 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39928 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39929 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39930 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39931 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39932 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39933 (P9V_BUILTIN_VEC_REVB): Likewise.
39934 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39935 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39936 (p9_xxbrq_v16qi): Likewise.
39937 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39938 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39939 (p9_xxbrh_v8hi): Likewise.
39940 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39941 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39942 vec_revb built-in functions.
39944 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39946 PR rtl-optimization/78952
39947 * config/i386/i386.md (any_extract): New code iterator.
39948 (*insvqi_2): Use any_extract for source operand.
39949 (*insvqi_3): Use any_shiftrt for source operand.
39951 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39953 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39955 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39957 2017-01-18 Matthias Klose <doko@ubuntu.com>
39959 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39961 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39963 * config/rs6000/altivec.h (vec_bperm): Change #define.
39964 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39965 (altivec_vbpermq2): New define_insn.
39966 (altivec_vbpermd): Likewise.
39967 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39968 function interface.
39969 (VBPERMD): Likewise.
39970 (VBPERM): New polymorphic function interface.
39971 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39972 Add entries for P9V_BUILTIN_VEC_VBPERM.
39973 * doc/extend.texi: Add interfaces for vec_bperm.
39975 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39977 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39978 first letter of error messages.
39979 (s390_resolve_overloaded_builtin): Likewise.
39980 * config/s390/s390.c (s390_expand_builtin): Likewise.
39981 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39982 (s390_valid_target_attribute_inner_p): Likewise.
39983 * config/s390/s390.md ("tabort"): Likewise.
39985 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39987 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39988 (ISA_AVOID_DIV_HILO): New macro.
39989 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39990 (ISA_HAS_DDIV): Likewise.
39992 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39994 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39996 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39998 * doc/invoke.texi (fabi-version): Spelling fix.
40000 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40003 * doc/invoke.texi (fabi-version): Mention mangling fix for
40006 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40009 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40011 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40014 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40015 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40017 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40019 (stack_protect_test): Ditto.
40020 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40021 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40023 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40024 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40025 -mstack-protector-guard-offset=.
40026 (RS/6000 and PowerPC Options): Ditto.
40028 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40030 * config/i386/i386.h (MASK_CLASS_P): New define.
40031 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40032 there are no registers from different register sets also when
40033 mask registers are used. Update function comment.
40034 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40035 to (*k/*r) and (*k/*km) alternatives.
40037 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40039 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40040 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40041 (EH_RETURN_HANDLER_RTX): New define.
40042 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40043 Force frame pointer in EH return functions.
40044 (aarch64_expand_epilogue): Add barrier for eh_return.
40045 (aarch64_final_eh_return_addr): Remove.
40046 (aarch64_eh_return_handler_rtx): New function.
40047 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40049 (aarch64_eh_return_handler_rtx): New prototype.
40051 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40053 * config/rs6000/altivec.h (vec_rlmi): New #define.
40054 (vec_vrlnm): Likewise.
40055 (vec_rlnm): Likewise.
40056 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40057 (UNSPEC_VRLNM): Likewise.
40058 (VIlong): New mode iterator.
40059 (altivec_vrl<VI_char>mi): New define_insn.
40060 (altivec_vrl<VI_char>nm): Likewise.
40061 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40063 (VRLDNM): Likewise.
40064 (RLNM): New polymorphic function entry.
40065 (VRLWMI): New monomorphic function entry.
40066 (VRLDMI): Likewise.
40067 (RLMI): New polymorphic function entry.
40068 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40069 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40070 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40073 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40076 * dwarf2out.c (field_byte_offset): Restore the
40077 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40078 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40079 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40082 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40085 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40087 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40090 * dwarf2out.c (add_data_member_location_attribute): For constant
40091 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40092 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40093 DW_AT_byte_size attributes.
40095 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40097 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40098 after forcing to constant memory when the code model is medium.
40100 2017-01-17 Julia Koval <julia.koval@intel.com>
40103 * config/i386/avx512fintrin.h
40104 (_mm512_i32gather_ps): Change __addr type to void const*.
40105 (_mm512_mask_i32gather_ps): Ditto.
40106 (_mm512_i32gather_pd): Ditto.
40107 (_mm512_mask_i32gather_pd): Ditto.
40108 (_mm512_i64gather_ps): Ditto.
40109 (_mm512_mask_i64gather_ps): Ditto.
40110 (_mm512_i64gather_pd): Ditto.
40111 (_mm512_mask_i64gather_pd): Ditto.
40112 (_mm512_i32gather_epi32): Ditto.
40113 (_mm512_mask_i32gather_epi32): Ditto.
40114 (_mm512_i32gather_epi64): Ditto.
40115 (_mm512_mask_i32gather_epi64): Ditto.
40116 (_mm512_i64gather_epi32): Ditto.
40117 (_mm512_mask_i64gather_epi32): Ditto.
40118 (_mm512_i64gather_epi64): Ditto.
40119 (_mm512_mask_i64gather_epi64): Ditto.
40120 (_mm512_i32scatter_ps): Change __addr type to void*.
40121 (_mm512_mask_i32scatter_ps): Ditto.
40122 (_mm512_i32scatter_pd): Ditto.
40123 (_mm512_mask_i32scatter_pd): Ditto.
40124 (_mm512_i64scatter_ps): Ditto.
40125 (_mm512_mask_i64scatter_ps): Ditto.
40126 (_mm512_i64scatter_pd): Ditto.
40127 (_mm512_mask_i64scatter_pd): Ditto.
40128 (_mm512_i32scatter_epi32): Ditto.
40129 (_mm512_mask_i32scatter_epi32): Ditto.
40130 (_mm512_i32scatter_epi64): Ditto.
40131 (_mm512_mask_i32scatter_epi64): Ditto.
40132 (_mm512_i64scatter_epi32): Ditto.
40133 (_mm512_mask_i64scatter_epi32): Ditto.
40134 (_mm512_i64scatter_epi64): Ditto.
40135 (_mm512_mask_i64scatter_epi64): Ditto.
40136 * config/i386/avx512pfintrin.h
40137 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40138 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40139 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40140 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40141 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40142 (_mm512_prefetch_i32scatter_ps): Ditto.
40143 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40144 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40145 (_mm512_prefetch_i64scatter_pd): Ditto.
40146 (_mm512_prefetch_i64scatter_ps): Ditto.
40147 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40148 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40149 * config/i386/avx512vlintrin.h
40150 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40151 (_mm_mmask_i32gather_ps): Ditto.
40152 (_mm256_mmask_i32gather_pd): Ditto.
40153 (_mm_mmask_i32gather_pd): Ditto.
40154 (_mm256_mmask_i64gather_ps): Ditto.
40155 (_mm_mmask_i64gather_ps): Ditto.
40156 (_mm256_mmask_i64gather_pd): Ditto.
40157 (_mm_mmask_i64gather_pd): Ditto.
40158 (_mm256_mmask_i32gather_epi32): Ditto.
40159 (_mm_mmask_i32gather_epi32): Ditto.
40160 (_mm256_mmask_i32gather_epi64): Ditto.
40161 (_mm_mmask_i32gather_epi64): Ditto.
40162 (_mm256_mmask_i64gather_epi32): Ditto.
40163 (_mm_mmask_i64gather_epi32): Ditto.
40164 (_mm256_mmask_i64gather_epi64): Ditto.
40165 (_mm_mmask_i64gather_epi64): Ditto.
40166 (_mm256_i32scatter_ps): Change __addr type to void*.
40167 (_mm256_mask_i32scatter_ps): Ditto.
40168 (_mm_i32scatter_ps): Ditto.
40169 (_mm_mask_i32scatter_ps): Ditto.
40170 (_mm256_i32scatter_pd): Ditto.
40171 (_mm256_mask_i32scatter_pd): Ditto.
40172 (_mm_i32scatter_pd): Ditto.
40173 (_mm_mask_i32scatter_pd): Ditto.
40174 (_mm256_i64scatter_ps): Ditto.
40175 (_mm256_mask_i64scatter_ps): Ditto.
40176 (_mm_i64scatter_ps): Ditto.
40177 (_mm_mask_i64scatter_ps): Ditto.
40178 (_mm256_i64scatter_pd): Ditto.
40179 (_mm256_mask_i64scatter_pd): Ditto.
40180 (_mm_i64scatter_pd): Ditto.
40181 (_mm_mask_i64scatter_pd): Ditto.
40182 (_mm256_i32scatter_epi32): Ditto.
40183 (_mm256_mask_i32scatter_epi32): Ditto.
40184 (_mm_i32scatter_epi32): Ditto.
40185 (_mm_mask_i32scatter_epi32): Ditto.
40186 (_mm256_i32scatter_epi64): Ditto.
40187 (_mm256_mask_i32scatter_epi64): Ditto.
40188 (_mm_i32scatter_epi64): Ditto.
40189 (_mm_mask_i32scatter_epi64): Ditto.
40190 (_mm256_i64scatter_epi32): Ditto.
40191 (_mm256_mask_i64scatter_epi32): Ditto.
40192 (_mm_i64scatter_epi32): Ditto.
40193 (_mm_mask_i64scatter_epi32): Ditto.
40194 (_mm256_i64scatter_epi64): Ditto.
40195 (_mm256_mask_i64scatter_epi64): Ditto.
40196 (_mm_i64scatter_epi64): Ditto.
40197 (_mm_mask_i64scatter_epi64): Ditto.
40198 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40199 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40200 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40201 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40202 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40203 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40204 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40205 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40206 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40207 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40208 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40209 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40210 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40211 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40212 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40213 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40214 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40215 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40216 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40217 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40218 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40219 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40220 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40221 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40222 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40223 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40224 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40225 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40226 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40227 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40228 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40229 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40230 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40231 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40232 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40233 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40234 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40235 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40236 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40237 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40238 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40239 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40240 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40241 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40242 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40243 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40244 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40245 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40246 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40247 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40248 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40249 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40250 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40251 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40252 definitions accordingly.
40254 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40255 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40258 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40261 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40264 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40265 hard regs for inner regno.
40267 2017-01-17 Martin Liska <mliska@suse.cz>
40270 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40271 assumption and add comment.
40273 2017-01-17 Nathan Sidwell <nathan@acm.org>
40275 * ipa-visibility.c (localize_node): New function, broken out of ...
40276 (function_and_variable_visibility): ... here. Call it.
40278 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40280 PR middle-end/77445
40281 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40282 correctly set frequency of oudgoing edge.
40283 (duplicate_thread_path): Fix profile updating.
40285 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40288 * configure.ac: Add GCC_BASE_VER.
40289 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40290 version from BASE-VER file.
40291 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40292 (gcc.o): Depend on $(BASEVER).
40293 * common.opt (dumpfullversion): New option.
40294 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40295 * doc/invoke.texi: Document -dumpfullversion.
40296 * doc/install.texi: Document --with-gcc-major-version-only.
40297 * configure: Regenerated.
40299 2017-01-17 Richard Biener <rguenther@suse.de>
40301 PR tree-optimization/71433
40302 * tree-vrp.c (register_new_assert_for): Merge same asserts
40303 on all incoming edges.
40304 (process_assert_insertions_for): Handle insertions at the
40307 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40309 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40310 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40312 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40315 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40318 2017-01-17 Alan Modra <amodra@gmail.com>
40321 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40322 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40323 symbolic stack limit when pic.
40325 2017-01-16 Martin Sebor <msebor@redhat.com>
40327 PR tree-optimization/78608
40328 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40330 2017-01-16 Jeff Law <law@redhat.com>
40333 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40334 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40335 for several include directories that may be relative to sysroot.
40336 * config/i386/x-mingw32 (gplus_includedir): Define.
40337 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40338 (native_system_includedir): Likewise.
40339 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40340 override if TARGET_SYSTEM_ROOT is defined.
40341 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40343 PR tree-optimization/79090
40344 PR tree-optimization/33562
40345 PR tree-optimization/61912
40346 PR tree-optimization/77485
40347 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40348 and computed trims into the dump file.
40350 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40352 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40354 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40357 * gimplify.c (gimplify_init_constructor): If want_value and
40358 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40362 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40363 sequence. Formatting fixes.
40364 (doloop_optimize): Formatting fixes.
40367 * gcc.c (debug_level_greater_than_spec_func): New function.
40368 (static_spec_functions): Add debug-level-gt spec function.
40369 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40371 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40372 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40373 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40374 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40375 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40376 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40378 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40380 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40381 QImode fixups to general and mask registers only.
40383 2017-01-16 Carl Love <cel@us.ibm.com>
40385 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40386 for built-in functions
40387 vector signed char vec_nabs (vector signed char)
40388 vector signed short vec_nabs (vector signed short)
40389 vector signed int vec_nabs (vector signed int)
40390 vector signed long long vec_nabs (vector signed long long)
40391 vector float vec_nabs (vector float)
40392 vector double vec_nabs (vector double)
40393 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40395 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40396 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40397 * doc/extend.texi: Update the documentation file for the new built-in
40400 2017-01-16 Martin Sebor <msebor@redhat.com>
40402 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40405 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40407 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40408 UNSPEC_VSX__XXSPLTD to require special splat handling.
40410 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40413 * system.h: Poison strndup.
40415 2017-01-16 Alan Modra <amodra@gmail.com>
40418 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40421 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40423 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40425 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40427 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40428 call recog here. Assert that INSN_CODE (insn) is non-negative.
40430 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40433 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40435 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40436 in the currently scheduled RTL fragment.
40438 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40440 PR rtl-optimization/78751
40441 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40444 2017-01-14 Jeff Law <law@redhat.com>
40446 PR tree-optimization/79090
40447 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40448 variable length stores.
40449 (compute_trims): Delete dead assignment to *trim_tail.
40450 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40453 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40455 PR rtl-optimization/78626
40456 PR rtl-optimization/78727
40457 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40458 of a block, and split such blocks after everything else is finished.
40460 2017-01-14 Alan Modra <amodra@gmail.com>
40463 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40464 target legitimate_combined_insn.
40465 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40466 (rs6000_legitimate_combined_insn): New function.
40467 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40469 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40470 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40471 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40473 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40475 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40477 2017-01-13 Jeff Law <law@redhat.com>
40479 PR tree-optimization/33562
40480 PR tree-optimization/61912
40481 PR tree-optimization/77485
40482 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40484 (delete_dead_assignment): Likewise.
40485 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40486 statement to delete_dead_call and delete_dead_assignment.
40488 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40491 * substring-locations.c (format_warning_va): Strengthen case 1 so
40492 that both endpoints of the substring must be within the format
40493 range for just the substring to be printed.
40495 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40497 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40498 * config/i386/i386.c (ix86_target_string): Add missing options
40499 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40500 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40501 flags_other and ix86_target_other to flags2_other. Display unknown
40503 (ix86_valid_target_attribute_inner_p): Add missing options and
40504 reorder options by implied ISAs, as in ix86_target_string.
40506 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40508 * hash-table.h (hash_table::too_empty_p): New function.
40509 (hash_table::expand): Use it.
40510 (hash_table::traverse): Likewise.
40511 (hash_table::empty_slot): Use sizeof (value_type) instead of
40512 sizeof (PTR) to convert bytes to elements. Shrink the table
40513 if the current size is excessive for the current number of
40516 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40518 * ira-costs.c (record_reg_classes): Break from the inner loop
40519 early once alt_fail is known to be true. Update outer loop
40520 handling accordingly.
40522 2017-01-13 Jeff Law <law@redhat.com>
40524 * tree-ssa-dse.c (decrement_count): New function.
40525 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40526 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40527 when we know the partially dead statement is a mem* function.
40529 PR tree-optimization/61912
40530 PR tree-optimization/77485
40531 * tree-ssa-dse.c: Include expr.h.
40532 (maybe_trim_constructor_store): New function.
40533 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40535 PR tree-optimization/33562
40536 PR tree-optimization/61912
40537 PR tree-optimization/77485
40538 * doc/invoke.texi: Document new dse-max-object-size param.
40539 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40540 * tree-ssa-dse.c: Include params.h.
40541 (dse_store_status): New enum.
40542 (initialize_ao_ref_for_dse): New, partially extracted from
40544 (valid_ao_ref_for_dse, normalize_ref): New.
40545 (setup_live_bytes_from_ref, compute_trims): Likewise.
40546 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40547 (maybe_trim_partially_dead_store): Likewise.
40548 (maybe_trim_complex_store): Likewise.
40549 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40550 Track what bytes live from the original store. Return tri-state
40551 for dead, partially dead or live.
40552 (dse_dom_walker): Add constructor, destructor and new private members.
40553 (delete_dead_call, delete_dead_assignment): New extracted from
40555 (dse_optimize_stmt): Make a member of dse_dom_walker.
40556 Use initialize_ao_ref_for_dse.
40558 PR tree-optimization/33562
40559 PR tree-optimization/61912
40560 PR tree-optimization/77485
40561 * sbitmap.h (bitmap_count_bits): Prototype.
40562 (bitmap_clear_range, bitmap_set_range): Likewise.
40563 * sbitmap.c (bitmap_clear_range): New function.
40564 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40566 2017-01-13 Martin Liska <mliska@suse.cz>
40569 * function.c (set_cfun): Add new argument force.
40570 * function.h (set_cfun): Likewise.
40571 * ipa-inline-transform.c (inline_call): Use the function when
40572 strict alising from is dropped for function we inline to.
40574 2017-01-13 Richard Biener <rguenther@suse.de>
40576 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40577 for dumping GIMPLE INTEGER_CSTs.
40579 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40581 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40582 to 201112L since C++17.
40584 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40587 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40588 if -fsanitize=kernel-address is present.
40590 2017-01-13 Richard Biener <rguenther@suse.de>
40592 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40593 as _Literal ( type ) number in case usual suffixes do not
40594 preserve all information.
40596 2017-01-13 Richard Biener <rguenther@suse.de>
40598 PR tree-optimization/77283
40599 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40600 and ssa-iterators.h.
40601 (is_feasible_trace): Implement a cost model based on joiner
40604 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40607 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40608 char or short to __float128/_Float128 directly.
40610 2017-01-12 Martin Sebor <msebor@redhat.com>
40612 to -Wformat-overflow.
40613 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40614 (min_bytes_remaining): Same.
40615 (get_string_length): Same.
40616 (format_string): Same.
40617 (format_directive): Same.
40619 (pass_sprintf_length::handle_gimple_call): Same.
40621 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40623 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40624 info.nowrite calls with no lhs that can't throw. Return bool
40625 whether gsi_remove has been called or not.
40626 (pass_sprintf_length::handle_gimple_call): Return bool whether
40627 try_substitute_return_value called gsi_remove. Formatting fix.
40628 (pass_sprintf_length::execute): Don't use gsi_remove if
40629 handle_gimple_call returned true.
40632 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40633 be removed due to side-effects, don't remove following barrier nor
40634 turn the successor edge into fallthru edge.
40636 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40639 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40640 element-reversing loads and stores as not swappable.
40642 2017-01-12 Nathan Sidwell <nathan@acm.org>
40643 Nicolai Stange <nicstange@gmail.com>
40645 * combine.c (try_combine): Don't ignore result of overlap checking
40646 loop. Combine overlap & asm check into single loop.
40648 2017-01-12 Richard Biener <rguenther@suse.de>
40650 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40651 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40653 2017-01-12 Richard Biener <rguenther@suse.de>
40655 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40656 and TS_TARGET_OPTION directly derive from TS_BASE.
40657 * tree-core.h (tree_optimization_option): Derive from tree_base.
40658 (tree_target_option): Likewise.
40660 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40662 * config/i386/i386.c (memory_address_length): Increase len
40663 only when rip_relative_addr_p returns false.
40665 2017-01-11 Julia Koval <julia.koval@intel.com>
40667 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40668 (OPTION_MASK_ISA_SGX_SET): New.
40669 (ix86_handle_option): Handle OPT_msgx.
40670 * config.gcc: Added sgxintrin.h.
40671 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40672 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40673 * config/i386/i386.c (ix86_target_string): Add -msgx.
40675 (ix86_option_override_internal): Handle new options.
40676 (ix86_valid_target_attribute_inner_p): Add sgx.
40677 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40678 * config/i386/i386.opt: Add msgx.
40679 * config/i386/sgxintrin.h: New file.
40680 * config/i386/x86intrin.h: Add sgxintrin.h.
40682 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40685 * fold-const.c (maybe_nonzero_address): Return 1 for function
40687 (tree_single_nonzero_warnv_p): Don't handle function local objects
40691 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40694 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40697 * opts.c: Include "spellcheck.h"
40698 (struct string_fragment): New struct.
40699 (struct edit_distance_traits<const string_fragment &>): New
40701 (get_closest_sanitizer_option): New function.
40702 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40704 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40706 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40708 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40709 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40710 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40711 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40712 for initial die_offset if dwarf_split_debug_info.
40713 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40714 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40716 (output_skeleton_debug_sections): Formatting fix. Use
40717 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40718 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40720 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40722 * config/arm/cortex-a53.md: Add bypasses for
40723 cortex_a53_r2f_cvt.
40724 (cortex_a53_r2f): Only use for transfers.
40725 (cortex_a53_f2r): Likewise.
40726 (cortex_a53_r2f_cvt): Add reservation for conversions.
40727 (cortex_a53_f2r_cvt): Likewise.
40729 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40731 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40732 to all inlined functions, change static to extern.
40734 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40737 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40739 (arm_assemble_integer): Likewise.
40741 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40743 * config.gcc: Use new awk script to check CPU, FPU and architecture
40744 parameters for --with-... options.
40745 * config/arm/parsecpu.awk: New file
40746 * config/arm/arm-cpus.in: New file.
40747 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40749 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40751 * config/arm/t-arm: Update dependency rules.
40752 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40753 of processing .def files.
40754 * config/arm/genopt.sh: Deleted.
40755 * config/arm/gentune.sh: Deleted.
40756 * config/arm/arm-cores.def: Deleted.
40757 * config/arm/arm-arches.def: Deleted.
40758 * config/arm/arm-fpus.def: Deleted.
40759 * config/arm/arm-tune.md: Regenerated.
40760 * config/arm/arm-tables.opt: Regenerated.
40761 * config/arm/arm-cpu.h: New generated file.
40762 * config/arm/arm-cpu-data.h: New generated file.
40763 * config/arm/arm-cpu-cdata.h: New generated file.
40765 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40768 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40770 (input_varpool_node): Unpack dynamically_initialized bit.
40772 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40774 PR rtl-optimization/79032
40775 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40776 the alignment of the adjusted memory reference against that of MODE,
40777 instead of the alignment of the original memory reference.
40779 2017-01-11 Martin Jambor <mjambor@suse.cz>
40781 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40783 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40784 decorated functions.
40786 2017-01-11 Richard Biener <rguenther@suse.de>
40788 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40789 set range/nonnull info for PHI results. Do not set it on
40790 stmts marked for removal.
40792 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40794 * expr.c (store_field): In the bitfield case, fetch the return value
40795 from the registers before applying a single big-endian adjustment.
40796 Always do a final load for a BLKmode value not larger than a word.
40798 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40801 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40802 that we correctly handle column numbers greater than
40803 LINE_MAP_MAX_COLUMN_NUMBER.
40805 2017-01-10 Martin Sebor <msebor@redhat.com>
40807 PR middle-end/78245
40808 * gimple-ssa-sprintf.c (get_destination_size): Call
40809 {init,fini}object_sizes.
40810 * tree-object-size.c (addr_object_size): Adjust.
40811 (pass_through_call): Adjust.
40812 (pass_object_sizes::execute): Adjust.
40813 * tree-object-size.h (fini_object_sizes): Declare.
40815 2017-01-10 Martin Sebor <msebor@redhat.com>
40817 PR tree-optimization/78775
40818 * builtins.c (get_size_range): Move...
40819 * calls.c: ...to here.
40820 (alloc_max_size): Accept zero argument.
40821 (operand_signed_p): Remove.
40822 (maybe_warn_alloc_args_overflow): Call get_size_range.
40823 * calls.h (get_size_range): Declare.
40825 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40827 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40828 from TI's devices.csv file as of September 2016.
40829 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40831 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40833 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40834 * doc/invoke.texi: Likewise.
40835 * doc/md.texi: Likewise.
40836 * doc/objc.texi: Likewise.
40838 2017-01-10 Joshua Conner <joshconner@google.com>
40840 * config/arm/fuchsia-elf.h: New file.
40841 * config/fuchsia.h: New file.
40842 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40843 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40845 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40847 2016-01-10 Richard Biener <rguenther@suse.de>
40849 PR tree-optimization/79034
40850 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40851 Propagate out degenerate PHIs in the joiner.
40853 2017-01-10 Martin Liska <mliska@suse.cz>
40855 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40856 (sort_congruence_classes_by_decl_uid): Likewise.
40857 (sort_congruence_class_groups_by_decl_uid): Likewise.
40858 (sem_item_optimizer::merge_classes): Sort class, groups in these
40859 classes and members in the groups by DECL_UID of declarations.
40860 This would make merge operations stable.
40862 2017-01-10 Martin Liska <mliska@suse.cz>
40864 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40865 usage of m_classes_vec.
40866 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40867 (sem_item_optimizer::get_group_by_hash): Likewise.
40868 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40869 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40870 (sem_item_optimizer::verify_classes): Likewise.
40871 (sem_item_optimizer::process_cong_reduction): Likewise.
40872 (sem_item_optimizer::dump_cong_classes): Likewise.
40873 (sem_item_optimizer::merge_classes): Likewise.
40874 * ipa-icf.h (congruence_class_hash): Rename from
40875 congruence_class_group_hash. Remove declaration of m_classes_vec.
40877 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40879 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40880 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40881 * config.gcc: Add avx512vpopcntdqintrin.h.
40882 * config/i386/avx512vpopcntdqintrin.h: New.
40883 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40884 * config/i386/i386-builtin-types.def: Add new types.
40885 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40886 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40887 __builtin_ia32_vpopcountq_v8di_mask): New.
40888 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40889 __AVX512VPOPCNTDQ__.
40890 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40891 (PTA_AVX512VPOPCNTDQ): Define.
40892 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40893 TARGET_AVX512VPOPCNTDQ_P): Define.
40894 * config/i386/i386.opt: Add mavx512vpopcntdq.
40895 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40896 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40898 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40900 PR middle-end/77484
40901 * predict.def (PRED_CALL): Set to 67.
40903 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40905 * expr.c (store_field): In the bitfield case, if the value comes from
40906 a function call and is of an aggregate type returned in registers, do
40907 not modify the field mode; extract the value in all cases if the mode
40908 is BLKmode and the size is not larger than a word.
40910 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40913 * config/i386/cpuid.h: Fix undefined behavior.
40915 2017-01-04 Jeff Law <law@redhat.com>
40917 PR tree-optimization/79007
40918 PR tree-optimization/67955
40919 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40920 conservative for pt.null when flag_non_call_exceptions is on.
40922 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40924 PR translation/79019
40925 PR translation/79020
40926 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40927 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40928 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40930 * config/avr/avr.opt (maccumulate-args): Likewise.
40931 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40932 * common.opt (freport-bug): Likewise.
40933 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40934 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40935 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40936 translatable string.
40937 * config/i386/i386.c (function_value_32): Likewise.
40938 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40939 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40941 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40942 * common/config/msp430/msp430-common.c (msp430_handle_option):
40944 * symtab.c (symtab_node::verify_base): Likewise.
40945 * opts.c (set_debug_level): Likewise.
40946 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40947 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40948 missing whitespace to translatable strings.
40949 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40950 * config/sh/superh.h: Likewise.
40951 * config/i386/xopintrin.h: Likewise.
40952 * config/i386/znver1.md: Likewise.
40953 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40954 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40955 * double-int.h (struct double_int): Likewise.
40956 * double-int.c (div_and_round_double): Likewise.
40957 * wide-int.cc: Likewise.
40958 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40959 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40960 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40961 (crossjumps_occurred): ... this.
40962 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40965 PR tree-optimization/78899
40966 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40967 returning bool return struct loop *, NULL for failure and the new
40969 (versionable_outer_loop_p): Don't version outer loop if it has
40970 dont_vectorized bit set.
40971 (tree_if_conversion): When versioning outer loop, ensure
40972 tree_if_conversion is performed also on the inner loop of the
40973 non-vectorizable outer loop copy.
40974 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40975 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40976 prevent vectorization of it.
40977 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40978 the outer loop vectorization of the non-scalar version is attempted
40979 before vectorization of the inner loop in scalar version. If
40980 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40981 vectorization of its inner loop.
40982 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40983 has 2 inner loops, rename also on edges from bb whose single pred
40984 is outer_loop->header. Fix typo in function comment.
40986 2017-01-09 Martin Sebor <msebor@redhat.com>
40989 * asan.c (asan_emit_stack_protection): Increase local buffer size
40990 to avoid snprintf truncation warning.
40992 2017-01-09 Andrew Pinski <apinski@cavium.com>
40994 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40995 to reference thunderx2t99 for the tuning structure
40996 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40998 (thunderx2t99_extra_costs): This.
40999 * config/aarch64/aarch64-tune.md: Regenerate.
41000 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41001 (vulcan_addrcost_table): This.
41002 (vulcan_regmove_cost): Rename to ...
41003 (thunderx2t99_regmove_cost): This.
41004 (vulcan_vector_cost): Rename to ...
41005 (thunderx2t99_vector_cost): this.
41006 (vulcan_branch_cost): Rename to ...
41007 (thunderx2t99_branch_cost): This.
41008 (vulcan_tunings): Rename to ...
41009 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41010 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41012 2017-01-09 Martin Jambor <mjambor@suse.cz>
41016 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41017 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41018 (propagate_vr_accross_jump_function): Use the above function for all
41019 value range computations for pass-through jump functions and type
41020 converasion from explicit value range values.
41021 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41022 parameters from TYPE_ARG_TYPES.
41023 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41024 (ipa_write_node_info): Stream type of the actual argument.
41025 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41027 2017-01-09 Martin Liska <mliska@suse.cz>
41030 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41031 (lookup_compiler): Do not show error message with have_E.
41033 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41035 PR tree-optimization/78938
41036 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41037 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41038 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41039 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41042 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41044 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41047 2017-01-09 Richard Biener <rguenther@suse.de>
41049 PR tree-optimization/78997
41050 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41051 name condition properly.
41053 2017-01-09 Richard Biener <rguenther@suse.de>
41056 * dwarf2out.c (is_cxx): New overload with context.
41057 (is_naming_typedef_decl): Use it.
41059 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41061 * invoke.texi (Option Summary): Correct spacing in option lists
41062 and add line breaks to fix over-long lines.
41064 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41066 PR middle-end/17660
41068 * extend.texi (Common Variable Attributes): Add xref to GCC
41069 Internals manual to explain mode attribute keywords.
41071 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41074 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41075 and Preprocessor Options.
41076 (Options for Linking): Document -pthread here....
41077 (RS/6000 and PowerPC Options): ...not here.
41078 (Solaris 2 Options): ...or here.
41079 * doc/cppopts.texi: Document -pthread.
41081 2017-01-08 Martin Sebor <msebor@redhat.com>
41083 PR middle-end/77708
41084 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41085 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41086 New member functions.
41087 (format_directive): Used them.
41089 (pass_sprintf_length::handle_gimple_call): Same.
41090 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41091 to avoid truncation for any argument.
41092 (extract_affine_mul): Same.
41093 * tree.c (get_file_function_name): Same.
41095 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41097 PR middle-end/77484
41098 * predict.def (PRED_INDIR_CALL): Set to 86.
41100 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41102 PR preprocessor/54124
41103 * doc/cppopts.texi: Reformat -d subtable to list the full name
41104 of the options. Add cross-reference to the docs for the general
41105 compiler -d options.
41106 * doc/invoke.texi (Developer Options): Add cross-reference to the
41107 preprocessor-specific -d option documentation.
41109 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41111 PR preprocessor/13498
41112 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41113 redudant material, and reflect new command-line options.
41114 (System Headers): Likewise.
41116 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41118 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41119 -isystem, and -idirafter. Copy-edit.
41120 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41121 default for -ftrack-macro-expansion. Delete obsolete and
41122 badly-formatted implementation details about -fdebug-cpp output.
41123 * doc/cppwarnopts.texi: Copy-edit.
41125 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41128 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41129 that the transition from a max line width >= 1<<10 to narrower
41130 lines works correctly.
41132 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41134 * doc/options.texi (PerFunction): New.
41135 * opt-functions.awk (switch_flags): Map both Optimization and
41136 PerFunction to CL_OPTIMIZATION.
41137 * opth-gen.awk: Test for PerFunction flag along with
41139 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41140 it only when the latter is present. Skip those that don't in
41141 the hash function generator.
41142 * common.opt (fvar-tracking): Mark as PerFunction instead of
41144 (fvar-tracking-assignments): Likewise.
41145 (fvar-tracking-assignments-toggle): Likewise.
41146 (fvar-tracking-uninit): Likewise.
41148 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41150 PR translation/79018
41151 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41154 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41157 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41158 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41159 TARGET_LONG_JUMP_TABLE_OFFSETS.
41160 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41161 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41162 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41163 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41164 * config/m68k/m68k.md (tablejump expander): Likewise.
41165 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41166 TARGET_LONG_JUMP_TABLE_OFFSETS.
41167 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41168 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41170 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41171 David Holsgrove <david.holsgrove@xilinx.com>
41173 * common/config/microblaze/microblaze-common.c
41174 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41175 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41177 * config/microblaze/microblaze.c (microblaze_must_save_register)
41178 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41180 (microblaze_eh_return): New function.
41181 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41182 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41183 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41184 * config/microblaze/microblaze.md (eh_return): New pattern.
41186 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41188 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41189 GCC_DIAGNOSTIC_STRINGIFY): Define.
41191 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41193 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41195 * config/arm/arm.md (<mcrr>): New.
41197 * config/arm/arm.c (arm_arch5te): New.
41198 (arm_option_override): Set arm_arch5te.
41199 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41201 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41202 (arm_mcrr_qualifiers): ... this. New.
41203 (MRRC_QUALIFIERS): Define to...
41204 (arm_mrrc_qualifiers): ... this. New.
41205 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41207 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41208 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41209 (MRRCI, mrrc, MRRC): New.
41210 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41211 VUNSPEC_MRRC2): New.
41213 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41215 * config/arm/arm.md (<mcr>): New.
41217 * config/arm/arm.c (arm_coproc_builtin_available): Add
41218 support for mcr, mrc, mcr2 and mrc2.
41219 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41220 (arm_mcr_qualifiers): ... this. New.
41221 (MRC_QUALIFIERS): Define to ...
41222 (arm_mrc_qualifiers): ... this. New.
41223 (MCR_QUALIFIERS): Define to ...
41224 (arm_mcr_qualifiers): ... this. New.
41225 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41227 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41228 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41229 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41230 VUNSPEC_MRC2): New.
41232 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41234 * config/arm/arm.md (*ldc): New.
41238 * config/arm/arm.c (arm_coproc_builtin_available): Add
41239 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41240 (arm_coproc_ldc_stc_legitimate_address): New.
41241 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41242 'qualifier_const_pointer'.
41243 (LDC_QUALIFIERS): Define to...
41244 (arm_ldc_qualifiers): ... this. New.
41245 (STC_QUALIFIERS): Define to...
41246 (arm_stc_qualifiers): ... this. New.
41247 * config/arm/arm-protos.h
41248 (arm_coproc_ldc_stc_legitimate_address): New.
41249 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41250 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41251 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41252 stc2, stcl, stc2l): New.
41253 * config/arm/constraints.md (Uz): New.
41254 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41255 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41256 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41257 VUNSPEC_STC2L): New.
41259 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41261 * config/arm/arm.md (<cdp>): New.
41262 * config/arm/arm.c (neon_const_bounds): Rename this ...
41263 (arm_const_bounds): ... this.
41264 (arm_coproc_builtin_available): New.
41265 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41266 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41267 (CDP_QUALIFIERS): Define to...
41268 (arm_cdp_qualifiers): ... this. New.
41270 (arm_expand_builtin_args): Add case for 6 arguments.
41271 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41272 (arm_const_bounds): ... this.
41273 (arm_coproc_builtin_available): New.
41274 * config/arm/arm_acle.h (__arm_cdp): New.
41276 * config/arm/arm_acle_builtins.def (cdp): New.
41278 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41279 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41280 'arm_const_bounds'.
41281 * config/arm/types.md (coproc): New.
41282 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41283 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41284 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41285 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41287 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41289 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41290 (UBINOP_QUALIFIERS): New.
41292 (acle_builtin_data): New. Change comment.
41293 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41294 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41295 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41296 arm_acle_builtins.def.
41297 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41298 (arm_init_acle_builtins): New.
41299 (CRC32_BUILTIN): Remove.
41300 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41301 crc32cb, crc32ch and crc32cw.
41302 (arm_init_crc32_builtins): Remove.
41303 (arm_init_builtins): Use arm_init_acle_builtins rather
41304 than arm_init_crc32_builtins.
41305 (arm_expand_acle_builtin): New.
41306 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41307 * config/arm/arm_acle_builtins.def: New.
41309 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41311 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41312 (arm_builtin_datum): ... this.
41313 (arm_init_neon_builtin): Rename to ...
41314 (arm_init_builtin): ... this. Add a new parameters PREFIX
41315 and USE_SIG_IN_NAME.
41316 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41317 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41318 'arm_builtin_datum'.
41319 (arm_init_vfp_builtins): Likewise.
41320 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41321 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41322 (arm_expand_neon_args): Rename to ...
41323 (arm_expand_builtin_args): ... this. Rename builtin_arg
41324 enum values and differentiate between ARG_BUILTIN_MEMORY
41325 and ARG_BUILTIN_NEON_MEMORY.
41326 (arm_expand_neon_builtin_1): Rename to ...
41327 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41328 values, arm_expand_builtin_args and add bool parameter NEON.
41329 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41330 (arm_expand_vfp_builtin): Likewise.
41331 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41333 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41335 PR middle-end/77484
41336 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41337 * predict.c (tree_estimate_probability_bb): Reverse direction of
41338 polymorphic call predictor.
41340 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41342 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41343 (determine_pass_name_match): ...this new function and...
41344 (should_skip_pass_p): ...this new function.
41346 2017-01-06 Nathan Sidwell <nathan@acm.org>
41348 * ipa-visibility.c (function_and_variable_visibility): Reformat
41349 comments and long lines. Remove extrneous if.
41350 * symtab.c (symtab_node::make_decl_local): Fix code format.
41351 (symtab_node::set_section_for_node): Fix comment typo.
41353 2017-01-06 Martin Liska <mliska@suse.cz>
41356 * lra-constraints.c: Rename invariant to lra_invariant.
41357 * predict.c (set_even_probabilities): Initialize e to NULL.
41359 2017-01-05 Martin Sebor <msebor@redhat.com>
41361 PR tree-optimization/78910
41362 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41363 (format_integer): Correct off-by-one error in the handling
41364 of precision with negative numbers in signed conversions..
41366 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41368 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41370 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41372 PR tree-optimization/71016
41373 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41374 factor_out_conditional_conversion. Formatting fix.
41375 (factor_out_conditional_conversion): Add cond_stmt argument.
41376 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41377 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41380 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41382 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41383 read-rtl-function.o, and selftest-rtl.o.
41384 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41385 (selftest::aarch64_test_loading_full_dump): New function.
41386 (selftest::aarch64_run_selftests): New function.
41387 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41388 selftest::aarch64_run_selftests.
41389 * config/i386/i386.c
41390 (selftest::ix86_test_loading_dump_fragment_1): New function.
41391 (selftest::ix86_test_loading_call_insn): New function.
41392 (selftest::ix86_test_loading_full_dump): New function.
41393 (selftest::ix86_test_loading_unspec): New function.
41394 (selftest::ix86_run_selftests): Call the new functions.
41395 * emit-rtl.c (maybe_set_max_label_num): New function.
41396 * emit-rtl.h (maybe_set_max_label_num): New decl.
41397 * function.c (instantiate_decls): Guard call to
41398 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41399 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41401 * gensupport.c (gen_reader::gen_reader): Pass "false"
41402 for new "compact" param of rtx_reader.
41403 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41404 rather than an empty string for NULL strings.
41405 * read-md.c: Potentially include config.h rather than bconfig.h.
41406 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41407 (have_error): New global, copied from errors.c.
41408 (md_reader::read_name): Rename to...
41409 (md_reader::read_name_1): ...this, adding "out_loc" param,
41410 and converting "missing name or number" to returning false, rather
41412 (md_reader::read_name): Reimplement in terms of read_name_1.
41413 (md_reader::read_name_or_nil): New function.
41414 (md_reader::read_string): Handle "(nil)" by returning NULL.
41415 (md_reader::md_reader): Add new param "compact".
41416 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41417 (md_reader::read_file): New method.
41418 * read-md.h (md_reader::md_reader): Add new param "compact".
41419 (md_reader::read_file): New method.
41420 (md_reader::is_compact): New accessor.
41421 (md_reader::read_name): Convert return type from void to file_location.
41422 (md_reader::read_name_or_nil): New decl.
41423 (md_reader::read_name_1): New decl.
41424 (md_reader::m_compact): New field.
41425 (noop_reader::noop_reader): Pass "false" for new "compact" param
41427 (rtx_reader::rtx_reader): Add new "compact" param.
41428 (rtx_reader::read_rtx_operand): Make virtual and convert return
41429 type from void to rtx.
41430 (rtx_reader::read_until): New decl.
41431 (rtx_reader::handle_any_trailing_information): New virtual function.
41432 (rtx_reader::postprocess): New virtual function.
41433 (rtx_reader::finalize_string): New virtual function.
41434 (rtx_reader::m_in_call_function_usage): New field.
41435 (rtx_reader::m_reuse_rtx_by_id): New field.
41436 * read-rtl-function.c: New file.
41437 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41438 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41439 (selftest::verify_three_block_rtl_cfg): New decl.
41440 * read-rtl-function.h: New file.
41441 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41442 For host, include function.h, memmodel.h, and emit-rtl.h.
41443 (one_time_initialization): New function.
41444 (struct compact_insn_name): New struct.
41445 (compact_insn_names): New array.
41446 (find_code): Handle insn codes in compact dumps.
41447 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41448 (bind_subst_iter_and_attr): Likewise.
41449 (add_condition_to_string): Likewise.
41450 (add_condition_to_rtx): Likewise.
41451 (apply_attribute_uses): Likewise.
41452 (add_current_iterators): Likewise.
41453 (apply_iterators): Likewise.
41454 (initialize_iterators): Guard usage of apply_subst_iterator with
41455 #ifdef GENERATOR_FILE.
41456 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41457 (md_reader::read_mapping): Likewise.
41458 (add_define_attr_for_define_subst): Likewise.
41459 (add_define_subst_attr): Likewise.
41460 (read_subst_mapping): Likewise.
41461 (check_code_iterator): Likewise.
41462 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41464 (one_time_initialization): New function.
41465 (rtx_reader::read_until): New method.
41466 (read_flags): New function.
41467 (parse_reg_note_name): New function.
41468 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41469 Handle reuse_rtx ids.
41470 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41471 Add parsing support for RTL dumps, mirroring the special-cases in
41472 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41473 values, and calling handle_any_trailing_information.
41474 (rtx_reader::read_rtx_operand): Convert return type from void
41475 to rtx, returning return_rtx. Handle case 'e'. Call
41476 finalize_string on XSTR and XTMPL fields.
41477 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41478 "(nil)" values were omitted. Call the postprocess vfunc on the
41480 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41481 class ctor. Initialize m_in_call_function_usage. Call
41482 one_time_initialization.
41483 * rtl-tests.c (selftest::test_uncond_jump): Call
41484 set_new_first_and_last_insn.
41485 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41486 * selftest-rtl.c: New file.
41487 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41488 (selftest::get_insn_by_uid): New decl.
41489 * selftest-run-tests.c (selftest::run_tests): Call
41490 read_rtl_function_c_tests.
41491 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41492 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41495 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41497 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41498 operands in a special way. Assert that pos+len <= mode precision.
41500 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41502 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41503 3 argument Alias with unlimited for the negative form.
41504 (fno-vect-cost-model): Removed.
41506 2017-01-05 Martin Liska <mliska@suse.cz>
41508 * hsa-gen.c (gen_hsa_divmod): New function.
41509 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41511 2017-01-05 Martin Liska <mliska@suse.cz>
41514 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41517 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41519 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41520 small constant length operands.
41522 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41524 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41525 between loop iterations.
41527 2017-01-05 Martin Liska <mliska@suse.cz>
41530 * gimplify.c (gimplify_decl_expr): Compare to
41531 asan_poisoned_variables instread of checking flags.
41532 (gimplify_target_expr): Likewise.
41533 (gimplify_expr): Likewise.
41534 (gimplify_function_tree): Conditionally initialize
41535 asan_poisoned_variables.
41537 2017-01-04 Jeff Law <law@redhat.com>
41539 PR tree-optimizatin/78812
41540 * rtl.h (contains_mem_rtx_p): Prototype.
41541 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41542 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41543 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41544 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41545 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41547 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41549 * input.c (assert_char_at_range): Default-initialize actual_range.
41551 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41553 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41554 to match the caller.
41556 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41558 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41559 insns after final jump in test to emit dummy move.
41561 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41563 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41564 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41566 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41568 * multiple_target.c (create_dispatcher_calls): Init e_next.
41569 * tree-ssa-loop-split.c (split_loop): Init border.
41570 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41573 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41578 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41579 (altivec_register_operand): Do not return true if the operand
41580 contains a SUBREG mixing SImode and SFmode.
41581 (vsx_register_operand): Likewise.
41582 (vsx_reg_sfsubreg_ok): New predicate.
41583 (vfloat_operand): Do not return true if the operand contains a
41584 SUBREG mixing SImode and SFmode.
41585 (vint_operand): Likewise.
41586 (vlogical_operand): Likewise.
41587 (gpc_reg_operand): Likewise.
41588 (int_reg_operand): Likewise.
41589 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41590 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41591 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41593 (rs6000_emit_move_si_sf_subreg): New helper function.
41594 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41595 fixup SUBREGs involving SImode and SFmode.
41596 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41597 numbers for the new peephole2 optimization.
41598 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41599 the GLIBC math library that do AND/IOR/XOR operations on single
41600 precision floating point.
41601 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41602 target macros to say whether we need to avoid SUBREGs mixing
41604 (TARGET_ALLOW_SF_SUBREG): Likewise.
41605 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41606 (UNSPEC_SI_FROM_SF): Likewise.
41607 (iorxor): Change spacing.
41608 (and_ior_xor): New iterator for AND, IOR, and XOR.
41609 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41610 (movdi_from_sf_zero_ext): Likewise.
41611 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41612 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41613 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41614 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41615 (fms<mode>4): Likewise.
41616 (fnma<mode>4): Likewise.
41617 (fnms<mode>4): Likewise.
41618 (nfma<mode>4): Likewise.
41619 (nfms<mode>4): Likewise.
41621 2017-01-04 Marek Polacek <polacek@redhat.com>
41624 * doc/invoke.texi: Document -Wpointer-compare.
41626 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41628 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41631 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41632 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41635 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41638 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41639 documentation of the powerpc_popcntb_ok attribute.
41640 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41641 code to issue warning messages if a requested CPU configuration is
41642 not supported by the binary (assembler and loader) toolchain.
41643 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41644 made to define a built-in function that has been disabled.
41645 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41646 made to define a built-in function that has been disabled.
41647 (altivec_init_builtins): Add comment explaining why definition
41648 of the DST built-in functions is not preceded by an assertion
41649 check. Add assertions to prevent ICE if attempts are made to
41650 define an altivec predicate or an abs* built-in function that has
41652 (htm_init_builtins): Add comment explaining why definition of the
41653 htm built-in functions is not preceded by an assertion check.
41655 2017-01-04 Jeff Law <law@redhat.com>
41657 PR tree-optimizatin/67955
41658 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41659 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41660 the points-to solution does not include pt_null. Use DECL_PT_UID
41663 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41665 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41666 Use gen_int_mode instead of gen_lopwart for const_int operands.
41668 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41670 PR tree-optimization/71563
41671 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41672 out of range value - has low bits known to be zero.
41674 2017-01-04 Alan Modra <amodra@gmail.com>
41676 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41677 * configure: Regenerate.
41678 * config.in: Regenerate.
41680 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41683 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41684 a substring of the message, but strcmp with the whole message. Ifdef
41685 ENABLE_NLS, translate the message first using dgettext.
41687 2017-01-03 Jeff Law <law@redhat.com>
41689 PR tree-optimizatin/78856
41690 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41691 (mark_threaded_blocks): Remove code to truncate thread paths that
41692 cross multiple loop headers. Instead invalidate the cached loop
41693 iteration information and handle case of a thread path walking
41694 into an irreducible region.
41696 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41699 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41700 assertions. Add support for doing the signbit if the IEEE 128-bit
41701 floating point value is in a GPR.
41702 * config/rs6000/rs6000.md (Fsignbit): Delete.
41703 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41704 Update the length attribute if the value is in a GPR.
41705 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41706 the sign or zero extension instruction, since the value is always 0/1.
41707 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41710 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41711 extracting SImode to a GPR register so that we can generate a
41712 store, limit the vector to be in a traditional Altivec register
41713 for the vextuwrx instruction.
41715 2017-01-03 Ian Lance Taylor <iant@google.com>
41717 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41719 2017-01-03 Martin Sebor <msebor@redhat.com>
41721 PR tree-optimization/78696
41722 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41723 precision. Use MPFR for %f for greater fidelity. Correct handling
41725 (pass_sprintf_length::compute_format_length): Set width and precision
41726 specified by asrerisk to void_node for vararg functions.
41727 (try_substitute_return_value): Adjust dump output.
41729 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41731 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41733 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41735 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41736 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41737 -mlra/-mno-lra was passed to the compiler.
41739 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41741 PR rtl-optimization/65618
41742 * emit-rtl.c (try_split): Move initialization of "before" and
41743 "after" to just before the call to emit_insn_after_setloc.
41745 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41747 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41749 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41751 * dwarf2out.c (gen_enumeration_type_die): When
41752 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41754 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41756 PR tree-optimization/78965
41757 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41758 Change first argument from const call_info & to call_info &. For %n
41759 set info.nowrite to false.
41761 PR middle-end/78901
41762 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41763 possibly throwing calls.
41765 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41766 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41767 and fns handling, rather than in a separate case SSA_NAME.
41769 2017-01-02 Jeff Law <law@redhat.com>
41771 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41772 fixes for first_period and second_period variables.
41774 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41777 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41778 (*insvqi_1): New insn pattern.
41779 (*insvqi_1_mem_rex64): Ditto.
41780 (*insvqi_2): Ditto.
41781 (*insvqi_3): Rename from *insvqi.
41783 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41785 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41787 * doc/cfg.texi (Edges): Remove reference to Java.
41788 (Maintaining the CFG): Ditto.
41790 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41792 PR middle-end/77674
41793 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41794 transparent aliases.
41796 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41798 PR middle-end/77484
41799 * predict.def (PRED_CALL): Update hitrate.
41800 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41801 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41802 into direct/indirect/polymorphic variants.
41804 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41806 Update copyright years.
41808 * gcc.c (process_command): Update copyright notice dates.
41809 * gcov-dump.c (print_version): Ditto.
41810 * gcov.c (print_version): Ditto.
41811 * gcov-tool.c (print_version): Ditto.
41812 * gengtype.c (create_file): Ditto.
41813 * doc/cpp.texi: Bump @copying's copyright year.
41814 * doc/cppinternals.texi: Ditto.
41815 * doc/gcc.texi: Ditto.
41816 * doc/gccint.texi: Ditto.
41817 * doc/gcov.texi: Ditto.
41818 * doc/install.texi: Ditto.
41819 * doc/invoke.texi: Ditto.
41821 Copyright (C) 2017 Free Software Foundation, Inc.
41823 Copying and distribution of this file, with or without modification,
41824 are permitted in any medium without royalty provided the copyright
41825 notice and this notice are preserved.