1 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
4 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
7 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
11 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
12 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
13 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
14 cond_taken_branch_cost 3->4.
16 2018-01-01 Jakub Jelinek <jakub@redhat.com>
18 PR tree-optimization/83581
19 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
20 TODO_cleanup_cfg if any changes have been made.
23 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
24 convert_modes if target mode has the right side, but different mode
28 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
29 last argument when extracting from CONCAT. If either from_real or
30 from_imag is NULL, use expansion through memory. If result is not
31 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
32 the parts directly to inner mode, if even that fails, use expansion
36 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
37 check for bswap in mode rather than HImode and use that in expand_unop
40 2017-12-31 Jakub Jelinek <jakub@redhat.com>
43 * config/i386/i386.c (ix86_attribute_table): Remove excess
44 initializer for "shared" attribute.
46 2017-12-30 Tom de Vries <tom@codesourcery.com>
49 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
51 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
52 offload_funcs entries that no longer have a corresponding cgraph_node.
53 Mark the remaining ones as DECL_PRESERVE_P.
54 (output_lto): Call prune_offload_funcs.
56 2017-12-30 Jakub Jelinek <jakub@redhat.com>
58 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
59 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
60 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
61 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
62 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
63 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
64 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
65 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
66 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
67 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
68 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
69 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
70 vaesenc_<mode>, vpclmulqdq_<mode>,
71 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
73 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
75 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
76 _Float<N> and _Float<N>X variants for rounding built-in
78 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
79 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
80 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
81 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
82 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
83 * builtins.c (mathfn_built_in_2): Likewise.
84 * internal-fn.def (CEIL): Likewise.
86 (NEARBYINT): Likewise.
90 * convert.c (convert_to_integer_1): Likewise.
91 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
92 (integer_valued_real_call_p): Likewise.
93 * fold-const-call.c (fold_const_call_ss): Likewise.
94 * gencfn-macros.c (print_case_cfn): Change CFN and operator
95 printers to take a const char * suffix instead of a bool.
96 (print_define_operator_list): Likewise.
97 (fltall_suffixes): New list of suffixes, that include the
98 traditional suffixes as well as all of the _Float<N> and
100 (main): For _Float<N> and _Float<N>X functions, emit both
101 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
102 has the _Float<N> and _Float<N>X case names or operators. The
103 <name>_ALL variant has both the traditional and the
104 _Float<N>/_Float<N>X case names or operators.
105 * match.pd (COPYSIGN optimizations): Provide optimizations for
106 _Float<N> and _Float<N>X types where possible.
107 (MIN/MAX optimizations): Likewise.
108 (sqrt optimizations): Likewise.
109 (rounding optimizations): Likewise.
111 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
113 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
114 gen_int_mode rather than GEN_INT.
115 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
117 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
119 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
120 instead of CONSTANT_P.
121 (gen_vec_series): Likewise.
122 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
124 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
126 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
129 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
131 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
132 (aarch64_ld1x2<VDC:mode>): Likewise.
133 (aarch64_simd_ld1<mode>_x2): Likewise.
134 (aarch64_simd_ld1<mode>_x2): Likewise.
135 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
136 (vld1_s8_x2): Likewise.
137 (vld1_u16_x2): Likewise.
138 (vld1_s16_x2): Likewise.
139 (vld1_u32_x2): Likewise.
140 (vld1_s32_x2): Likewise.
141 (vld1_u64_x2): Likewise.
142 (vld1_s64_x2): Likewise.
143 (vld1_f16_x2): Likewise.
144 (vld1_f32_x2): Likewise.
145 (vld1_f64_x2): Likewise.
146 (vld1_p8_x2): Likewise.
147 (vld1_p16_x2): Likewise.
148 (vld1_p64_x2): Likewise.
149 (vld1q_u8_x2): Likewise.
150 (vld1q_s8_x2): Likewise.
151 (vld1q_u16_x2): Likewise.
152 (vld1q_s16_x2): Likewise.
153 (vld1q_u32_x2): Likewise.
154 (vld1q_s32_x2): Likewise.
155 (vld1q_u64_x2): Likewise.
156 (vld1q_s64_x2): Likewise.
157 (vld1q_f16_x2): Likewise.
158 (vld1q_f32_x2): Likewise.
159 (vld1q_f64_x2): Likewise.
160 (vld1q_p8_x2): Likewise.
161 (vld1q_p16_x2): Likewise.
162 (vld1q_p64_x2): Likewise.
164 2017-12-27 Martin Liska <mliska@suse.cz>
166 PR tree-optimization/83552
167 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
168 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
170 2017-12-27 Tom de Vries <tom@codesourcery.com>
172 * config/nvptx/nvptx.c (nvptx_option_override): Disable
173 -gstatement-frontiers.
175 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
177 PR rtl-optimization/83513
178 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
179 before priority comparison.
181 2017-12-25 Jakub Jelinek <jakub@redhat.com>
184 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
185 ix86_isa_flags2 to ix86_isa_flags.
186 * config/i386/i386-c.c (ix86_target_macros_internal): Test
187 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
188 isa_flags rather than isa_flags2.
189 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
190 and -mavx512bitalg from isa2_opts to isa_opts.
191 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
192 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
193 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
195 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
196 OPTION_MASK_ISA_AVX512F ored with another option similarly to
197 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
198 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
199 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
200 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
201 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
202 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
203 builtins from bdesc_args2 to bdesc_args section.
204 (__builtin_ia32_compressstoreuqi512_mask,
205 __builtin_ia32_compressstoreuhi512_mask,
206 __builtin_ia32_compressstoreuqi256_mask,
207 __builtin_ia32_expandloadqi512_mask,
208 __builtin_ia32_expandloadqi512_maskz,
209 __builtin_ia32_expandloadhi512_mask,
210 __builtin_ia32_expandloadhi512_maskz,
211 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
212 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
213 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
214 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
215 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
216 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
217 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
218 __builtin_ia32_vpshldv_v32hi_maskz,
219 __builtin_ia32_vpopcountb_v64qi_mask,
220 __builtin_ia32_vpopcountw_v32hi_mask,
221 __builtin_ia32_vpshufbitqmb512_mask,
222 __builtin_ia32_vpshufbitqmb256_mask): Add
223 " | OPTION_MASK_ISA_AVX512BW".
224 (__builtin_ia32_expandloadqi256_mask,
225 __builtin_ia32_expandloadqi256_maskz,
226 __builtin_ia32_vpopcountb_v32qi_mask): Add
227 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
228 (__builtin_ia32_expandloadhi256_mask,
229 __builtin_ia32_expandloadhi256_maskz,
230 __builtin_ia32_expandloadqi128_mask,
231 __builtin_ia32_expandloadqi128_maskz,
232 __builtin_ia32_expandloadhi128_mask,
233 __builtin_ia32_expandloadhi128_maskz,
234 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
235 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
236 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
237 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
238 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
239 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
240 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
241 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
242 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
243 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
244 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
245 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
246 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
247 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
248 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
249 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
250 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
251 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
252 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
253 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
254 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
255 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
256 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
257 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
258 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
259 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
260 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
261 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
262 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
263 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
264 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
265 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
266 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
267 __builtin_ia32_vpopcountw_v8hi_mask): Add
268 " | OPTION_MASK_ISA_AVX512VL".
269 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
270 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
271 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
272 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
273 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
274 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
275 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
276 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
277 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
278 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
279 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
280 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
281 _mm_mask_bitshuffle_epi64_mask): Likewise.
282 * common/config/i386/i386-common.c
283 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
284 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
285 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
286 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
287 OPTION_MASK_ISA_AVX512BITALG_UNSET.
288 (OPTION_MASK_ISA2_AVX512F_UNSET,
289 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
290 (ix86_handle_option): For -mno-general-regs-only, clear from
291 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
292 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
293 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
294 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
295 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
296 for moving from ix86_isa_flags2 to ix86_isa_flags.
298 2017-12-23 Jakub Jelinek <jakub@redhat.com>
301 * fold-const.c (struct contains_label_data): New type.
302 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
303 inside of a SWITCH_BODY seen during the walk.
304 (contains_label_p): Use walk_tree instead of
305 walk_tree_without_duplicates, prepare data for contains_label_1 and
308 2017-12-22 Jakub Jelinek <jakub@redhat.com>
311 * tree-iterator.c (alloc_stmt_list): Start with cleared
312 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
316 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
317 enable avx512f explicitly in #pragma GCC target.
318 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
319 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
320 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
321 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
322 __builtin_ia32_vpdpbusds_v8si_mask,
323 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
324 __builtin_ia32_vpdpbusds_v4si_mask,
325 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
326 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
327 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
328 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
329 __builtin_ia32_vpdpwssds_v8si_mask,
330 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
331 __builtin_ia32_vpdpwssds_v4si_mask,
332 __builtin_ia32_vpdpwssds_v4si_maskz): Use
333 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
334 just OPTION_MASK_ISA_AVX512VNNI.
336 2017-12-22 Martin Jambor <mjambor@suse.cz>
339 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
342 2017-12-22 Julia Koval <julia.koval@intel.com>
343 Sebastian Peryt <sebastian.peryt@intel.com>
345 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
346 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
347 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
348 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
349 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
350 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
351 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
352 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
353 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
354 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
355 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
356 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
357 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
358 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
359 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
360 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
361 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
362 _mm256_maskz_popcnt_epi64): New intrinsics.
363 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
364 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
365 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
366 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
367 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
368 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
369 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
370 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
371 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
372 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
373 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
374 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
375 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
376 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
377 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
378 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
379 __builtin_ia32_vpshufbitqmb256_mask,
380 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
381 * config/i386/i386-c.c (__AVX512BITALG__): New.
382 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
383 (ix86_valid_target_attribute_inner_p): Ditto.
384 (ix86_expand_args_builtin): Handle new types.
385 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
386 * config/i386/i386.opt: Add -mavx512bitalg.
387 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
388 avx512bitalgintrin.h.
389 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
390 (vpopcount<mode><mask_name>): Add more types.
391 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
392 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
394 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
396 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
397 Or in OPTION_MASK_ISA_AVX512F_SET.
398 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
399 OPTION_MASK_ISA_AVX512VNNI_UNSET.
400 (ix86_handle_option): Adjust for
401 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
402 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
404 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
405 isa_flag instead of isa_flag2.
406 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
407 isa_opts2 to isa_opts.
408 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
411 2017-12-22 Mike Stump <mikestump@comcast.net>
412 Eric Botcazou <ebotcazou@adacore.com>
414 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
416 2017-12-21 Jakub Jelinek <jakub@redhat.com>
419 * config/i386/i386.c (ix86_function_arg_boundary): Return
420 PARM_BOUNDARY for TYPE_EMPTY_P types.
423 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
424 if navail is >= dir.len.
426 2017-12-21 Steve Ellcey <sellcey@cavium.com>
428 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
431 2017-12-21 Jakub Jelinek <jakub@redhat.com>
433 PR rtl-optimization/80747
434 PR rtl-optimization/83512
435 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
436 succ edge from ENTRY, copy partition from e->dest to the newly
438 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
439 ENTRY, use BB_PARTITION of its successor block as current_partition.
440 Don't copy partition when splitting succ edge from ENTRY.
442 PR tree-optimization/83523
443 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
444 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
445 (convert_mult_to_fma): Likewise.
447 PR tree-optimization/83521
448 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
449 gimple_build_assign without code on result of
450 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
453 2017-12-21 Andrew Pinski <apinski@cavium.com>
454 Steve Ellcey <sellcey@cavium.com>
456 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
457 multi-arch for ilp32.
459 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
462 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
463 constraints to operand 2.
464 (*ashl<mode>3_mask_1): Ditto.
465 (*<shift_insn><mode>3_mask): Ditto.
466 (*<shift_insn><mode>3_mask_1): Ditto.
467 (*<rotate_insn><mode>3_mask): Ditto.
468 (*<rotate_insn><mode>3_mask_1): Ditto.
470 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
472 * reorg.c (make_return_insns): Reemit each insn with its own location.
474 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
476 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
477 generation for cases where splatting a value is not useful.
478 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
479 across a vec_duplicate and a paradoxical subreg forming a vector
480 mode to a vec_concat.
482 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
484 * combine.c (simplify_set): Do not transform subregs to zero_extends
485 if the destination is not a scalar int mode.
487 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
490 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
491 shared zero if the input has overflowed.
493 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
495 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
496 support for Cortex-A55 and Cortex-A75.
498 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
500 * common/config/arm/arm-common.c (compare_opt_names): Add function
501 comment. Use strcmp instead of manual loop.
503 2017-12-21 Martin Liska <mliska@suse.cz>
505 PR gcov-profile/83509
506 * gcov-dump.c (dump_gcov_file): Do not read info about
507 support_unexecuted_blocks for gcda files.
509 2017-12-21 Jakub Jelinek <jakub@redhat.com>
511 PR rtl-optimization/82973
512 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
513 (valid_for_const_vector_p): ... this.
514 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
515 (valid_for_const_vector_p): ... this. Adjust function comment.
516 (gen_vec_duplicate): Adjust caller.
517 * optabs.c (expand_vector_broadcast): Likewise.
518 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
519 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
521 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
522 instead of GET_CODE == CONST_FIXED.
523 (simplify_subreg): Use CONST_FIXED_P macro instead of
524 GET_CODE == CONST_FIXED.
527 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
528 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
529 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
530 (ix86_option_override_internal): Adjust for
531 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
532 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
533 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
534 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
535 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
536 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
537 def_builtin2 for CET builtins.
538 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
539 up formatting in IX86_BUILTIN_RDPID code.
540 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
541 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
542 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
543 ix86_isa_flags2 to ix86_isa_flags.
544 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
546 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
547 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
548 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
550 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
551 Or in OPTION_MASK_ISA_AVX512F_SET.
552 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
553 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
554 (ix86_handle_option): Adjust for
555 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
556 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
558 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
559 Alan Hayward <alan.hayward@arm.com>
560 David Sherwood <david.sherwood@arm.com>
562 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
563 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
564 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
565 Track polynomial offsets.
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-vect-data-refs.c (vect_compute_data_ref_alignment):
572 Treat drb->init as a poly_int. Fail if its misalignment wrt
573 vector_alignment isn't known.
575 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
576 Alan Hayward <alan.hayward@arm.com>
577 David Sherwood <david.sherwood@arm.com>
579 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
581 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
582 (vect_loop_versioning): Take the loop versioning threshold as a
584 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
585 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
586 versioning_threshold.
587 (vect_analyze_loop_2): Compute the loop versioning threshold
588 whenever loop versioning is needed, and store it in the new
589 field rather than combining it with the cost model threshold.
590 (vect_transform_loop): Update call to vect_loop_versioning.
591 Try to combine the loop versioning and cost thresholds here.
593 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
594 Alan Hayward <alan.hayward@arm.com>
595 David Sherwood <david.sherwood@arm.com>
597 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
598 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
599 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
600 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
601 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
602 (group_compare_offset): Update accordingly.
603 (split_small_address_groups_p): Likewise.
604 (record_use): Take addr_offset as a poly_uint64 rather than
605 an unsigned HOST_WIDE_INT.
606 (strip_offset): Return the offset as a poly_uint64 rather than
607 an unsigned HOST_WIDE_INT.
608 (record_group_use, split_address_groups): Track polynomial offsets.
609 (add_iv_candidate_for_use): Likewise.
610 (addr_offset_valid_p): Take the offset as a poly_int64 rather
611 than a HOST_WIDE_INT.
612 (strip_offset_1): Return the offset as a poly_int64 rather than
615 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
616 Alan Hayward <alan.hayward@arm.com>
617 David Sherwood <david.sherwood@arm.com>
619 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
620 rather than a HOST_WIDE_INT.
621 * tree.c (get_binfo_at_offset): Likewise.
623 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
624 Alan Hayward <alan.hayward@arm.com>
625 David Sherwood <david.sherwood@arm.com>
627 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
628 rather than a HOST_WIDE_INT.
629 * tree-sra.c (build_ref_for_offset): Likewise.
631 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
632 Alan Hayward <alan.hayward@arm.com>
633 David Sherwood <david.sherwood@arm.com>
635 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
637 * tree.c (mem_ref_offset): Likewise.
638 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
639 * builtins.c (get_object_alignment_2): Likewise.
640 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
641 * gimple-fold.c (get_base_constructor): Likewise.
642 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
643 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
645 * ipa-polymorphic-call.c
646 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
647 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
648 (get_ancestor_addr_info): Likewise.
649 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
650 * match.pd: Likewise.
651 * tree-data-ref.c (dr_analyze_innermost): Likewise.
652 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
653 * tree-eh.c (tree_could_trap_p): Likewise.
654 * tree-object-size.c (addr_object_size): Likewise.
655 * tree-ssa-address.c (copy_ref_info): Likewise.
656 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
657 (indirect_refs_may_alias_p): Likewise.
658 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
659 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
660 (non_rewritable_mem_ref_base): Likewise.
661 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
662 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
663 * varasm.c (decode_addr_const): Likewise.
665 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
666 Alan Hayward <alan.hayward@arm.com>
667 David Sherwood <david.sherwood@arm.com>
669 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
670 offsets for MEM_REFs.
672 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
673 Alan Hayward <alan.hayward@arm.com>
674 David Sherwood <david.sherwood@arm.com>
676 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
677 as a poly_uint64 rather than an unsigned int.
678 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
680 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
681 Alan Hayward <alan.hayward@arm.com>
682 David Sherwood <david.sherwood@arm.com>
684 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
686 (decode_addr_const): Update accordingly.
688 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
689 Alan Hayward <alan.hayward@arm.com>
690 David Sherwood <david.sherwood@arm.com>
692 * tree.h (bit_field_size, bit_field_offset): New functions.
693 * hsa-gen.c (gen_hsa_addr): Use them.
694 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
695 (simplify_vector_constructor): Likewise.
696 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
697 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
698 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
699 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
702 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
703 Alan Hayward <alan.hayward@arm.com>
704 David Sherwood <david.sherwood@arm.com>
706 * expr.h (emit_group_load, emit_group_load_into_temps)
707 (emit_group_store): Take the size as a poly_int64 rather than an int.
708 * expr.c (emit_group_load_1, emit_group_load): Likewise.
709 (emit_group_load_into_temp, emit_group_store): Likewise.
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 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
716 an unsigned int to a poly_uint64.
717 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
718 sizes as poly_uint64s rather than unsigned ints.
719 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
722 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
723 Alan Hayward <alan.hayward@arm.com>
724 David Sherwood <david.sherwood@arm.com>
726 * lra-constraints.c (emit_inc): Change inc_amount from an int
729 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
730 Alan Hayward <alan.hayward@arm.com>
731 David Sherwood <david.sherwood@arm.com>
733 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
735 (add_stack_var, stack_var_cmp, partition_stack_vars)
736 (dump_stack_var_partition): Update accordingly.
737 (alloc_stack_frame_space): Take the size as a poly_int64 rather
738 than a HOST_WIDE_INT.
739 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
740 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
741 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
742 rather than a HOST_WIDE_INT.
744 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
745 Alan Hayward <alan.hayward@arm.com>
746 David Sherwood <david.sherwood@arm.com>
748 * target.def (return_pops_args): Treat both the input and output
749 sizes as poly_int64s rather than HOST_WIDE_INTS.
750 * targhooks.h (default_return_pops_args): Update accordingly.
751 * targhooks.c (default_return_pops_args): Likewise.
752 * doc/tm.texi: Regenerate.
753 * emit-rtl.h (incoming_args): Change pops_args, size and
754 outgoing_args_size from int to poly_int64_pod.
755 * function.h (expr_status): Change x_pending_stack_adjust and
756 x_stack_pointer_delta from int to poly_int64.
757 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
758 (ARGS_SIZE_RTX): Update accordingly.
759 * calls.c (highest_outgoing_arg_in_use): Change from int to
761 (stack_usage_watermark, stored_args_watermark): New variables.
762 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
763 (emit_call_1): Change the stack_size and rounded_stack_size
764 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
766 (save_fixed_argument_area): Check stack_usage_watermark.
767 (initialize_argument_information): Change old_pending_adj from
768 a HOST_WIDE_INT * to a poly_int64_pod *.
769 (compute_argument_block_size): Return the size as a poly_int64
771 (finalize_must_preallocate): Track polynomial argument sizes.
772 (compute_argument_addresses): Likewise.
773 (internal_arg_pointer_based_exp): Track polynomial offsets.
774 (mem_overlaps_already_clobbered_arg_p): Rename to...
775 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
776 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
777 Check stored_args_used_watermark.
778 (load_register_parameters): Update accordingly.
779 (check_sibcall_argument_overlap_1): Likewise.
780 (combine_pending_stack_adjustment_and_call): Take the unadjusted
781 args size as a poly_int64 rather than an int. Return a bool
782 indicating whether the optimization was possible and return
783 the new adjustment by reference.
784 (check_sibcall_argument_overlap): Track polynomail argument sizes.
785 Update stored_args_watermark.
786 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
787 (expand_call): Likewise. Maintain stack_usage_watermark and
788 stored_args_watermark. Update calls to
789 combine_pending_stack_adjustment_and_call.
790 (emit_library_call_value_1): Handle polynomial argument sizes.
791 Call stack_region_maybe_used_p and mark_stack_region_used.
792 Maintain stack_usage_watermark.
793 (store_one_arg): Likewise. Update call to
794 mem_overlaps_already_clobbered_arg_p.
795 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
797 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
798 * config/microblaze/microblaze.c (microblaze_function_prologue):
800 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
801 TARGET_RETURN_POPS_ARGS interface.
802 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
804 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
805 * config/i386/i386.c (ix86_return_pops_args): Update for new
806 TARGET_RETURN_POPS_ARGS interface.
807 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
808 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
809 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
810 TARGET_RETURN_POPS_ARGS interface.
811 * config/vax/vax.c (vax_return_pops_args): Likewise.
812 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
813 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
814 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
815 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
816 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
817 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
818 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
819 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
820 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
821 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
822 and x_stack_pointer_delta from int to poly_int64.
823 * dojump.c (do_pending_stack_adjust): Update accordingly.
824 * explow.c (allocate_dynamic_stack_space): Handle polynomial
825 stack_pointer_deltas.
826 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
827 (pad_to_arg_alignment): Track polynomial offsets.
828 (assign_parm_find_stack_rtl): Likewise.
829 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
830 * toplev.c (output_stack_usage): Update reference to
831 current_function_pushed_stack_size.
833 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
834 Alan Hayward <alan.hayward@arm.com>
835 David Sherwood <david.sherwood@arm.com>
837 * function.c (in_arg_offset, var_offset, dynamic_offset)
838 (out_arg_offset, cfa_offset): Change from int to poly_int64.
839 (instantiate_new_reg): Return the new offset as a poly_int64_pod
840 rather than a HOST_WIDE_INT.
841 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
842 (instantiate_virtual_regs_in_insn): Likewise.
844 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
845 Alan Hayward <alan.hayward@arm.com>
846 David Sherwood <david.sherwood@arm.com>
848 * rtl.h (get_args_size, add_args_size_note): New functions.
849 (find_args_size_adjust): Return a poly_int64 rather than a
851 (fixup_args_size_notes): Likewise. Make the same change to the
852 end_args_size parameter.
853 * rtlanal.c (get_args_size, add_args_size_note): New functions.
854 * builtins.c (expand_builtin_trap): Use add_args_size_note.
855 * calls.c (emit_call_1): Likewise.
856 * explow.c (adjust_stack_1): Likewise.
857 * cfgcleanup.c (old_insns_match_p): Update use of
858 find_args_size_adjust.
859 * combine.c (distribute_notes): Track polynomial arg sizes.
860 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
861 end_true_args_size, beg_delay_args_size and end_delay_args_size
862 from HOST_WIDE_INT to poly_int64.
863 (add_cfi_args_size): Take the args_size as a poly_int64 rather
864 than a HOST_WIDE_INT.
865 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
866 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
867 polynomial arg sizes.
868 * emit-rtl.c (try_split): Use get_args_size.
869 * recog.c (peep2_attempt): Likewise.
870 * reload1.c (reload_as_needed): Likewise.
871 * expr.c (find_args_size_adjust): Return the adjustment as a
872 poly_int64 rather than a HOST_WIDE_INT.
873 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
874 to a poly_int64 and change the return type in the same way.
875 (emit_single_push_insn): Track polynomial arg sizes.
877 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
878 Alan Hayward <alan.hayward@arm.com>
879 David Sherwood <david.sherwood@arm.com>
881 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
882 from HOST_WIDE_INT to poly_int64.
883 * expr.c (push_block, emit_push_insn): Likewise.
885 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
886 Alan Hayward <alan.hayward@arm.com>
887 David Sherwood <david.sherwood@arm.com>
889 * function.h (frame_space): Change start and length from HOST_WIDE_INT
891 (get_frame_size): Return the size as a poly_int64 rather than a
893 (frame_offset_overflow): Take the offset as a poly_int64 rather
894 than a HOST_WIDE_INT.
895 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
896 (assign_stack_temp): Likewise for the size.
897 * function.c (get_frame_size): Return a poly_int64 rather than
899 (frame_offset_overflow): Take the offset as a poly_int64 rather
900 than a HOST_WIDE_INT.
901 (try_fit_stack_local): Take the start, length and size as poly_int64s
902 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
903 rather than a HOST_WIDE_INT.
904 (add_frame_space): Take the start and end as poly_int64s rather than
906 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
907 (assign_stack_temp): Likewise for the size.
908 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
910 (find_temp_slot_from_address): Handle polynomial offsets.
911 (combine_temp_slots): Likewise.
912 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
914 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
915 poly_int64 rather than a HOST_WIDE_INT.
916 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
917 than a HOST_WIDE_INT.
918 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
919 polynomial frame offsets.
920 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
921 as a poly_int64 rather than an int.
922 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
923 * config/v850/v850-protos.h (compute_frame_size): Likewise.
924 * config/v850/v850.c (compute_frame_size): Likewise.
925 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
926 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
927 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
928 * config/pa/pa.c (pa_compute_frame_size): Likewise.
929 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
930 rather than a HOST_WIDE_INT.
931 * explow.c (get_dynamic_stack_base): Likewise.
932 * final.c (final_start_function): Use the constant lower bound
933 of the frame size for -Wframe-larger-than.
934 * ira.c (do_reload): Adjust for new get_frame_size return type.
935 * lra.c (lra): Likewise.
936 * reload1.c (reload): Likewise.
937 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
938 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
939 * rtlanal.c (get_initial_register_offset): Return the offset as
940 a poly_int64 rather than a HOST_WIDE_INT.
942 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
943 Alan Hayward <alan.hayward@arm.com>
944 David Sherwood <david.sherwood@arm.com>
946 * reload1.c (elim_table): Change initial_offset, offset and
947 previous_offset from HOST_WIDE_INT to poly_int64_pod.
948 (offsets_at): Change the target array's element type from
949 HOST_WIDE_INT to poly_int64_pod.
950 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
951 (elimination_costs_in_insn, update_eliminable_offsets)
952 (verify_initial_elim_offsets, set_offsets_for_label)
953 (init_eliminable_invariants): Update after above changes.
955 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
956 Alan Hayward <alan.hayward@arm.com>
957 David Sherwood <david.sherwood@arm.com>
959 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
960 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
961 (decomposition): Change start and end from HOST_WIDE_INT
963 (decompose, immune_p): Update accordingly.
964 (find_inc_amount): Return a poly_int64 rather than an int.
965 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
968 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
969 Alan Hayward <alan.hayward@arm.com>
970 David Sherwood <david.sherwood@arm.com>
972 * tree.h (get_inner_reference): Return the bitsize and bitpos
973 as poly_int64_pods rather than HOST_WIDE_INT.
974 * fold-const.h (ptr_difference_const): Return the pointer difference
975 as a poly_int64_pod rather than a HOST_WIDE_INT.
976 * expr.c (get_inner_reference): Return the bitsize and bitpos
977 as poly_int64_pods rather than HOST_WIDE_INT.
978 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
980 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
981 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
982 (optimize_bit_field_compare): Update call to get_inner_reference.
983 (decode_field_reference): Likewise.
984 (fold_unary_loc): Track polynomial offsets and sizes.
985 (split_address_to_core_and_offset): Return the bitpos as a
986 poly_int64_pod rather than a HOST_WIDE_INT.
987 (ptr_difference_const): Likewise for the pointer difference.
988 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
989 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
990 * dbxout.c (dbxout_expand_expr): Likewise.
991 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
992 (loc_list_from_tree_1, fortran_common): Likewise.
993 * gimple-laddress.c (pass_laddress::execute): Likewise.
994 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
995 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
996 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
997 * tree-affine.c (tree_to_aff_combination): Likewise.
998 (get_inner_reference_aff): Likewise.
999 * tree-data-ref.c (split_constant_offset_1): Likewise.
1000 (dr_analyze_innermost): Likewise.
1001 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1002 * tree-sra.c (ipa_sra_check_caller): Likewise.
1003 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1004 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1005 (instrument_bool_enum_load, instrument_object_size): Likewise.
1006 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1007 to get_inner_reference.
1008 * hsa-gen.c (gen_hsa_addr): Likewise.
1009 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1010 * tsan.c (instrument_expr): Likewise.
1011 * match.pd: Update call to ptr_difference_const.
1013 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1014 Alan Hayward <alan.hayward@arm.com>
1015 David Sherwood <david.sherwood@arm.com>
1017 * fold-const.c (fold_comparison): Track sizes and offsets as
1018 poly_int64s rather than HOST_WIDE_INTs when folding address
1021 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1022 Alan Hayward <alan.hayward@arm.com>
1023 David Sherwood <david.sherwood@arm.com>
1025 * expr.h (get_bit_range): Return the bitstart and bitend as
1026 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1027 as a poly_int64 rather than a HOST_WIDE_INT.
1028 * expr.c (get_bit_range): Likewise.
1029 (expand_assignment): Update call accordingly.
1030 * fold-const.c (optimize_bit_field_compare): Likewise.
1032 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1033 Alan Hayward <alan.hayward@arm.com>
1034 David Sherwood <david.sherwood@arm.com>
1036 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1037 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1038 as an optional final parameter.
1039 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1040 (aarch64_legitimate_address_p): Likewise.
1041 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1042 instead of an rtx code.
1043 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1044 (aarch64_legitimate_address_hook_p): Likewise.
1045 (aarch64_print_ldpstp_address): Likewise.
1046 (aarch64_print_operand_address): Likewise.
1047 (aarch64_address_cost): Likewise.
1048 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1049 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1050 (aarch64_mem_pair_lanes_operand): Likewise.
1052 2017-12-20 Richard Biener <rguenther@suse.de>
1054 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1055 update_stmt_if_modified.
1057 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1059 PR tree-optimization/83491
1060 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1061 before walking uses. Improve coding style and comments.
1063 2017-12-20 Tom de Vries <tom@codesourcery.com>
1065 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1067 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1070 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1072 (pass_ipa_free_fn_summary): Add small_p private data member,
1073 initialize to false in the ctor.
1074 (pass_ipa_free_fn_summary::clone,
1075 pass_ipa_free_fn_summary::set_pass_param,
1076 pass_ipa_free_fn_summary::gate): New methods.
1077 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1078 | TODO_dump_symtab if small_p.
1079 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1080 entry and add another instance of the pass with false parm after
1082 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1083 ipa_free_fn_summary here.
1085 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1087 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1089 2017-12-20 Martin Sebor <msebor@redhat.com>
1092 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1093 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1094 (expand_builtin_strncmp): Same.
1096 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1099 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1100 allowed debug stmts before labels.
1101 (expand_gimple_basic_block): Likewise.
1102 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1103 * gimple-iterator.h (gsi_after_labels): Likewise.
1104 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1105 rename reused variable, and simplify using gsi_move_before.
1106 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1107 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1108 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1109 (gimple_verify_flow_info, gimple_block_label): Likewise.
1110 (make_blocks): Move debug markers after adjacent labels.
1111 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1112 allowed debug insns outside blocks.
1113 * df-scan.c (df_insn_delete): Likewise.
1114 * lra-constraints.c (update_ebb_live_info): Likewise.
1115 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1116 (vt_initialize, delete_vta_debug_insns): Likewise.
1117 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1119 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1120 Alan Hayward <alan.hayward@arm.com>
1121 David Sherwood <david.sherwood@arm.com>
1123 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1124 (round_up_to_byte_boundary): Likewise.
1125 * expr.h (get_bit_range): Add temporary shim.
1126 * gimple-ssa-store-merging.c (store_operand_info): Change the
1127 bitsize, bitpos, bitregion_start and bitregion_end fields from
1128 unsigned HOST_WIDE_INT to poly_uint64.
1129 (merged_store_group): Likewise load_align_base.
1130 (compatible_load_p, compatible_load_p): Update accordingly.
1131 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1132 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1133 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1134 bitregion_start and bitregion_end as poly_uint64s rather than
1135 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1136 (handled_load): Take the bitsize, bitpos,
1137 bitregion_start and bitregion_end as poly_uint64s rather than
1138 unsigned HOST_WIDE_INTs.
1139 (pass_store_merging::process_store): Update call to
1140 mem_valid_for_store_merging.
1142 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1143 Alan Hayward <alan.hayward@arm.com>
1144 David Sherwood <david.sherwood@arm.com>
1146 * builtins.c (get_object_alignment_2): Track polynomial offsets
1147 and sizes. Update the alignment handling.
1149 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1150 Alan Hayward <alan.hayward@arm.com>
1151 David Sherwood <david.sherwood@arm.com>
1153 * tree.h (get_inner_reference): Add a version that returns the
1154 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1155 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1156 the case in which bitpos is not associated with the first byte.
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-affine.h (get_inner_reference_aff): Return the size as a
1164 * tree-affine.c (get_inner_reference_aff): Likewise.
1165 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1166 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1168 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1169 Alan Hayward <alan.hayward@arm.com>
1170 David Sherwood <david.sherwood@arm.com>
1172 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1173 HOST_WIDE_INT rather than a poly_int64.
1175 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1176 Alan Hayward <alan.hayward@arm.com>
1177 David Sherwood <david.sherwood@arm.com>
1179 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1180 HOST_WIDE_INT to poly_int64_pod.
1181 (perform_symbolic_merge): Update accordingly.
1182 (bswap_replace): Likewise.
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 * tree-affine.h (aff_tree::offset): Change from widest_int
1190 (wide_int_ext_for_comb): Delete.
1191 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1192 constants as poly_widest_int rather than widest_int.
1193 (aff_combination_constant_multiple_p): Return the multiplier
1194 as a poly_widest_int.
1195 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1197 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1198 version static and add an overload for poly_widest_int.
1199 (aff_combination_const, aff_combination_add_cst)
1200 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1201 the constants as poly_widest_int rather than widest_int.
1202 (tree_to_aff_combination): Generalize INTEGER_CST case to
1204 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1205 (aff_combination_add_product, aff_combination_mult): Handle
1207 (aff_combination_constant_multiple_p): Return the multiplier
1208 as a poly_widest_int.
1209 * tree-predcom.c (determine_offset): Return the offset as a
1211 (split_data_refs_to_components, suitable_component_p): Update
1213 (valid_initializer_p): Update call to
1214 aff_combination_constant_multiple_p.
1215 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1216 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1217 as a poly_int64 rather than a HOST_WIDE_INT.
1218 (get_address_cost): Handle polynomial offsets.
1219 (iv_elimination_compare_lt): Likewise.
1220 (rewrite_use_nonlinear_expr): Likewise.
1222 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1223 Alan Hayward <alan.hayward@arm.com>
1224 David Sherwood <david.sherwood@arm.com>
1226 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1227 as a poly_int64_pod rather than a HOST_WIDE_INT.
1228 (get_addr_base_and_unit_offset): Likewise.
1229 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1230 (get_addr_base_and_unit_offset): Likewise.
1231 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1232 to poly_int64 in example.
1233 * fold-const.c (fold_binary_loc): Update call to
1234 get_addr_base_and_unit_offset.
1235 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1236 (maybe_canonicalize_mem_ref_addr): Likewise.
1237 (gimple_fold_stmt_to_constant_1): Likewise.
1238 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1240 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1241 * match.pd: Likewise.
1242 * omp-low.c (lower_omp_target): Likewise.
1243 * tree-sra.c (build_ref_for_offset): Likewise.
1244 (build_debug_ref_for_model): Likewise.
1245 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1246 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1247 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1248 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1249 (constant_pointer_difference): Likewise.
1250 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1251 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1252 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1253 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1254 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1255 (set_ssa_val_to): Likewise.
1256 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1257 (maybe_diag_stxncpy_trunc): Likewise.
1258 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1259 * tree.c (build_simple_mem_ref_loc): Likewise.
1260 (array_at_struct_end_p): Likewise.
1262 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1263 Alan Hayward <alan.hayward@arm.com>
1264 David Sherwood <david.sherwood@arm.com>
1266 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1267 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1268 (get_ref_base_and_extent_hwi): Declare.
1269 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1270 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1271 (get_ref_base_and_extent_hwi): New function.
1272 * cfgexpand.c (expand_debug_expr): Update call to
1273 get_ref_base_and_extent.
1274 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1275 * gimple-fold.c (get_base_constructor): Return the offset as a
1276 poly_int64_pod rather than a HOST_WIDE_INT.
1277 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1278 * ipa-polymorphic-call.c
1279 (ipa_polymorphic_call_context::set_by_invariant)
1280 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1281 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1282 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1283 rather than get_ref_base_and_extent.
1284 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1285 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1286 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1288 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1289 call to get_ref_base_and_extent.
1290 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1291 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1292 (stmt_kills_ref_p): Likewise.
1293 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1294 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1296 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1297 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1298 when calling native_encode_expr.
1299 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1300 call to get_ref_base_and_extent.
1301 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1302 get_ref_base_and_extent.
1303 * var-tracking.c (track_expr_p): Likewise.
1305 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1306 Alan Hayward <alan.hayward@arm.com>
1307 David Sherwood <david.sherwood@arm.com>
1309 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1310 HOST_WIDE_INT to poly_int64_pod.
1311 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1312 polynomail parameter offsets.
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 * gengtype.c (main): Handle poly_int64_pod.
1319 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1320 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1321 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1322 from HOST_WIDE_INT to poly_int64_pod.
1323 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1324 (copy_cfa): New function.
1325 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1326 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1327 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1328 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1329 (initial_return_save): Treat offsets as poly_ints.
1330 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1331 offset is nonconstant.
1332 (reg_save): Take the offset as a poly_int64. Fall back to
1333 DW_CFA_expression for nonconstant offsets.
1334 (queue_reg_save): Take the offset as a poly_int64.
1335 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1337 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1338 Alan Hayward <alan.hayward@arm.com>
1339 David Sherwood <david.sherwood@arm.com>
1341 * rtl.h (operand_subword, operand_subword_force): Take the offset
1342 as a poly_uint64 an unsigned int.
1343 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1345 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1346 Alan Hayward <alan.hayward@arm.com>
1347 David Sherwood <david.sherwood@arm.com>
1349 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1350 'p' format code. Use INT_LIST rather than SUBREG as the example of
1351 a code with an XINT and an XEXP. Remove the implication that
1352 accessing an rtx field using XINT is expected to work.
1353 * rtl.def (SUBREG): Change format from "ei" to "ep".
1354 * rtl.h (rtunion::rt_subreg): New field.
1355 (XCSUBREG): New macro.
1356 (SUBREG_BYTE): Use it.
1357 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1358 Update constructor accordingly.
1359 (subreg_shape::operator ==): Update accordingly.
1360 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1361 than an unsigned int.
1362 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1363 a poly_uint64 rather than an unsigned int.
1364 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1365 than an unsigned int.
1366 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1367 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1368 an unsigned int. Take the sizes as poly_uint64s.
1369 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1370 an unsigned int. Take the shift as a poly_uint64 rather than
1372 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1373 as a poly_uint64 rather than an unsigned int.
1374 (simplify_subreg_regno): Likewise.
1375 (byte_lowpart_offset): Return the memory offset as a poly_int64
1377 (subreg_memory_offset): Likewise. Take the subreg offset as a
1378 poly_uint64 rather than an unsigned int.
1379 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1380 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1381 poly_uint64 rather than an unsigned int.
1382 * rtl.c (rtx_format): Describe 'p' in comment.
1383 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1384 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1385 offset as a poly_uint64 rather than an unsigned int.
1386 (byte_lowpart_offset): Return the memory offset as a poly_int64
1388 (subreg_memory_offset): Likewise. Take the subreg offset as a
1389 poly_uint64 rather than an unsigned int.
1390 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1391 mode sizes as poly_uint64s rather than unsigned ints. Return a
1392 poly_uint64 rather than an unsigned int.
1393 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1394 (copy_insn_1): Handle 'p'.
1395 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1396 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1397 an unsigned int. Return the shift in the same way.
1398 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1400 (subreg_size_offset_from_lsb): Take the sizes and shift as
1401 poly_uint64s rather than unsigned ints. Return the offset as
1403 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1404 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1406 * rtlhash.c (add_rtx): Handle 'p'.
1407 * genemit.c (gen_exp): Likewise.
1408 * gengenrtl.c (type_from_format, gendef): Likewise.
1409 * gensupport.c (subst_pattern_match, get_alternatives_number)
1410 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1411 (subst_dup): Likewise.
1412 * gengtype.c (adjust_field_rtx_def): Likewise.
1413 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1414 (match_pattern_2): Likewise.
1415 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1416 (rtx_test::subreg_field): New function.
1417 (operator ==, safe_to_hoist_p, transition_parameter_type)
1418 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1419 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1420 * genpeep.c (match_rtx): Likewise.
1421 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1422 (rtx_writer::print_rtx_operand): Handle 'p'.
1423 (print_value): Handle SUBREG.
1424 * read-rtl.c (apply_int_iterator): Likewise.
1425 (rtx_reader::read_rtx_operand): Handle 'p'.
1426 * alias.c (rtx_equal_for_memref_p): Likewise.
1427 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1428 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1430 * calls.c (expand_call): Likewise.
1431 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1432 (make_extraction, gen_lowpart_for_combine): Likewise.
1433 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1435 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1436 rather than an unsigned int. Treat subreg offsets as poly_ints.
1437 (exp_equiv_p): Handle 'p'.
1438 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1439 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1440 * dse.c (find_shift_sequence): Likewise.
1441 * dwarf2out.c (rtl_for_decl_location): Likewise.
1442 * expmed.c (extract_low_bits): Likewise.
1443 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1444 (expand_expr_real_2): Likewise.
1445 * final.c (alter_subreg): Likewise.
1446 (leaf_renumber_regs_insn): Handle 'p'.
1447 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1448 Treat subreg offsets as poly_ints.
1449 * fwprop.c (forward_propagate_and_simplify): Likewise.
1450 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1451 * ira.c (get_subreg_tracking_sizes): Likewise.
1452 * ira-conflicts.c (go_through_subreg): Likewise.
1453 * ira-lives.c (process_single_reg_class_operands): Likewise.
1454 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1455 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1456 as a poly_uint64 rather than an unsigned int.
1457 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1458 subreg offsets as poly_ints.
1459 * lra-constraints.c (operands_match_p): Handle 'p'.
1460 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1461 * lra-spills.c (assign_mem_slot): Likewise.
1462 * postreload.c (move2add_valid_value_p): Likewise.
1463 * recog.c (general_operand, indirect_operand): Likewise.
1464 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1465 (copyprop_hardreg_forward_1): Likewise.
1466 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1467 (record_subregs_of_mode): Likewise.
1468 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1469 * reload.c (operands_match_p): Handle 'p'.
1470 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1471 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1472 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1473 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1474 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1475 (simplify_const_poly_int_tests<N>::run): Likewise.
1476 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1477 a poly_uint64 rather than an unsigned int.
1478 * valtrack.c (debug_lowpart_subreg): Likewise.
1479 * var-tracking.c (var_lowpart): Likewise.
1480 (loc_cmp): Handle 'p'.
1482 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1483 Alan Hayward <alan.hayward@arm.com>
1484 David Sherwood <david.sherwood@arm.com>
1486 * ira.c (get_subreg_tracking_sizes): New function.
1487 (init_live_subregs): Take an integer size rather than a register.
1488 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1489 to init_live_subregs.
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 * expr.c (store_constructor_field): Change bitsize from a
1496 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1497 HOST_WIDE_INT to a poly_int64.
1498 (store_constructor): Change size from a HOST_WIDE_INT to
1500 (store_field): Likewise bitsize and bitpos.
1502 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1503 Alan Hayward <alan.hayward@arm.com>
1504 David Sherwood <david.sherwood@arm.com>
1506 * expmed.h (store_bit_field): Change bitregion_start and
1507 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1508 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1509 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1510 (store_fixed_bit_field, store_split_bit_field): Likewise.
1511 * expr.c (store_constructor_field, store_field): Likewise.
1512 (optimize_bitfield_assignment_op): Likewise. Make the same change
1513 to bitsize and bitpos.
1514 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1515 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1516 same change in the constructor arguments.
1517 (get_best_mode): Change bitregion_start and bitregion_end from
1518 unsigned HOST_WIDE_INT to poly_uint64.
1519 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1520 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1522 (bit_field_mode_iterator::next_mode): Update for new types
1523 of m_bitregion_start and m_bitregion_end.
1524 (get_best_mode): Change bitregion_start and bitregion_end from
1525 unsigned HOST_WIDE_INT to poly_uint64.
1527 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1528 Alan Hayward <alan.hayward@arm.com>
1529 David Sherwood <david.sherwood@arm.com>
1531 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1532 accepts poly_uint64 offsets.
1533 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1534 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1535 * expmed.c (lowpart_bit_field_p): Likewise.
1536 (extract_bit_field_as_subreg): New function, split out from...
1537 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1538 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1539 extractions, check that BITSIZE matches the size of the extracted
1540 value and that BITNUM is an exact multiple of that size.
1541 If all else fails, try forcing the value into memory if
1542 BITNUM is variable, and adjusting the address so that the
1543 offset is constant. Split the part that can only handle constant
1544 bitsize and bitnum out into...
1545 (extract_integral_bit_field): ...this new function.
1546 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1547 rather than unsigned HOST_WIDE_INTs.
1549 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1550 Alan Hayward <alan.hayward@arm.com>
1551 David Sherwood <david.sherwood@arm.com>
1553 * expmed.h (store_bit_field): Take bitsize and bitnum as
1554 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1555 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1556 that returns the byte size.
1557 (store_bit_field_1): Take bitsize and bitnum as
1558 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1559 to simple_mem_bitfield_p. Split the part that can only handle
1560 constant bitsize and bitnum out into...
1561 (store_integral_bit_field): ...this new function.
1562 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1563 than unsigned HOST_WIDE_INTs.
1564 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1566 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1567 Alan Hayward <alan.hayward@arm.com>
1568 David Sherwood <david.sherwood@arm.com>
1570 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1571 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1573 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1574 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1575 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1576 offset as a poly_int64 rather than an int.
1577 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1578 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1579 * lra-constraints.c (equiv_address_substitution): Track offsets
1581 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1582 (curr_insn_transform): Handle the new form of sp_offset.
1583 * lra-eliminations.c (lra_elim_table): Change previous_offset
1584 and offset from HOST_WIDE_INT to poly_int64.
1585 (print_elim_table, update_reg_eliminate): Update accordingly.
1586 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1587 (get_elimination): Update accordingly.
1588 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1589 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1590 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1591 poly_int64 offsets generally.
1592 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1593 (mark_not_eliminable, init_elimination): Update accordingly.
1594 (remove_reg_equal_offset_note): Return a bool and pass the new
1595 offset back by pointer as a poly_int64.
1596 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1597 rather than a HOST_WIDE_INT.
1598 (do_remat): Track offsets poly_int64s.
1599 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1601 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1602 Alan Hayward <alan.hayward@arm.com>
1603 David Sherwood <david.sherwood@arm.com>
1605 * rtl.h (mem_attrs): Add a default constructor. Change size and
1606 offset from HOST_WIDE_INT to poly_int64.
1607 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1608 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1609 (widen_memory_access): Take the sizes and offsets as poly_int64s
1610 rather than HOST_WIDE_INTs.
1611 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1612 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1614 (adjust_offset_for_component_ref): Change the offset from a
1615 HOST_WIDE_INT to a poly_int64.
1616 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1617 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1618 * dce.c (find_call_stack_args): Likewise.
1619 * dse.c (record_store): Likewise.
1620 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1621 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1622 * read-rtl-function.c (test_loading_mem): Likewise.
1623 * rtlanal.c (may_trap_p_1): Likewise.
1624 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1625 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1626 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1627 (mem_attrs::mem_attrs): New function.
1628 (set_mem_attributes_minus_bitpos): Change bitpos from a
1629 HOST_WIDE_INT to poly_int64.
1630 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1631 (clear_mem_offset, clear_mem_size, change_address)
1632 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1633 initialize mem_attrs.
1634 (set_mem_offset, set_mem_size, adjust_address_1)
1635 (adjust_automodify_address_1, offset_address, widen_memory_access):
1636 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1638 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1639 Alan Hayward <alan.hayward@arm.com>
1640 David Sherwood <david.sherwood@arm.com>
1642 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1643 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1644 rather than 0 to represent an unknown size. Assert that the size
1645 is known when the mode isn't BLKmode.
1646 (may_trap_p_1): Use -1 for unknown sizes.
1647 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1649 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1650 Alan Hayward <alan.hayward@arm.com>
1651 David Sherwood <david.sherwood@arm.com>
1653 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1654 to poly_int64. Update commentary for positions_needed.large.
1655 (read_info_type): Change offset and width from HOST_WIDE_INT
1657 (set_usage_bits): Likewise.
1658 (canon_address): Return the offset as a poly_int64 rather than
1659 a HOST_WIDE_INT. Use strip_offset_and_add.
1660 (set_all_positions_unneeded, any_positions_needed_p): Use
1661 positions_needed.large to track stores with non-constant widths.
1662 (all_positions_needed_p): Likewise. Take the offset and width
1663 as poly_int64s rather than ints. Assert that rhs is nonnull.
1664 (record_store): Cope with non-constant offsets and widths.
1665 Nullify the rhs of an earlier store if we can't tell which bytes
1667 (find_shift_sequence): Take the access_size and shift as poly_int64s
1669 (get_stored_val): Take the read_offset and read_width as poly_int64s
1670 rather than HOST_WIDE_INTs.
1671 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1672 non-constant offsets and widths.
1674 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1675 Alan Hayward <alan.hayward@arm.com>
1676 David Sherwood <david.sherwood@arm.com>
1678 * inchash.h (inchash::hash::add_poly_int): New function.
1679 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1680 Use poly_int64 rather than HOST_WIDE_INT.
1681 (ao_ref::max_size_known_p): New function.
1682 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1683 rather than HOST_WIDE_INT.
1684 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1685 to temporaries until its interface is adjusted to match.
1686 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1687 (aliasing_component_refs_p, decl_refs_may_alias_p)
1688 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1689 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1690 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1692 * alias.c (ao_ref_from_mem): Likewise.
1693 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1694 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1695 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1696 (maybe_trim_complex_store, maybe_trim_constructor_store)
1697 (live_bytes_read, dse_classify_store): Likewise.
1698 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1699 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1700 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1701 (vn_reference_lookup_3): Likewise.
1702 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1704 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1705 Alan Hayward <alan.hayward@arm.com>
1706 David Sherwood <david.sherwood@arm.com>
1708 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1709 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1710 instead of ranges_overlap_p.
1712 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1713 Alan Hayward <alan.hayward@arm.com>
1714 David Sherwood <david.sherwood@arm.com>
1716 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1717 sizes as poly_int64s rather than HOST_WIDE_INTs.
1719 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1720 Alan Hayward <alan.hayward@arm.com>
1721 David Sherwood <david.sherwood@arm.com>
1723 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1724 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1725 * gimple-fold.c (fold_ctor_reference): Likewise.
1727 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1728 Alan Hayward <alan.hayward@arm.com>
1729 David Sherwood <david.sherwood@arm.com>
1731 * target.def (dwarf_poly_indeterminate_value): New hook.
1732 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1733 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1734 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1735 * doc/tm.texi: Regenerate.
1736 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1737 offset as a poly_int64.
1738 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1739 offset as a poly_int64.
1740 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1741 Take the offset as a poly_int64.
1742 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1743 (frame_pointer_fb_offset): Change to a poly_int64.
1744 (int_loc_descriptor): Take the offset as a poly_int64. Use
1745 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1746 (based_loc_descr): Take the offset as a poly_int64.
1747 Use strip_offset_and_add to handle (plus X (const)).
1748 Use new_reg_loc_descr instead of an open-coded version of the
1749 previous implementation.
1750 (mem_loc_descriptor): Handle CONST_POLY_INT.
1751 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1752 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1754 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1755 Alan Hayward <alan.hayward@arm.com>
1756 David Sherwood <david.sherwood@arm.com>
1758 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1759 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1760 * inchash.h (inchash::hash::add_poly_hwi): New function.
1761 * gengtype.c (main): Register poly_int64.
1762 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1763 offset as a poly_int.
1764 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1765 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1766 offset as a poly_int64.
1767 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1768 * print-rtl.c (print_poly_int): New function.
1769 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1771 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1773 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1774 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1775 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1776 Enforce track_offset_p for parts of a PARALLEL.
1777 (vt_add_function_parameter): Use const_offset for the final
1778 offset to track. Use get_tracked_reg_offset for the parts
1781 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1782 Alan Hayward <alan.hayward@arm.com>
1783 David Sherwood <david.sherwood@arm.com>
1785 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1786 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1787 * doc/tm.texi: Regenerate.
1788 * hooks.h (hook_bool_uint_uint_true): Delete.
1789 (hook_bool_puint64_puint64_true): Declare.
1790 * hooks.c (hook_bool_uint_uint_true): Delete.
1791 (hook_bool_puint64_puint64_true): New function.
1792 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1793 instead of unsigned ints.
1794 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1795 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1797 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1798 Alan Hayward <alan.hayward@arm.com>
1799 David Sherwood <david.sherwood@arm.com>
1801 * optabs.h (expand_operand): Add an int_value field.
1802 (create_expand_operand): Add an int_value parameter and use it
1803 to initialize the new expand_operand field.
1804 (create_integer_operand): Replace with a declaration of a function
1805 that accepts poly_int64s. Move the implementation to...
1806 * optabs.c (create_integer_operand): ...here.
1807 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1808 the mode preserves the value of int_value, instead of calling
1809 const_int_operand on the rtx. Use gen_int_mode to generate
1812 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1813 Alan Hayward <alan.hayward@arm.com>
1814 David Sherwood <david.sherwood@arm.com>
1816 * dumpfile.h (dump_dec): Declare.
1817 * dumpfile.c (dump_dec): New function.
1818 * pretty-print.h (pp_wide_integer): Turn into a function and
1819 declare a poly_int version.
1820 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1822 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1823 Alan Hayward <alan.hayward@arm.com>
1824 David Sherwood <david.sherwood@arm.com>
1826 * doc/generic.texi (POLY_INT_CST): Document.
1827 * tree.def (POLY_INT_CST): New tree code.
1828 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1829 * tree-core.h (tree_poly_int_cst): New struct.
1830 (tree_node): Add a poly_int_cst field.
1831 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1832 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1833 instead of a wide_int_ref.
1834 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1836 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1837 instead of an unsigned HOST_WIDE_INT.
1838 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1839 (ptrdiff_tree_p): Declare.
1840 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1841 extern inline implementations if the target doesn't use POLY_INT_CST.
1842 (poly_int_tree_p): New function.
1843 (wi::unextended_tree): New class.
1844 (wi::int_traits <unextended_tree>): New override.
1845 (wi::extended_tree): Add a default constructor.
1846 (wi::extended_tree::get_tree): New function.
1847 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1848 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1849 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1850 (wi::tree_to_poly_wide_ref): New typedefs.
1851 (wi::ints_for): Provide overloads for extended_tree and
1853 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1854 (wi::to_wide): New functions.
1855 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1856 * tree.c (poly_int_cst_hasher): New struct.
1857 (poly_int_cst_hash_table): New variable.
1858 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1859 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1861 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1862 (init_ttree): Initialize poly_int_cst_hash_table.
1863 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1864 a poly_int64 instead of a HOST_WIDE_INT.
1865 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1866 instead of an unsigned HOST_WIDE_INT.
1867 (wide_int_to_tree): Rename to...
1868 (wide_int_to_tree_1): ...this.
1869 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1870 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1871 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1872 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1873 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1874 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1876 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1877 (streamer_read_tree_body): Likewise.
1878 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1879 (streamer_write_tree_body): Likewise.
1880 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1881 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1882 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1883 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1884 * gimple-expr.h (is_gimple_constant): Likewise.
1885 * gimplify.c (maybe_with_size_expr): Likewise.
1886 * print-tree.c (print_node): Likewise.
1887 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1888 * tree-pretty-print.c (dump_generic_node): Likewise.
1889 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1890 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1891 * tree-vrp.c (compare_values_warnv): Likewise.
1892 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1893 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1894 (force_expr_to_var_cost): Likewise.
1895 * tree-ssa-loop.c (for_each_index): Likewise.
1896 * fold-const.h (build_invariant_address, size_int_kind): Take a
1897 poly_int64 instead of a HOST_WIDE_INT.
1898 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1899 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1901 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1902 (int_const_binop_2): New function, split out from...
1903 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1904 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1905 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1906 * gimple-ssa-strength-reduction.c (slsr_process_add)
1907 (slsr_process_mul): Check for INTEGER_CSTs before using them
1909 * stor-layout.c (bits_from_bytes): New function.
1910 (bit_from_pos): Use it.
1911 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1912 by BITS_PER_UNIT to get the TYPE_SIZE.
1913 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1914 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1916 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1917 Alan Hayward <alan.hayward@arm.com>
1918 David Sherwood <david.sherwood@arm.com>
1920 * doc/rtl.texi (const_poly_int): Document. Also document the
1921 rtl sharing behavior.
1922 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1923 * rtl.h (const_poly_int_def): New struct.
1924 (rtx_def::u): Add a cpi field.
1925 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1926 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1927 (wi::rtx_to_poly_wide_ref): New typedef
1928 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1929 (poly_int_rtx_p): New functions.
1930 (trunc_int_for_mode): Declare a poly_int64 version.
1931 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1932 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1934 (strip_offset): Declare.
1935 (strip_offset_and_add): New function.
1936 * rtl.def (CONST_POLY_INT): New rtx code.
1937 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1938 (shared_const_p): Use poly_int_rtx_p.
1939 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1941 (gen_int_shift_amount): Likewise.
1942 * emit-rtl.c (const_poly_int_hasher): New class.
1943 (const_poly_int_htab): New variable.
1944 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1945 (const_poly_int_hasher::hash): New function.
1946 (const_poly_int_hasher::equal): Likewise.
1947 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1948 (immed_wide_int_const): Rename to...
1949 (immed_wide_int_const_1): ...this and make static.
1950 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1951 instead of a wide_int_ref.
1952 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1953 (gen_lowpart_common): Handle CONST_POLY_INT.
1954 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1955 * cselib.c (cselib_hash_rtx): Likewise.
1956 * dwarf2out.c (const_ok_for_output_1): Likewise.
1957 * expr.c (convert_modes): Likewise.
1958 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1959 * rtlhash.c (add_rtx): Likewise.
1960 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1961 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1962 Handle existing CONST_POLY_INT rtxes.
1963 * expmed.h (expand_shift): Take a poly_int64 instead of a
1965 * expmed.c (expand_shift): Likewise.
1966 * rtlanal.c (strip_offset): New function.
1967 (commutative_operand_precedence): Give CONST_POLY_INT the same
1968 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1970 * rtl-tests.c (const_poly_int_tests): New struct.
1971 (rtl_tests_c_tests): Use it.
1972 * simplify-rtx.c (simplify_const_unary_operation): Handle
1974 (simplify_const_binary_operation): Likewise.
1975 (simplify_binary_operation_1): Fold additions of symbolic constants
1976 and CONST_POLY_INTs.
1977 (simplify_subreg): Handle extensions and truncations of
1979 (simplify_const_poly_int_tests): New struct.
1980 (simplify_rtx_c_tests): Use it.
1981 * wide-int.h (storage_ref): Add default constructor.
1982 (wide_int_ref_storage): Likewise.
1983 (trailing_wide_ints): Use GTY((user)).
1984 (trailing_wide_ints::operator[]): Add a const version.
1985 (trailing_wide_ints::get_precision): New function.
1986 (trailing_wide_ints::extra_size): Likewise.
1988 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1989 Alan Hayward <alan.hayward@arm.com>
1990 David Sherwood <david.sherwood@arm.com>
1992 * emit-rtl.h (gen_int_shift_amount): Declare.
1993 * emit-rtl.c (gen_int_shift_amount): New function.
1994 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1996 * calls.c (shift_return_value): Likewise.
1997 * cse.c (fold_rtx): Likewise.
1998 * dse.c (find_shift_sequence): Likewise.
1999 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2000 (expand_shift, expand_smod_pow2): Likewise.
2001 * lower-subreg.c (shift_cost): Likewise.
2002 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2003 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2004 (expand_vec_perm_var): Likewise.
2005 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2006 (simplify_binary_operation_1): Likewise.
2007 * combine.c (try_combine, find_split_point, force_int_to_mode)
2008 (simplify_shift_const_1, simplify_shift_const): Likewise.
2009 (change_zero_ext): Likewise. Use simplify_gen_binary.
2011 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2013 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2015 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2017 * doc/invoke.texi (ARM Options): Document accepted extension options
2018 for -march=armv8.3-a.
2020 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2023 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2024 or --with-float=softfp, set the default CPU to arm10e.
2026 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2028 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2029 * config/visium/predicates.md (const_shift_operand): Likewise.
2030 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2031 (visium_legitimize_reload_address): Likewise.
2033 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2036 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2038 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2041 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2043 2017-12-20 Julia Koval <julia.koval@intel.com>
2045 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2046 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2047 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2048 * config.gcc: Include vpclmulqdqintrin.h.
2049 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2050 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2052 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2053 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2054 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2055 * config/i386/i386.c (isa2_opts): Add -mcx16.
2056 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2057 (ix86_option_override_internal): Move mcx16 to flags2.
2058 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2059 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2060 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2061 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2062 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2063 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2064 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2065 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2066 * doc/invoke.texi: Add -mvpclmulqdq.
2068 2017-12-20 Tom de Vries <tom@codesourcery.com>
2071 * config/i386/i386.c (ix86_static_chain): Move
2072 DECL_STATIC_CHAIN test ...
2073 * calls.c (rtx_for_static_chain): ... here. New function.
2074 * calls.h (rtx_for_static_chain): Declare.
2075 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2076 instead of targetm.calls.static_chain.
2077 * df-scan.c (df_get_entry_block_def_set): Same.
2079 2017-12-19 Tom de Vries <tom@codesourcery.com>
2081 PR tree-optimization/83493
2082 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2085 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2087 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2088 inform with hardcoded english plural handling.
2090 2017-12-18 Jeff Law <law@redhat.com>
2092 PR tree-optimization/83477
2093 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2094 a non-virtual PHI, always push a new range.
2096 2017-12-19 Martin Sebor <msebor@redhat.com>
2099 * builtins.c (compute_objsize): Handle non-constant offsets.
2101 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2103 PR tree-optimization/83444
2104 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2105 character load case, if get_stridx on MEM_REF's operand doesn't
2106 look usable, retry with get_addr_stridx.
2108 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2111 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2112 (variable_tracking_main_1): Keep markers even when VTA fails.
2115 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2116 even if there are markers before it.
2117 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2119 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2121 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2122 typical order conditions.
2123 * sel-sched.c (extract_new_fences_from): Likewise.
2124 * config/visium/constraints.md (J, K, L): Likewise.
2125 * config/visium/predicates.md (const_shift_operand): Likewise.
2126 * config/visium/visium.c (visium_legitimize_address,
2127 visium_legitimize_reload_address): Likewise.
2128 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2129 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2130 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2131 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2132 SET_DUMP_DETAIL): Likewise.
2133 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2134 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2135 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2136 avr_set_core_architecture, avr_set_current_function,
2137 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2138 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2139 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2140 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2141 avr_map_decompose, avr_fold_builtin): Likewise.
2142 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2143 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2144 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2145 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2146 * config/m32c/m32c.c (m32c_conditional_register_usage,
2147 m32c_address_cost): Likewise.
2148 * config/m32c/predicates.md (shiftcount_operand,
2149 longshiftcount_operand): Likewise.
2150 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2151 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2152 can_use_cdx_ldstw): Likewise.
2153 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2154 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2156 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2157 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2158 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2159 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2160 * config/vax/vax.c (adjacent_operands_p): Likewise.
2161 * config/ft32/constraints.md (L, b, KA): Likewise.
2162 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2164 * cfgexpand.c (expand_stack_alignment): Likewise.
2165 * gcse.c (insert_expr_in_table): Likewise.
2166 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2167 * cgraphunit.c (cgraph_node::expand): Likewise.
2168 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2169 * emit-rtl.c (add_insn): Likewise.
2170 * input.c (dump_location_info): Likewise.
2171 * passes.c (NEXT_PASS): Likewise.
2172 * read-rtl-function.c (parse_note_insn_name,
2173 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2175 * sched-rgn.c (sched_rgn_init): Likewise.
2176 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2177 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2178 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2180 * explow.c (eliminate_constant_term): Likewise.
2181 * final.c (leaf_renumber_regs_insn): Likewise.
2182 * cfgrtl.c (print_rtl_with_bb): Likewise.
2183 * genhooks.c (emit_init_macros): Likewise.
2184 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2185 * tree-data-ref.c (conflict_fn): Likewise.
2186 * selftest.c (assert_streq): Likewise.
2187 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2188 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2189 fold_binary_loc, multiple_of_p): Likewise.
2190 * reload.c (push_reload, find_equiv_reg): Likewise.
2191 * et-forest.c (et_nca, et_below): Likewise.
2192 * dbxout.c (dbxout_symbol_location): Likewise.
2193 * reorg.c (relax_delay_slots): Likewise.
2194 * dojump.c (do_compare_rtx_and_jump): Likewise.
2195 * gengtype-parse.c (type): Likewise.
2196 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2197 simplify_const_relational_operation): Likewise.
2198 * reload1.c (do_output_reload): Likewise.
2199 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2200 * gengtype.c (type_for_name): Likewise.
2201 * gimple-ssa-sprintf.c (format_directive): Likewise.
2203 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2206 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2207 accessing it. Adjust comment.
2209 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2212 * predict.c (zero_one_minusone): New function.
2213 (apply_return_prediction): Avoid return prediction for functions
2214 returning only -1, 0 and 1 values, unless they only return -1 and 0
2217 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2219 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2220 fall-through warning.
2221 (arc_compute_frame_size): Remove unused variables.
2222 (arc_print_operand): Fix fprintif format.
2223 (arc_can_follow_jump): Clean fall-through warning.
2225 2017-12-19 Marek Polacek <polacek@redhat.com>
2228 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2231 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2233 * config/arc/arc.c (overriderregs): New variable.
2234 (arc_override_options): Track fixed/call saved/call options.
2235 (arc_conditional_register_usage): Check against overrideregs
2236 variable whenever we change register properties.
2238 2017-12-19 Nathan Sidwell <nathan@acm.org>
2240 * opts.c (finish_options): Don't prefix dump_base_name if it
2241 already contains directories.
2243 2017-12-19 Martin Liska <mliska@suse.cz>
2245 PR rtl-optimization/82675
2246 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2247 more element in sbitmap.
2249 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2251 * gimplify.c (gimplify_expr): Use error_operand_p.
2253 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2256 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2259 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2261 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2262 the jump is to a label.
2264 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2266 PR tree-optimization/83444
2267 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2272 * ipa-inline.c (flatten_remove_node_hook): New function.
2273 (ipa_inline): Keep only nodes with flatten attribute at the end of
2274 the array in the order from ipa_reverse_postorder, only walk that
2275 portion of array for flattening, if there is more than one such
2276 node, temporarily register a removal hook and ignore removed nodes.
2278 PR tree-optimization/80631
2279 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2280 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2281 IFN_REDUC_MAX or IFN_REDUC_MIN.
2283 2017-12-18 Jeff Law <law@redhat.com>
2285 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2286 record symbolic equivalences from backedges in the CFG.
2289 2017-11-19 Jeff Law <law@redhat.com>
2291 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2292 of degenerates resulting from ignoring an edge.
2294 2017-12-18 Martin Sebor <msebor@redhat.com>
2297 PR tree-optimization/78450
2298 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2299 (handle_builtin_strlen): Call it.
2301 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2303 PR rtl-optimization/83424
2304 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2306 2017-12-18 Marek Polacek <polacek@redhat.com>
2309 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2310 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2313 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2315 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2316 (umaddsidi4, umaddsidi_split): Likewise.
2318 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2320 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2323 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2326 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2328 2017-12-18 Richard Biener <rguenther@suse.de>
2330 PR tree-optimization/81877
2331 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2332 (outermost_indep_loop): Adjust.
2333 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2334 (can_sm_ref_p): Adjust.
2336 2017-12-18 Richard Biener <rguenther@suse.de>
2339 * tree.c (array_at_struct_end_p): Return true if the underlying
2340 object has space for at least one element in excess of what
2341 the array domain specifies.
2343 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2345 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2346 example to avoid overfull hbox.
2347 * doc/invoke.texi (Option Summary): Add missing @gol.
2348 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2351 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2352 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2354 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2355 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2356 (AArch64 Options): Document -mverbose-cost-dump.
2357 (ARM Options): Likewise, plus -mflip-thumb.
2359 2017-12-17 Martin Sebor <msebor@redhat.com>
2362 * gimple-ssa-warn-restrict.c
2363 (builtin_memref::offset_out_of_bounds): Correct the handling of
2366 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2368 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2370 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2372 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2374 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2376 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2379 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2381 * doc/invoke.texi: Fix some typos.
2383 2017-12-16 Martin Sebor <msebor@redhat.com>
2385 PR tree-optimization/78918
2386 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2387 * builtins.c (check_sizes): Rename...
2388 (check_access): ...to this. Rename function arguments for clarity.
2389 (check_memop_sizes): Adjust names.
2390 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2391 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2392 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2393 (check_strncat_sizes, expand_builtin_strncat): Same.
2394 (expand_builtin_strncpy, expand_builtin_memset): Same.
2395 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2396 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2397 (maybe_emit_sprintf_chk_warning): Same.
2398 (expand_builtin_strcpy): Adjust.
2399 (expand_builtin_stpcpy): Same.
2400 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2401 in pointer-checking forms of memcpy, memmove, and mempcpy.
2402 (gcall_to_tree_minimal, max_object_size): Define new functions.
2403 * builtins.h (max_object_size): Declare.
2404 * calls.c (alloc_max_size): Call max_object_size instead of
2405 hardcoding ssizetype limit.
2406 (get_size_range): Handle new argument.
2407 * calls.h (get_size_range): Add a new argument.
2408 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2409 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2410 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2412 (gimple_fold_builtin_memory_chk): Same.
2413 (gimple_fold_builtin_stxcpy_chk): New function.
2414 * gimple-ssa-warn-restrict.c: New source.
2415 * gimple-ssa-warn-restrict.h: New header.
2416 * gimple.c (gimple_build_call_from_tree): Propagate location.
2417 * passes.def (pass_warn_restrict): Add new pass.
2418 * tree-pass.h (make_pass_warn_restrict): Declare.
2419 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2421 (handle_builtin_strcat): Same.
2422 (strlen_optimize_stmt): Rename...
2423 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2424 stpncpy, strncpy, and their checking forms.
2426 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2428 PR rtl-optimization/82849
2429 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2430 and get_max_loop_iterations_int.
2432 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2433 Alan Hayward <alan.hayward@arm.com>
2434 David Sherwood <david.sherwood@arm.com>
2436 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2437 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2438 size as a poly_uint64.
2439 (mode_for_vector, mode_for_int_vector): Take the number of vector
2440 elements as a poly_uint64.
2441 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2442 size as a poly_uint64.
2443 (mode_for_vector, mode_for_int_vector): Take the number of vector
2444 elements as a poly_uint64.
2446 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2447 Alan Hayward <alan.hayward@arm.com>
2448 David Sherwood <david.sherwood@arm.com>
2450 * machmode.h (MACRO_MODE): New macro.
2451 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2452 * targhooks.c (default_libcall_value, default_secondary_reload)
2453 (default_memory_move_cost, default_register_move_cost)
2454 (default_class_max_nregs): Likewise.
2456 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2457 Alan Hayward <alan.hayward@arm.com>
2458 David Sherwood <david.sherwood@arm.com>
2460 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2461 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2463 * genautomata.c (main): Likewise.
2464 * genconditions.c (write_header): Likewise.
2465 * genemit.c (main): Likewise.
2466 * genextract.c (print_header): Likewise.
2467 * genopinit.c (main): Likewise.
2468 * genoutput.c (output_prologue): Likewise.
2469 * genpeep.c (main): Likewise.
2470 * genpreds.c (write_insn_preds_c): Likewise.
2471 * genrecog.c (writer_header): Likewise.
2472 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2473 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2474 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2475 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2476 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2477 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2478 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2479 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2480 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2481 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2482 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2483 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2484 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2485 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2486 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2487 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2488 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2489 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2490 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2491 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2492 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2493 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2494 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2495 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2496 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2497 * config/darwin.c (IN_TARGET_CODE): Likewise.
2498 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2499 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2500 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2501 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2502 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2503 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2504 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2505 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2506 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2507 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2508 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2509 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2510 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2511 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2512 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2513 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2514 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2515 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2516 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2517 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2518 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2519 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2520 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2521 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2522 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2523 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2524 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2525 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2526 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2527 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2528 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2529 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2530 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2531 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2532 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2533 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2534 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2535 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2536 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2537 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2538 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2539 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2540 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2541 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2542 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2543 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2544 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2545 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2546 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2547 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2548 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2549 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2550 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2551 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2552 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2553 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2554 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2555 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2556 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2557 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2558 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2559 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2560 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2561 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2562 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2563 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2564 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2565 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2566 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2567 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2568 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2569 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2570 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2571 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2572 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2573 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2574 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2575 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2576 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2577 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2578 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2579 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2580 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2581 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2582 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2583 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2584 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2585 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2586 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2587 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2588 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2589 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2590 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2591 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2592 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2593 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2594 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2595 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2596 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2597 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2598 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2599 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2600 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2601 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2602 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2603 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2605 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2606 Alan Hayward <alan.hayward@arm.com>
2607 David Sherwood <david.sherwood@arm.com>
2609 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2612 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2613 Alan Hayward <alan.hayward@arm.com>
2614 David Sherwood <david.sherwood@arm.com>
2616 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2617 * doc/md.texi (vec_series@var{m}): Document.
2618 * tree.def (VEC_SERIES_EXPR): New tree code.
2619 * tree.h (build_vec_series): Declare.
2620 * tree.c (build_vec_series): New function.
2621 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2622 * tree-pretty-print.c (dump_generic_node): Likewise.
2623 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2624 * tree-inline.c (estimate_operator_cost): Likewise.
2625 * expr.c (expand_expr_real_2): Likewise.
2626 * optabs-tree.c (optab_for_tree_code): Likewise.
2627 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2628 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2629 * expmed.c (make_tree): Handle VEC_SERIES.
2630 * optabs.def (vec_series_optab): New optab.
2631 * optabs.h (expand_vec_series_expr): Declare.
2632 * optabs.c (expand_vec_series_expr): New function.
2633 * tree-vect-generic.c (expand_vector_operations_1): Check that
2634 the operands also have vector type.
2636 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2637 Alan Hayward <alan.hawyard@arm.com>
2638 David Sherwood <david.sherwood@arm.com>
2640 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2641 (VEC_COND_EXPR): Add missing @tindex.
2642 * doc/md.texi (vec_duplicate@var{m}): Document.
2643 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2644 * tree.c (build_vector_from_val): Add stubbed-out handling of
2645 variable-length vectors, using VEC_DUPLICATE_EXPR.
2646 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2647 * cfgexpand.c (expand_debug_expr): Likewise.
2648 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2649 * tree-inline.c (estimate_operator_cost): Likewise.
2650 * tree-pretty-print.c (dump_generic_node): Likewise.
2651 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2652 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2653 (test_vec_duplicate_folding): New function.
2654 (fold_const_c_tests): Call it.
2655 * optabs.def (vec_duplicate_optab): New optab.
2656 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2657 * optabs.h (expand_vector_broadcast): Declare.
2658 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2659 vec_duplicate_optab.
2660 * expr.c (store_constructor): Try using vec_duplicate_optab for
2662 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2664 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2667 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2668 div/mod latencies a bit.
2670 2017-12-15 Jeff Law <law@redhat.com>
2672 PR tree-optimization/36550
2673 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2674 (mark_threaded_blocks): Rewrite code to avoid block copying when
2675 optimizing for size. Don't pessimize blocks which will be
2676 copied, but all the statements will be dead.
2678 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2680 PR tree-optimization/81165
2681 * tree-ssa-threadupdate.c (uses_in_bb): New.
2682 (estimate_threading_killed_stmts): New.
2683 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2684 * tree-ssa-threadedge.c
2685 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2688 2017-12-15 Jeff Law <law@redhat.com>
2690 PR tree-optimization/83410
2691 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2692 threads when parallelizing loops.
2694 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2696 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2698 * config/alpha/alpha.c (vms_attribute_table): Swap
2699 affects_type_identity and handler fields, adjust comments.
2700 * config/mips/mips.c (mips_attribute_table): Likewise.
2701 * config/visium/visium.c (visium_attribute_table): Likewise.
2702 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2703 * config/microblaze/microblaze.c (microblaze_attribute_table):
2705 * config/spu/spu.c (spu_attribute_table): Likewise.
2706 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2707 * config/arc/arc.c (arc_attribute_table): Likewise.
2708 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2709 * config/v850/v850.c (v850_handle_interrupt_attribute,
2710 v850_handle_data_area_attribute): Formatting fixes.
2711 (v850_attribute_table): Swap affects_type_identity and handler
2712 fields, adjust comments.
2713 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2714 * config/arm/arm.c (arm_attribute_table): Likewise.
2715 * config/avr/avr.c (avr_attribute_table): Likewise.
2716 * config/s390/s390.c (s390_attribute_table): Likewise.
2717 * config/sh/sh.c (sh_attribute_table): Likewise.
2718 * config/i386/i386.c (ix86_handle_cconv_attribute,
2719 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2720 (ix86_attribute_table): Swap affects_type_identity and handler
2721 fields, adjust comments.
2722 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2723 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2724 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2725 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2726 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2727 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2728 * config/rx/rx.c (rx_attribute_table): Likewise.
2729 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2730 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2731 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2732 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2733 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2734 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2735 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2736 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2737 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2738 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2739 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2741 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2743 (register_scoped_attributes, decl_attributes): Formatting fixes.
2745 PR tree-optimization/83269
2746 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2747 subtraction in arg0's type if type is signed and arg0 is unsigned.
2751 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2752 convert? on @0 instead of convert. Check type of @1, not @0.
2753 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2754 convert? on @0 instead of convert. Check type of @1, not @0.
2755 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2756 only optimize if either both @1 and @2 types are narrower
2757 precision, or both are wider or equal precision, and in the former
2758 case only if both have undefined overflow.
2760 2017-12-15 Richard Biener <rguenther@suse.de>
2763 * internal-fn.def (IFN_NOP): Add.
2764 * internal-fn.c (expand_NOP): Do nothing.
2765 * lto-streamer-in.c (input_function): Instead of removing
2766 sanitizer calls replace them with IFN_NOP calls.
2768 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2769 Alan Hayward <alan.hayward@arm.com>
2770 David Sherwood <david.sherwood@arm.com>
2772 * dse.c (store_info, read_info_type): Replace begin and end with
2774 (print_range): New function.
2775 (set_all_positions_unneeded, any_positions_needed_p)
2776 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2778 (record_store): Likewise. Optimize the case in which all positions
2780 (get_stored_val): Replace read_begin and read_end with read_offset
2782 (replace_read): Update call accordingly.
2784 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2786 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2787 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2788 (loop_cand::loop_cand): Initialize above members.
2789 (loop_cand::supported_operations): Delete.
2790 (loop_cand::can_interchange_p): Inline above function.
2791 (loop_cand::classify_simple_reduction): Record number of constant
2792 initialized simple reductions.
2793 (should_interchange_loops): New parameters. Check stmt cost of loops
2795 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2796 Update call to should_interchange_loops.
2797 (should_interchange_loop_nest): Update call to
2798 should_interchange_loops.
2800 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2803 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2804 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2805 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2807 2017-12-15 Julia Koval <julia.koval@intel.com>
2809 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2810 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2812 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2813 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2814 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2816 2017-12-15 Julia Koval <julia.koval@intel.com>
2818 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2819 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2820 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2821 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2822 _mm_aesenc_epi128): New intrinsics.
2824 2017-12-15 Julia Koval <julia.koval@intel.com>
2826 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2827 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2829 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2830 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2831 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2833 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2835 * gimple-ssa-strength-reduction.c (analyze_increments):
2836 Distinguish replacement costs for constant strides from those for
2839 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2841 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2843 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2845 * doc/invoke.texi: Document -Wcast-function-type.
2846 * recog.h (stored_funcptr): Change signature.
2847 * tree-dump.c (dump_node): Avoid warning.
2848 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2850 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2853 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2856 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2858 PR tree-optimization/83312
2859 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2860 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2861 "val" param, and to cope with arbitrary basic blocks.
2862 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2863 handle NULL_TREE for "val", dropping "bb" param.
2864 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2865 drop "bb" param. Handle NULL_TREE for "val".
2866 (find_case_label_for_value): Make "switch_stmt" param const.
2867 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2869 (vrp_prop::check_all_array_refs): Reimplement as...
2870 (check_array_bounds_dom_walker::before_dom_children): ...this new
2871 vfunc. Replace linear search through BB block list, excluding
2872 those with non-executable in-edges via dominator walk.
2874 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2876 * config/arm/arm.opt (mverbose-cost-dump): New option.
2877 * config/arm/arm.c (arm_rtx_costs): Use it.
2879 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2882 * reload1.c (emit_input_reload_insns): Skip debug markers.
2884 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2886 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2890 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2891 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2892 * config/sh/sh.c: Skip debug insns besides notes.
2893 * config/sh/sh.md: Likewise.
2894 * config/sh/sh_treg_combine.cc: Likewise.
2895 * config/sh/sync.md: Likewise.
2897 2017-12-14 Tom de Vries <tom@codesourcery.com>
2899 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2900 Add item for weak_undefined.
2902 2017-12-14 Richard Biener <rguenther@suse.de>
2904 PR tree-optimization/67842
2905 * sese.h (bb_in_region): Remove #if 0'ed code.
2907 2017-12-14 Richard Biener <rguenther@suse.de>
2909 PR tree-optimization/83326
2910 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2911 may_be_zero parameter and handle it by not marking the first
2912 peeled copy as not exiting the loop.
2913 (try_peel_loop): Likewise.
2914 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2915 to handle the case of constant or zero iterations and perform
2916 loop header copying on-the-fly.
2918 2017-12-14 Richard Biener <rguenther@suse.de>
2920 PR tree-optimization/83418
2921 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2922 Instead of asserting we don't get unfolded comparisons deal with them.
2924 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2927 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2928 basic blocks. Assert debug bind insns don't appear outside of bbs,
2929 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2932 PR tree-optimization/83198
2933 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2934 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2935 value if arg is a REAL_CST with incompatible type.
2937 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2938 Bin Cheng <bin.cheng@arm.com>
2941 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2943 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2946 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2948 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2949 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2951 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2952 Alan Hayward <alan.hayward@arm.com>
2953 David Sherwood <david.sherwood@arm.com>
2955 * poly-int.h: New file.
2956 * poly-int-types.h: Likewise.
2957 * coretypes.h: Include them.
2958 (POLY_INT_CONVERSION): Define.
2959 * target.def (estimated_poly_value): New hook.
2960 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2961 * doc/tm.texi: Regenerate.
2962 * doc/poly-int.texi: New file.
2963 * doc/gccint.texi: Include it.
2964 * doc/rtl.texi: Describe restrictions on subreg modes.
2965 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2966 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2967 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2968 * targhooks.h (default_estimated_poly_value): Declare.
2969 * targhooks.c (default_estimated_poly_value): New function.
2970 * target.h (estimated_poly_value): Likewise.
2971 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2972 (wi::unary_traits): Delete.
2973 (wi::binary_traits::signed_shift_result_type): Define for
2974 offset_int << HOST_WIDE_INT, etc.
2975 (generic_wide_int::operator <<=): Define for all types and use
2976 wi::lshift instead of <<.
2977 (wi::hwi_with_prec): Add a default constructor.
2978 (wi::ints_for): New class.
2979 (operator <<): Define for all wide-int types.
2980 (operator /): New function.
2981 (operator %): Likewise.
2982 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2983 (ASSERT_MAYBE_NE_AT): New macros.
2985 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2986 Dominik Vogt <vogt@linux.vnet.ibm.com>
2989 * emit-rtl.c (init_emit): Remove ??? comment.
2990 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2991 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2992 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2993 alignment of 3 virtual registers to BITS_PER_WORD.
2995 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2997 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2999 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3000 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3002 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3005 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3007 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3009 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3010 with the exception of debug begin stmt markers appear before
3014 * final.c (rest_of_handle_final): Call variable_tracking_main only
3015 if !flag_var_tracking.
3017 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3018 Jakub Jelinek <jakub@redhat.com>
3022 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3023 labels debug stmts that can only appear after labels.
3025 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3027 PR rtl-optimization/82398
3028 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3029 EXPR_USEFULNESS in priority comparison.
3031 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3033 PR rtl-optimization/83393
3034 * combine.c (move_deaths): If reg_stat points to a too new insn in
3035 last_death, do not use it: find the proper insn instead.
3037 2017-12-12 Jeff Law <law@redhat.com>
3039 PR tree-optimization/83298
3040 PR tree-optimization/83362
3041 PR tree-optimization/83383
3042 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3043 push_value_range a public interface. Add new argument to
3044 record_ranges_from_stmt.
3045 * gimple-ssa-evrp-analyze.c
3046 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3047 Update comments. Handle recording temporary equivalences.
3048 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3049 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3050 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3051 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3052 gimple-ssa-evrp-analyze.h.
3053 (record_temporary_equivalences_from_phis): Add new argument. When
3054 the PHI arg is an SSA_NAME, set the result's range to the range
3056 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3057 from statements too.
3058 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3059 Pass it down to children as needed.
3060 (thread_outgoing_edges): Likewise.
3061 (thread_across_edge): Likewise. Push/pop range state as needed.
3062 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3064 2017-12-12 Julia Koval <julia.koval@intel.com>
3066 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3067 (PTA_CANNONLAKE): Remove PTA_CLWB.
3069 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3072 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3073 (vcondv2div2df): Likewise.
3074 (vconduv2dfv2di): Likewise.
3076 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3078 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3079 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3080 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3081 ATTR_MATHFN_FPROUNDING_ERRNO.
3083 2017-12-12 Richard Biener <rguenther@suse.de>
3085 PR tree-optimization/83385
3086 * graphite-scop-detection.c (get_order, order): Remove.
3087 (bb_to_rpo): New global.
3089 (build_scops): Sort pbbs in RPO order.
3091 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3093 * combine.c (simplify_set): Do not transform subregs to zero_extends
3094 if the destination mode is a vector mode.
3096 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3098 PR tree-optimization/83359
3099 * tree-cfg.h (fold_loop_internal_call): Declare.
3100 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3101 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3102 (find_loop_dist_alias): New function.
3103 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3104 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3107 PR tree-optimization/80631
3108 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3109 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3110 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3111 hardcoding zero as the value if COND_EXPR is never true. For
3112 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3113 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3114 hardcoding MAX_EXPR as the reduction operation.
3115 (is_nonwrapping_integer_induction): Allow negative step.
3116 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3117 vect_create_epilog_for_reduction, if no value is suitable, don't
3118 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3120 2017-12-12 Richard Biener <rguenther@suse.de>
3122 PR tree-optimization/81889
3123 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3124 range info from the non-wrapping IV instead of just the range
3127 2017-12-12 Julia Koval <julia.koval@intel.com>
3129 * config.gcc: Add vaesintrin.h.
3130 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3132 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3133 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3135 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3136 * config/i386/immintrin.h: Include vaesintrin.h.
3137 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3138 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3139 _mm_aesdec_epi128): New intrinsics.
3141 2017-12-12 Julia Koval <julia.koval@intel.com>
3143 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3144 OPTION_MASK_ISA_VAES_UNSET): New.
3145 (ix86_handle_option): Handle -mvaes.
3146 * config/i386/cpuid.h: Define bit_VAES.
3147 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3148 * config/i386/i386-c.c (__VAES__): New.
3149 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3150 (ix86_valid_target_attribute_inner_p): Ditto.
3151 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3152 * config/i386/i386.opt: Add -mvaes.
3153 * doc/invoke.texi: Ditto.
3155 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3157 * debug.h (gcc_debug_hooks): Add inline_entry.
3158 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3159 * debug.c (do_nothing_debug_hooks): Likewise.
3160 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3161 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3162 (dwarf2_lineno_debug_hooks): Likewise.
3164 * common.opt (gstatement-frontiers): New, setting
3165 debug_nonbind_markers_p.
3166 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3167 * toplev.c (process_options): Autodetect value for debug statement
3169 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3170 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3172 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3173 markers. Integrate source bind into debug stmt expand loop.
3174 (pass_expand::execute): Check debug marker limit. Avoid deep
3175 TER and expand debug locations for debug bind insns only.
3176 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3178 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3179 * final.c (reemit_insn_block_notes): Take current block from
3180 nonbind markers. Declare note where it's first set.
3181 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3182 begin stmt markers if enabled.
3183 (notice_source_line): Handle nonbind markers. Fail if their
3184 location is unknown or that of builtins.
3185 (rest_of_handle_final): Convert begin stmt markers to notes if
3186 var-tracking didn't run.
3187 (rest_of_clean_state): Skip begin stmt markers.
3188 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3190 * function.c (allocate_struct_function): Set begin_stmt_markers.
3191 * function.h (struct function): Add debug_marker_count counter
3192 and debug_nonbind_markers flag.
3193 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3194 * gimple-low.c (lower_function_body): Adjust
3195 debug_nonbind_markers.
3196 (lower_stmt): Drop or skip gimple debug stmts.
3197 (lower_try_catch): Skip debug stmts.
3198 * gimple.c (gimple_build_debug_begin_stmt): New.
3199 (gimple_copy): Increment debug_marker_count if copying one.
3200 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3201 * gimplify.c (rexpr_location): New.
3202 (rexpr_has_location): New.
3203 (warn_switch_unreachable_r): Handle gimple debug stmts.
3204 (shortcut_cond_r): Call expr_location.
3206 (find_goto_label): New.
3207 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3209 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3211 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3212 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3213 (LANG_HOOKS_INITIALIZER): ... this.
3214 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3215 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3217 (update_ebb_live_info): Skip debug insn markers.
3218 * lra.c (debug_insn_static_data): Rename to...
3219 (debug_bind_static_data): ... this.
3220 (debug_marker_static_data): New.
3221 (lra_set_insn_recog_data): Select one of the above depending
3223 (lra_update_isn_regno_info): Don't assume debug insns have
3225 (push_insns): Skip debug insns.
3226 * lto-streamer-in.c (input_function): Drop debug stmts
3227 depending on active options. Adjust debug_nonbind_markers.
3228 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3229 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3230 begin stmt marker notes.
3231 (print_insn): Likewise.
3232 * recog.c (extract_insn): Recognize rtl for debug markers.
3233 * rtl.def (DEBUG_MARKER): New.
3234 * tree-inline.c: Include params.h.
3235 (remap_gimple_stmt): Handle nonbind markers.
3236 (maybe_move_debug_stmts_to_successors): Likewise.
3237 (copy_debug_stmt): Likewise.
3238 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3239 markers regardless of no side effects.
3240 (tsi_link_before): Don't update container's side effects when adding
3241 a begin stmt marker.
3242 (tsi_link_after): Likewise.
3243 (expr_first): Skip begin stmt markers.
3244 (expr_last): Likewise.
3245 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3246 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3247 Disregard nonbind markers.
3248 * tree.c (make_node_stat): Don't set side effects for begin stmt
3250 (build1_stat): Likewise.
3251 * tree.def (DEBUG_BEGIN_STMT): New.
3252 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3253 * var-tracking.c (delete_debug_insns): Renamed to...
3254 (delete_vta_debug_insns): ... this.
3255 (reemit_marker_as_note): New.
3256 (vt_initialize): Reemit markers.
3257 (delete_vta_debug_insns): Likewise.
3258 (vt_debug_insns_local): Reemit or delete markers.
3259 (variable_tracking_main_1): Likewise.
3260 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3261 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3262 (gimple_debug_nonbind_marker_p): New.
3263 (gimple_build_debug_bind): Adjust.
3264 (gimple_build_debug_begin_stmt): New.
3265 * doc/invoke.texi (max-debug-marker-count): New param.
3266 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3267 (debug_parameter_ref, debug_marker): New.
3268 (NOTE_INSN_BEGIN_STMT): New.
3269 (DEBUG_INSN): Describe begin stmt markers.
3271 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3272 without debug insns, we wouldn't, but clean up debug insns
3273 after a control flow insn nevertheless.
3275 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3276 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3277 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3278 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3279 (rtl_tidy_fallthru_edge): Likewise.
3280 (rtl_verify_fallthru): Likewise.
3281 (rtl_verify_bb_layout): Likewise.
3282 (skip_insns_after_block): Likewise.
3283 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3284 * dwarf2out.c: Include print-rtl.h.
3285 (dwarf2out_next_real_insn): New.
3286 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3287 Dump debug binds in asm comments.
3288 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3289 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3290 callers to use gsi_start_nondebug_bb instead.
3291 (gsi_after_labels): Skip gimple debug stmts.
3292 (gsi_start_nondebug): New.
3293 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3294 (proper_loop_form_for_interchange): Adjust.
3295 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3296 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3297 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3298 (collect_fallthrough_labels): Likewise.
3299 (should_warn_for_implicit_fallthrough): Likewise.
3300 (warn_implicit_fallthrough_r): Likewise.
3301 (expand_FALLTHROUGH_r): Likewise.
3302 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3303 (graphite_copy_stmts_from_block): Skip nonbind markers.
3304 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3305 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3306 * jump.c (clean_barriers): Skip debug insns.
3307 * omp-expand.c (expand_parallel_call): Skip debug insns.
3308 (expand_task_call): Likewise.
3309 (remove_exit_barrier): Likewise.
3310 (expand_omp_taskreg): Likewise.
3311 (expand_omp_for_init_counts): Likewise.
3312 (expand_omp_for_generic): Likewise.
3313 (expand_omp_for_static_nochunk): Likewise.
3314 (expand_omp_for_static_chunk): Likewise.
3315 (expand_omp_simd): Likewise.
3316 (expand_omp_taskloop_for_outer): Likewise.
3317 (expand_omp_taskloop_for_inner): Likewise.
3318 (expand_oacc_for): Likewise.
3319 (expand_omp_sections): Likewise.
3320 (expand_omp_single): Likewise.
3321 (expand_omp_synch): Likewise.
3322 (expand_omp_atomic_load): Likewise.
3323 (expand_omp_atomic_store): Likewise.
3324 (expand_omp_atomic_fetch_op): Likewise.
3325 (expand_omp_atomic_pipeline): Likewise.
3326 (expand_omp_atomic_mutex): Likewise.
3327 (expand_omp_target): Likewise.
3328 (grid_expand_omp_for_loop): Likewise.
3329 (grid_expand_target_grid_body): Likewise.
3330 (build_omp_regions_1): Likewise.
3331 * omp-low.c (check_combined_parallel): Skip debug stmts.
3332 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3333 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3334 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3336 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3337 (make_edges): Likewise.
3338 (cleanup_dead_labels): Likewise.
3339 (gimple_can_merge_blocks_p): Likewise.
3340 (stmt_starts_bb_p): Likewise.
3341 (gimple_block_label): Likewise.
3342 (gimple_redirect_edge_and_branch): Likewise.
3343 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3345 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3347 * tree-pretty-print (print_declaration): Omit initializer in slim
3349 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3351 (eliminate_unnecessary_stmts): Stabilize block removal order.
3352 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3353 * var-tracking.c (get_first_insn): New.
3354 (vt_emit_notes): Call it.
3355 (vt_initialize): Walk any insns before the first BB.
3356 (delete_debug_insns): Likewise.
3358 * gimple.h (enum gimple_debug_subcode): Add
3359 GIMPLE_DEBUG_BEGIN_STMT.
3360 (gimple_debug_begin_stmt_p): New.
3361 (gimple_debug_nonbind_marker_p): New.
3362 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3363 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3364 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3365 * insn-notes.def (BEGIN_STMT): New.
3366 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3367 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3368 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3369 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3370 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3371 (INSN_DEBUG_MARKER_KIND): New.
3372 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3373 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3374 (INSN_VAR_LOCATION_PTR): New.
3375 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3377 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3378 deps only if debug bind insns are enabled.
3379 (pass_expand::execute): Avoid deep TER and expand
3380 debug locations for debug bind insns only.
3381 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3382 debug stmts special handling down to debug bind stmts.
3383 * combine.c (try_combine): Narrow debug insns special handling
3384 down to debug bind insns.
3385 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3386 Narrow debug insns preexisting special handling down to debug
3388 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3389 handling down to debug bind insns.
3390 * function.c (instantiate_virtual_regs): Skip debug markers,
3391 adjust handling of debug binds.
3392 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3393 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3394 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3395 insns to debug bind insns.
3396 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3397 special handling of debug stmts to debug bind stmts.
3398 * ipa-split.c (split_function): Likewise.
3399 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3400 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3402 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3403 * regrename.c (build_def_use): Likewise.
3404 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3405 (pass_cprop_hardreg): Narrow special casing of debug insns to
3407 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3408 * reload1.c (reload): Likewise.
3409 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3410 casing of debug stmts to debug bind stmts.
3411 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3412 * ssa-iterators.h (num_imm_uses): Likewise.
3413 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3414 debug stmts to debug bind stmts.
3415 * tree-inline.c (tree_function_versioning): Narrow special casing
3416 of debug stmts to debug bind stmts.
3417 * tree-loop-distribution.c (generate_loops_for_partition):
3418 Narrow special casing of debug stmts to debug bind stmts.
3419 * tree-sra.c (analyze_access_subtree): Narrow special casing
3420 of debug stmts to debug bind stmts.
3421 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3422 stmts to debug bind stmts.
3423 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3424 casing of debug stmts to debug bind stmts.
3425 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3426 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3427 casing of debug stmts to debug bind stmts.
3428 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3430 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3431 debug stmts to debug bind stmts.
3432 (gimple_replace_ssa_lhs): Likewise.
3433 (insert_debug_temp_for_var_def): Likewise.
3434 (insert_debug_temps_for_defs): Likewise.
3435 (reset_debug_uses): Likewise.
3436 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3437 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3438 (adjust_debug_stmts): Likewise.
3439 (adjust_phi_and_debug_stmts): Likewise.
3440 (vect_do_peeling): Likewise.
3441 * tree-vect-loop.c (vect_transform_loop): Likewise.
3442 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3443 * var-tracking.c (adjust_mems): Narrow special casing of debug
3444 insns to debug bind insns.
3445 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3446 (compute_bb_dataflow, vt_find_locations): Likewise.
3447 (vt_expand_loc, emit_notes_for_changes): Likewise.
3448 (vt_init_cfa_base): Likewise.
3449 (vt_emit_notes): Likewise.
3450 (vt_initialize): Likewise.
3451 (vt_finalize): Likewise.
3453 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3454 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3455 (next_nonnote_nondebug_insn_bb): New.
3456 (prev_nonnote_nondebug_insn_bb): New.
3457 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3458 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3459 (prev_nonnote_nondebug_insn_bb): Declare.
3460 (next_nonnote_nondebug_insn_bb): Declare.
3461 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3462 * cfgrtl.c (get_last_bb_insn): Likewise.
3463 * lra.c (push_insns): Likewise.
3465 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3468 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3470 * selftest.c (selftest::test_runner::test_runner): New ctor.
3471 (selftest::test_runner::~test_runner): New dtor.
3472 * selftest.h (class selftest::test_runner): New class.
3474 2017-12-11 Carl Love <cel@us.ibm.com>
3476 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3477 vec_extract_fp32_from_shortl]): Add #defines.
3478 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3479 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3480 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3481 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3482 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3483 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3484 * doc/extend.texi: Add documentation for the added builtins.
3486 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3488 PR rtl-optimization/80693
3489 PR rtl-optimization/81019
3490 PR rtl-optimization/81020
3491 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3492 are not mentioned in i3. Place the REG_UNUSED note on i2,
3493 possibly modified to REG_DEAD, if it did not originate in i3.
3495 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3497 * recog.c (store_data_bypass_p_1): New function.
3498 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3499 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3501 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3503 PR rtl-optimization/83361
3504 * ifcvt.c (if_convert): Call fixup_partitions.
3506 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3508 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3509 early folding of splat_u{8,16,32}.
3511 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3513 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3514 output_operand_lossage first argument with capital letter.
3515 (aarch64_override_options): Don't start error and sorry first argument
3516 with capital letter.
3518 2017-12-11 Andi Kleen <ak@linux.intel.com>
3520 PR gcov-profile/83355
3521 * auto-profile.c (string_table::get_index_by_decl): Don't
3522 recurse when abstract origin points to itself.
3524 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3526 PR tree-optimization/83320
3527 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3528 (prune_datarefs_not_in_loop): Ditto.
3530 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3532 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3534 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3536 PR tree-optimization/83337
3537 * gimple-loop-interchange.cc (compute_access_stride): Handle
3538 bitfield DRs properly.
3540 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3542 PR tree-optimization/83338
3543 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3544 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3545 vector of unsigned integers to vector of signed integers.
3547 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3549 PR rtl-optimization/83317
3550 * lra-constraints.c (process_address_1): Add insn code check.
3552 2017-12-08 Michael Matz <matz@suse.de>
3554 Fix PR tree-optimization/83323
3555 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3556 head-controlled loops and loop BBs.
3557 * common.opt (funroll-and-jam): Remove, instead ...
3558 (floop-unroll-and-jam): ... reuse this option.
3559 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3560 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3561 (-floop-unroll-and-jam): ... this option.
3563 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3565 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3566 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3567 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3569 2017-12-08 Richard Biener <rguenther@suse.de>
3572 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3573 handle accesses outside of zero-sized vars.
3575 2017-12-08 Martin Jambor <mjambor@suse.cz>
3577 PR tree-optimization/83141
3578 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3579 test for MEM_REFs implicitely changing types with padding. Remove
3581 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3583 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3585 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3587 * config/avr/avr.c (avr_attribute_table): Likewise.
3588 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3589 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3590 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3591 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3592 of struct attribute_spec.
3593 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3595 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3598 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3600 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3603 * profile-count.c (profile_count::from_gcov_type): Move from
3604 profile-count.h; handle overflow.
3605 * profile-count.h (profile_count::from_gcov_type): Move offline.
3607 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3609 PR rtl-optimization/83304
3610 * combine.c (move_deaths): If we do not know where a register died,
3613 2017-12-08 Richard Biener <rguenther@suse.de>
3615 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3616 Provide -fopt-info-loop feedback when we interchange in a nest.
3618 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3620 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3621 for armv6 ARM CPU IDs.
3623 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3625 * common/config/arm/arm-common.c: Include <algorithm>.
3626 (INCLUDE_VECTOR): Define.
3627 (compare_opt_names): New function.
3628 (arm_rewrite_selected_arch): Only strip out extensions that can be
3629 expressed through -mfpu. Sort the remaining extensions
3632 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3634 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3635 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3636 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3637 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3639 2017-06-08 Tristan Gingold <gindold@adacore.com>
3642 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3643 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3645 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3647 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3648 of struct attribute_spec.
3650 2017-12-08 Julia Koval <julia.koval@intel.com>
3652 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3653 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3654 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3655 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3656 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3657 _mm_maskz_dpwssds_epi32): Ditto.
3659 2017-12-08 Richard Biener <rguenther@suse.de>
3661 PR tree-optimization/81303
3662 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3663 conditions try to create a comparison vector type matching
3664 the data vector type.
3665 (vectorizable_condition): Adjust.
3666 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3667 Leave invariant conditions alone in case we can vectorize those.
3669 2017-12-08 Julia Koval <julia.koval@intel.com>
3671 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3672 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3673 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3674 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3675 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3677 2017-12-08 Julia Koval <julia.koval@intel.com>
3679 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3680 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3681 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3682 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3683 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3684 _mm_maskz_dpbusds_epi32): New intrinsics.
3686 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3688 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3690 (ld<bh>io_signed): Likewise.
3691 (st<bhw>io): Likewise.
3692 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3693 SMALL_INT12 constant integer operand.
3695 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3697 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3699 (nios2_large_constant_allowed): ...this. Adjust uses.
3700 (nios2_plus_symbolic_constant_p): Rename to...
3701 (nios2_plus_large_constant_p): ...this. Adjust uses.
3702 (nios2_legitimate_address_p): Correct CONST_INT handling.
3703 (nios2_symbolic_memory_operand_p): Rename to...
3704 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3705 (nios2_large_constant_p): Check for large constant integers too.
3706 (nios2_split_large_constant): Handle constant integers.
3707 (nios2_split_symbolic_memory_operand): Rename to...
3708 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3709 (nios2_legitimize_constant_address): Handle constant integers.
3710 (r0rel_constant_p): Handle small constant integers.
3711 (nios2_print_operand_address): Handle r0-relative integer addresses.
3712 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3713 * config/nios2/nios2.md: Adjust for renamed functions.
3715 2017-12-07 Andrew Waterman <andrew@sifive.com>
3717 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3718 (TARGET_HAVE_SRODATA_SECTION): New define.
3719 (riscv_select_section): New function.
3721 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3722 Alexander Monakov <amonakov@ispras.ru>
3723 Jakub Jelinek <jakub@redhat.com>
3726 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3728 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3730 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3731 a bool success value. Don't call output_operand_lossage here.
3732 (aarch64_print_ldpstp_address): Return a bool success value.
3733 (aarch64_print_operand_address): Call output_addr_const if
3734 aarch64_print_address_internal fails.
3735 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3736 'y'; call output_operand_lossage instead. Call output_operand_lossage
3737 if aarch64_print_ldpstp_address fails.
3739 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3741 * tree-vector-builder.h
3742 (tree_vector_builder::binary_encoded_nelts): Declare.
3743 * tree-vector-builder.c
3744 (tree_vector_builder::binary_encoded_nelts): New function.
3745 * fold-const.c (negate_expr_p): Likewise.
3746 (operand_equal_p, fold_checksum_tree): Likewise.
3747 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3748 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3749 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3750 (uniform_vector_p): Likewise.
3751 * varasm.c (const_hash_1, compare_constant): Likewise.
3752 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3753 (valid_lattice_transition): Operate directly on the VECTOR_CST
3755 * ipa-icf.c: Include tree-vector-builder.h.
3756 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3757 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3759 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3761 * tree.c (build_vector): Delete.
3762 * tree.h (build_vector): Make static and move into the self-testing
3765 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3767 * vector-builder.h (vector_builder::derived): New const overload.
3768 (vector_builder::elt): New function.
3769 * tree-vector-builder.h (tree_vector_builder::type): New function.
3770 (tree_vector_builder::apply_step): Declare.
3771 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3773 * gimple-fold.h (tree_vector_builder): Declare.
3774 (gimple_build_vector): Take a tree_vector_builder instead of a
3775 type and vector of elements.
3776 * gimple-fold.c (gimple_build_vector): Likewise.
3777 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3779 (get_initial_defs_for_reduction): Likewise.
3780 (vectorizable_induction): Likewise.
3782 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3784 * tree-vector-builder.h
3785 (tree_vector_builder::new_binary_operation): Declare.
3786 * tree-vector-builder.c
3787 (tree_vector_builder::new_binary_operation): New function.
3788 * fold-const.c (fold_relational_const): Use it.
3789 (const_binop): Likewise. Check that both input vectors have
3790 the same number of elements, thus excluding things like WIDEN_SUM.
3791 Check whether it is possible to operate directly on the encodings
3794 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3796 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3797 new_unary_operation, operating only on the encoded elements.
3798 (const_unop): Likewise.
3799 (exact_inverse): Likewise.
3800 (distributes_over_addition_p): New function.
3801 (const_binop): Use tree_vector_builder and new_unary_operation
3802 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3803 on the encoded elements unless the encoding is strided and the
3804 operation does not distribute over addition.
3805 (fold_convert_const): Use tree_vector_builder and
3806 new_unary_operation. Operate only on the encoded elements
3807 for truncating integer conversions, or for non-stepped encodings.
3809 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3811 * config/sparc/sparc.c: Include tree-vector-builder.h.
3812 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3813 * expmed.c: Include tree-vector-builder.h.
3814 (make_tree): Use tree_vector_builder instead of build_vector.
3815 * fold-const.c: Include tree-vector-builder.h.
3816 (const_binop): Use tree_vector_builder instead of build_vector.
3817 (const_unop): Likewise.
3818 (native_interpret_vector): Likewise.
3819 (fold_vec_perm): Likewise.
3820 (fold_ternary_loc): Likewise.
3821 * gimple-fold.c: Include tree-vector-builder.h.
3822 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3824 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3825 (simplify_vector_constructor): Use tree_vector_builder instead
3827 * tree-vect-generic.c: Include tree-vector-builder.h.
3828 (add_rshift): Use tree_vector_builder instead of build_vector.
3829 (expand_vector_divmod): Likewise.
3830 (optimize_vector_constructor): Likewise.
3831 * tree-vect-loop.c: Include tree-vector-builder.h.
3832 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3833 of build_vector. Explicitly use a stepped encoding for
3835 * tree-vect-slp.c: Include tree-vector-builder.h.
3836 (vect_get_constant_vectors): Use tree_vector_builder instead
3838 (vect_transform_slp_perm_load): Likewise.
3839 (vect_schedule_slp_instance): Likewise.
3840 * tree-vect-stmts.c: Include tree-vector-builder.h.
3841 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3842 (vect_gen_perm_mask_any): Likewise.
3843 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3844 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3846 (build_vector_from_val): Likewise. Explicitly use a duplicate
3849 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3851 * doc/generic.texi (VECTOR_CST): Describe new representation of
3853 * vector-builder.h: New file.
3854 * tree-vector-builder.h: Likewise.
3855 * tree-vector-builder.c: Likewise.
3856 * Makefile.in (OBJS): Add tree-vector-builder.o.
3857 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3858 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3859 (tree_vector): Change the number of elements to
3860 vector_cst_encoded_nelts.
3861 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3862 (VECTOR_CST_ELTS): Delete.
3863 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3864 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3865 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3866 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3867 (VECTOR_CST_ENCODED_ELT): Likewise.
3868 (vector_cst_encoded_nelts): New function.
3869 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3870 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3871 (vector_cst_int_elt, vector_cst_elt): Declare.
3872 * tree.c: Include tree-vector-builder.h.
3873 (tree_code_size): Abort if passed VECTOR_CST.
3874 (tree_size): Update for new VECTOR_CST layout.
3875 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3876 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3877 (build_vector): Use tree_vector_builder.
3878 (vector_cst_int_elt, vector_cst_elt): New functions.
3879 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3880 encoded elements and then create the vector in the canonical form.
3881 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3882 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3883 (tree_c_tests): Call test_vector_cst_patterns.
3884 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3886 (hash_tree): Likewise.
3887 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3888 (streamer_write_tree_header): Likewise.
3889 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3890 (streamer_alloc_tree): Likewise. Update call to make_vector.
3891 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3893 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3895 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3896 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3897 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3899 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3900 Richard Biener <rguenther@suse.de>
3902 PR tree-optimization/81303
3903 * Makefile.in (gimple-loop-interchange.o): New object file.
3904 * common.opt (floop-interchange): Reuse the option from graphite.
3905 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3906 -floop-interchange and mention it for -O3.
3907 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3908 * gimple-loop-interchange.cc: New file.
3909 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3910 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3911 * passes.def (pass_linterchange): New pass.
3912 * timevar.def (TV_LINTERCHANGE): New time var.
3913 * tree-pass.h (make_pass_linterchange): New declaration.
3914 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3915 interchange. Record IV before/after increment in new parameters.
3916 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3917 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3919 (check_reduction_path): ...New function here.
3920 * tree-vectorizer.h (check_reduction_path): New declaration.
3922 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3925 PR rtl-optimization/80818
3926 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3927 always early clobbered.
3928 * lra-lives.c (process_bb_lives): Check input hard regs for early
3929 clobbered non-operand hard reg.
3931 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3934 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3935 types_compatible_p, just that TYPE_MODE is the same.
3937 2017-12-07 Martin Sebor <msebor@redhat.com>
3940 * attribs.c (empty_attribute_table): Initialize new member of
3941 struct attribute_spec.
3942 (decl_attributes): Add argument. Handle mutually exclusive
3943 combinations of attributes.
3944 (selftests::test_attribute_exclusions): New function.
3945 (selftests::attribute_c_tests): Ditto.
3946 * attribs.h (decl_attributes): Add default argument.
3947 * selftest.h (attribute_c_tests): Declare.
3948 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3949 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3951 * doc/extend.texi (Common Function Attributes): Update const and pure.
3952 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3953 of struct attribute_spec.
3954 * config/arc/arc.c (arc_attribute_table): Same.
3955 * config/arm/arm.c (arm_attribute_table): Same.
3956 * config/avr/avr.c ( avr_attribute_table): Same.
3957 * config/bfin/bfin.c (bfin_attribute_table): Same.
3958 * config/cr16/cr16.c (cr16_attribute_table): Same.
3959 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3960 * config/h8300/h8300.c (h8300_attribute_table): Same.
3961 * config/i386/i386.c (ix86_attribute_table): Same.
3962 * config/ia64/ia64.c (ia64_attribute_table): Same.
3963 * config/m32c/m32c.c (m32c_attribute_table): Same.
3964 * config/m32r/m32r.c (m32r_attribute_table): Same.
3965 * config/m68k/m68k.c (m68k_attribute_table): Same.
3966 * config/mcore/mcore.c (mcore_attribute_table): Same.
3967 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3968 * config/mips/mips.c (mips_attribute_table): Same.
3969 * config/msp430/msp430.c (msp430_attribute_table): Same.
3970 * config/nds32/nds32.c (nds32_attribute_table): Same.
3971 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3972 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3973 * config/rl78/rl78.c (rl78__attribute_table): Same.
3974 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3975 * onfig/rx/rx.c (rx_attribute_table): Same.
3976 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3977 * config/sh/sh.c (sh_attribute_table): Same.
3978 * config/sparc/sparc.c (sparc_attribute_table): Same.
3979 * config/spu/spu.c (spu_attribute_table): Same.
3980 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3981 * config/v850/v850.c (v850_attribute_table): Same.
3982 * config/visium/visium.c (visium_attribute_table): Same.
3984 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3987 * config/arm/arm.c (INCLUDE_STRING): Define.
3988 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3989 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3992 2017-12-07 Michael Matz <matz@suse.de>
3994 Add unroll and jam pass
3996 * gimple-loop-jam.c: New file.
3997 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3998 * common.opt (funroll-and-jam): New option.
3999 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4000 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4001 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4002 * passes.def: Add pass_loop_jam.
4003 * timevar.def (TV_LOOP_JAM): Add.
4004 * tree-pass.h (make_pass_loop_jam): Declare.
4005 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4006 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4007 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4008 to flow_loop_tree_node_add.
4009 (duplicate_subloops, copy_loops_to): Append to sibling list.
4010 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4011 * doc/invoke.texi (-funroll-and-jam): Document new option.
4012 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4014 2017-12-07 Richard Biener <rguenther@suse.de>
4016 PR tree-optimization/83296
4017 PR tree-optimization/67769
4018 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4019 flow sensitive info in an unrelated BB.
4020 (value_replacement): Use reset_flow_sensitive_info.
4021 (minmax_replacement): Reset flow sensitive info on the def
4022 we move. Do not reset flow sensitive info in the whole BB
4023 we move the stmt to.
4024 (abs_replacement): Likewise.
4026 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4029 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4030 rs6000_cpu to the given -mcpu=, or to the default processor.
4032 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4034 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4035 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4036 instead of rs6000_cpu_attr.
4037 (rs6000_adjust_cost): Ditto.
4038 (is_microcoded_insn): Ditto.
4039 (rs6000_adjust_priority): Ditto.
4040 (rs6000_issue_rate): Ditto.
4041 (rs6000_use_sched_lookahead): Ditto.
4042 (rs6000_use_sched_lookahead_guard): Ditto.
4043 (rs6000_sched_reorder): Ditto.
4044 (force_new_group): Ditto.
4045 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4046 (group_ending_nop): Ditto.
4048 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4050 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4051 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4052 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4053 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4055 (rs6000_reassociation_width): Ditto.
4056 (rs6000_emit_epilogue): Ditto.
4057 (rs6000_adjust_cost): Ditto.
4058 (is_microcoded_insn): Ditto.
4059 (is_cracked_insn): Ditto.
4060 (rs6000_adjust_priority): Ditto.
4061 (rs6000_sched_reorder): Ditto.
4062 (rs6000_sched_reorder2): Ditto.
4063 (insn_must_be_first_in_group): Ditto.
4064 (insn_must_be_last_in_group): Ditto.
4065 (rs6000_register_move_cost): Ditto.
4066 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4069 2017-12-07 Julia Koval <julia.koval@intel.com>
4071 * config.gcc: Add vaesintrin.h.
4072 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4074 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4075 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4077 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4078 * config/i386/immintrin.h: Include vaesintrin.h.
4079 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4080 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4081 _mm_aesdec_epi128): New intrinsics.
4083 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4085 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4086 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4087 c-family/c-spellcheck.cc.
4088 (best_macro_match::best_macro_match): Likewise.
4090 (struct edit_distance_traits<cpp_hashnode *>): Move to
4091 c-family/c-spellcheck.h.
4092 (class best_macro_match): Likewise.
4094 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4096 PR tree-optimization/83293
4097 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4098 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4099 might insert into empty bb.
4102 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4103 simplify for plus with :c added, and pointer_plus without that.
4104 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4105 with undefined overflow and the conversion is not widening,
4106 perform negation in utype and only convert to type afterwards.
4107 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4108 simplify for plus with :c added, and pointer_plus without that.
4109 If type is integral with undefined overflow and the conversion is
4110 not widening, perform minus in utype and only convert to type
4111 afterwards. Move the last pointer_diff_expr simplify into the
4114 2017-12-06 Martin Sebor <msebor@redhat.com>
4116 PR tree-optimization/82646
4117 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4118 strncpy, not maxlen.
4120 2017-12-06 Martin Sebor <msebor@redhat.com>
4122 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4125 PR tree-optimization/83075
4126 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4127 strncat/strncpy don't change length of source string.
4129 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4132 2017-11-29 Martin Aberg <maberg@gaisler.com>
4134 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4135 to prevent b2bst errata sequence.
4136 (sqrtdf2_fix): Likewise.
4138 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4140 PR tree-optimization/81945
4141 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4142 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4143 to dest_cfun has orig_loop_num set, either remap it to the new
4144 loop number if the loop got moved too, or clear it.
4146 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4148 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4150 (thunderx2t99_nothing): New insn reservation.
4151 (thunderx2t99_mrs): New insn reservation.
4152 (thunderx2t99_multiple): New insn reservation.
4153 (thunderx2t99_alu_basi): Add bfx to reservation.
4154 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4156 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4159 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4161 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4163 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4164 tree-ssa-pre.c::remove_dead_inserted_code.
4165 * tree-ssa-dce.h: New file.
4166 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4167 (remove_dead_inserted_code): Move and rename to function
4168 tree-ssa-dce.c::simple_dce_from_worklist.
4169 (pass_pre::execute): Update use.
4171 2017-12-05 Richard Biener <rguenther@suse.de>
4173 PR tree-optimization/83277
4174 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4175 to code-gen liveout vars.
4177 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4179 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4180 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4181 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4184 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4186 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4188 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4189 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4192 2017-12-05 Richard Biener <rguenther@suse.de>
4194 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4195 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4196 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4197 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4199 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4201 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4203 2017-12-05 Martin Liska <mliska@suse.cz>
4204 Jakub Jelinek <jakub@redhat.com>
4206 * doc/invoke.texi: Document the options.
4207 * flag-types.h (enum sanitize_code): Add
4208 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4209 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4210 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4211 * opts.c: Define new sanitizer options.
4212 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4213 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4215 2017-12-05 Julia Koval <julia.koval@intel.com>
4217 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4218 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4219 (ix86_handle_option): Handle -mavx512vnni.
4220 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4221 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4222 * config/i386/i386-c (__AVX512VNNI__): New.
4223 * config/i386/i386.c (ix86_target_string): Handle new option.
4224 (ix86_valid_target_attribute_inner_p): Handle new option.
4225 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4226 * config/i386/i386.opt (mavx512vnni): New option.
4228 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4231 * athlon.md: Disable for generic.
4232 * haswell.md: Enable for generic.
4233 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4234 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4236 (ix86_adjust_cost): Move generic to haswell path.
4238 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4240 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4241 instead of MEM_P in a couple more places. Fix formatting issues.
4243 2017-12-04 Jim Wilson <jimw@sifive.com>
4245 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4246 instead of GP_REG_LAST-1.
4247 (riscv_adjust_libcall_cfi_prologue): Likewise.
4248 (riscv_adjust_libcall_cri_epilogue): Likewise.
4249 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4252 2017-12-04 Luis Machado <luis.machado@linaro.org>
4254 * ipa-pure-const.c (check_decl): Add missing newline.
4255 (state_from_flags): Likewise.
4257 2017-12-04 Jeff Law <law@redhat.com>
4259 PR tree-optimizatin/78496
4260 * gimple-ssa-evrp-analyze.h
4261 (evrp_range_analyzer::get_vr_values): Simplify.
4262 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4263 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4264 and gimple-ssa-evrp-analyze.h.
4265 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4266 (simplify_stmt_for_jump_threading): Copy a blob of code from
4267 tree-vrp.c to use ranges to simplify statements.
4268 (dom_opt_dom_walker::before_dom_children): Call
4269 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4270 (dom_opt_dom_walker::after_dom_children): Similarly for
4271 evrp_range_analyzer::leave.
4272 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4275 * gimple-ssa-evrp-analyze.c
4276 (evrp_range_analyzer::extract_range_from_stmt): Always use
4277 vr_values::update_value_range so preexisting range info is
4278 medged with new range info, even if the new range is VR_VARYING.
4280 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4282 * combine.c: Adjust comment.
4283 (use_crosses_set_p): Delete.
4284 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4285 (try_combine): Ditto.
4287 2017-12-04 Richard Biener <rguenther@suse.de>
4289 PR tree-optimization/83255
4290 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4291 Re-add zero-iteration check.
4293 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4295 PR rtl-optimization/83245
4296 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4297 hard registers as earlyclobber, also if not in an asm.
4299 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4303 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4306 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4307 rs6000_cpu based on cpu_index, not tune_index.
4309 2017-12-04 Richard Biener <rguenther@suse.de>
4311 PR tree-optimization/83238
4312 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4313 code match comment, rejecting invalid SESE regions.
4315 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4317 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4318 require base operand is a REG_POINTER prior to reload on targets
4319 with non-equivalent space registers.
4321 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4323 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4324 (update_specialized_profile): Fix updating of counts.
4325 (perhaps_add_new_callers): Likewise.
4327 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4330 * x86-tune.def: Remove obsolette FIXMEs.
4331 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4332 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4333 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4335 (X86_TUNE_PAD_RETURNS): Disable for generic.
4337 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4339 PR tree-optimization/83170
4340 PR tree-optimization/83241
4341 * gimple-ssa-store-merging.c
4342 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4343 gimple_vuse (ins_stmt) in case it has changed.
4344 (imm_store_chain_info::output_merged_store): Likewise.
4346 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4350 * tree-cfg.c (pass_warn_function_return::execute): Handle
4351 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4352 with BUILTINS_LOCATION.
4356 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4357 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4358 instead of DECL_MODE.
4360 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4361 last argument to rtx pointer.
4362 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4363 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4364 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4365 depending on the chosen ISAs.
4366 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4367 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4368 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4370 * config/i386/sse.md (mov<mode>_internal): Likewise.
4371 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4373 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4375 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4377 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4379 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4380 parameters from prototype.
4381 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4382 parameters. Don't print a comment.
4383 (emit_fusion_gpr_load): Adjust.
4384 (emit_fusion_load_store): Adjust.
4385 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4386 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4387 comment on the second line.
4389 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4392 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4393 rs6000_cpu based on cpu_index, not tune_index.
4395 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4397 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4398 which_alternative instead of which_alternative + 1.
4399 (output_asm_insn): Print an extra tab if the template is short.
4401 2017-12-01 Jim Wilson <jimw@sifive.com>
4403 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4405 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4406 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4408 * doc/tm.texi: Regenerate.
4410 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4413 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4414 whether we can allocate pseudos before trying to fix an address.
4415 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4416 memory address is indexed or indirect.
4417 (floatuns_<mode>si2_hw2): Likewise.
4419 2017-12-01 Jason Merrill <jason@redhat.com>
4421 * Makefile.in (TAGS): Add c-family/*.cc.
4423 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4425 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4426 (call_value_insn): Likewise.
4427 (sibcall_insn): Likewise.
4428 (sibcall_value_insn): Likewise.
4429 (movsi_aarch64): Likewise.
4430 (movdi_aarch64): Likewise.
4431 (add_losym_): Likewise.
4432 (ldr_got_small_): Likewise.
4433 (ldr_got_small_sidi): Likewise.
4434 (ldr_got_small_28k_): Likewise.
4435 (ldr_got_small_28k_sidi): Likewise.
4436 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4437 Move output_addr_const to symbolic case. Add error check.
4439 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4441 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4442 * config/s390/s390.md: Disable MVC merging peephole if it would
4443 disable operand forwarding.
4444 (new peephole2): Split MVCs if it would turn them into up to 2
4447 2017-12-01 Richard Biener <rguenther@suse.de>
4449 PR tree-optimization/83232
4450 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4451 detection of same access. Instead of breaking the group here
4452 do not consider the duplicate. Add comment explaining real fix.
4454 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4456 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4458 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4460 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4462 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4464 * function.h (struct function): Remove cilk_frame_decl,
4465 is_cilk_function and calls_cilk_spawn fields.
4466 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4468 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4469 cilk_elemental field.
4470 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4471 * target.def: Adjust comment.
4472 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4473 Don't test cilk_elemental.
4475 PR tree-optimization/83233
4476 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4477 bswap_stat name for the struct.
4480 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4481 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4482 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4483 SWITCH_BREAK_LABEL_P set on the label.
4484 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4485 added for default case if it was missing and not all cases covered.
4486 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4487 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4489 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4490 to avoid -Wimplicit-fallthrough warning.
4491 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4492 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4494 PR tree-optimization/83221
4495 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4497 (init_reassoc): Formatting fix.
4500 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4501 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4505 * tree-cfg.c: Include asan.h.
4506 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4508 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4510 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4512 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4514 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4515 vpcmpeqd instruction.
4517 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4519 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4520 registers type for 128bit mode.
4522 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4524 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4526 * gengtype-state.c (read_state_pair): Likewise.
4527 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4528 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4529 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4530 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4531 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4532 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4533 * builtins.c (fold_builtin_3): Likewise.
4534 * graphite-scop-detection.c
4535 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4536 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4539 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4542 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4543 parameter. Return true if ignore_decl_rtl_set_p is true and other
4544 conditions are satisfied.
4545 * asan.h (asan_protect_global): Add new parameter.
4546 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4547 to asan_protect_global calls.
4549 2017-11-30 Jim Wilson <jimw@sifive.com>
4551 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4552 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4553 -mstrict-align, add info on default value. Delete redundant lines for
4554 -mabi. Add missing -mexplicit-relocs docs.
4556 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4558 * config/arc/arc.md (trap): New pattern.
4560 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4562 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4563 instruction to end into a delay slot.
4564 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4565 can be placed into a delay slot against reg_note.
4567 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4569 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4570 labels number of usages.
4572 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4574 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4576 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4578 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4581 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4582 multiplication by power of 2 constant into two shifts + comparison.
4584 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4587 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4589 2017-11-30 Richard Biener <rguenther@suse.de>
4591 PR tree-optimization/83202
4592 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4593 allow_peel argument and guard peeling.
4594 (canonicalize_loop_induction_variables): Likewise.
4595 (canonicalize_induction_variables): Pass false.
4596 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4597 peeling from cunrolli.
4599 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4601 * combine.c (try_combine): Print a message to dump file whenever
4602 I0, I1, or I2 cannot be combined into I3.
4604 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4606 PR rtl-optimization/83156
4607 PR rtl-optimization/82621
4608 * combine.c (try_combine): Don't split an I2 if one of the dests is
4609 set again before I3. Allow unused dests.
4611 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4613 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4615 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4617 PR rtl-optimization/80818
4618 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4619 recursively. Use insn code for clobber.
4620 (lra_set_insn_recog_data): Pass the new arg to
4621 collect_non_operand_hard_regs.
4622 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4624 (lra_update_insn_regno_info): Pass insn to
4625 add_regs_to_insn_regno_info.
4627 2017-11-29 Jim Wilson <jimw@sifive.com>
4628 Andrew Waterman <andrew@sifive.com>
4630 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4631 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4632 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4633 use SINGLE_SHIFT_COST.
4634 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4635 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4637 2017-11-29 Julia Koval <julia.koval@intel.com>
4639 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4640 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4641 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4642 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4643 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4644 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4645 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4646 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4647 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4648 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4649 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4650 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4651 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4652 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4653 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4654 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4655 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4656 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4657 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4658 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4659 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4660 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4661 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4662 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4663 __builtin_ia32_vpshldv_v2di_mask,
4664 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4665 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4666 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4668 2017-11-29 Julia Koval <julia.koval@intel.com>
4670 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4671 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4672 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4673 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4674 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4675 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4676 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4677 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4678 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4679 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4680 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4681 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4682 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4683 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4684 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4685 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4686 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4687 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4688 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4689 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4690 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4692 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4694 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4695 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4698 2017-11-29 Martin Aberg <maberg@gaisler.com>
4700 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4701 to prevent b2bst errata sequence.
4702 (sqrtdf2_fix): Likewise.
4704 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4706 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4707 (div_sqrt_insn_p): New function.
4708 (sparc_do_work_around_errata): Insert NOP instructions to
4709 prevent sequences that could trigger the TN-0013 errata for
4710 certain LEON3 processors.
4711 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4712 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4713 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4714 (in_branch_delay): Prevent div and sqrt in delay slot if
4716 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4718 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4720 * config/sparc/sparc.c (atomic_insn_p): New function.
4721 (sparc_do_work_around_errata): Insert NOP instructions to
4722 prevent sequences that could trigger the TN-0010 errata for
4724 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4725 instruction referable in atomic_insns_p.
4727 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4729 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4730 (atomic_compare_and_swap_leon3_1): Likewise.
4733 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4735 * config/sparc/sparc.c (fpop_insn_p): New function.
4736 (sparc_do_work_around_errata): Insert NOP instructions to
4737 prevent sequences that could trigger the TN-0012 errata for
4739 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4740 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4741 (in_branch_annul_delay): Prevent floating-point instructions
4742 in delay slot of annulled integer branch.
4744 2017-11-29 Richard Biener <rguenther@suse.de>
4746 PR tree-optimization/83202
4747 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4749 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4750 nodes vectorized to the same stmts multiple times.
4751 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4752 (vect_analyze_slp_instance): Adjust.
4753 (scalar_stmts_to_slp_tree_map_t): New typedef.
4754 (vect_schedule_slp_instance): Add a map recording the SLP node
4755 representing the vectorized stmts for a set of scalar stmts.
4756 Avoid code-generating redundancies.
4757 (vect_schedule_slp): Allocate map and pass it down.
4759 2017-11-29 Nathan Sidwell <nathan@acm.org>
4762 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4765 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4767 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4768 cases for printing LDP/STP memory addresses.
4769 (aarch64_print_address_internal): Renamed from
4770 aarch64_print_operand_address, added parameter, add Pmode check.
4771 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4772 (aarch64_print_operand_address): Indirect to
4773 aarch64_print_address_internal.
4774 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4775 'y' operand output specifier.
4777 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4780 * tree.c (build_simple_mem_ref_loc): Handle
4781 get_addr_base_and_unit_offset returning a MEM_REF.
4784 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4787 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4790 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4792 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4794 (layout::move_to_column): Likewise.
4796 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4799 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4800 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4802 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4803 Martin Jambor <mjambor@suse.cz>
4806 * tree.h (expr_type_first_operand_type_p): Declare
4807 * tree.c (expr_type_first_operand_type_p): New function.
4808 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4809 (ipa_value_from_jfunc): Adjust declaration.
4810 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4811 Use it as result type for arithmetics, unless it is NULL in which case
4812 be more conservative.
4813 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4814 ipa_get_jf_pass_through_result.
4815 (propagate_vals_across_pass_through): Likewise.
4816 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4817 is to propagate_vals_across_pass_through.
4818 (propagate_constants_across_call): Pass PARM_TYPE to
4819 propagate_scalar_across_jump_function.
4820 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4821 ipa_value_from_jfunc.
4822 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4823 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4824 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4825 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4826 target_type, pass it to ipa_value_from_jfunc.
4827 (update_indirect_edges_after_inlining): Pass parameter type to
4828 try_make_edge_direct_simple_call.
4830 2017-11-28 Jeff Law <law@redhat.com>
4832 * gimple-ssa-evrp-analyze.c
4833 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4834 refine ranges if scev_initialized_p returns true.
4835 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4837 2017-11-28 Julia Koval <julia.koval@intel.com>
4839 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4840 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4841 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4842 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4843 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4844 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4845 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4846 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4847 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4848 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4849 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4850 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4851 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4852 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4853 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4854 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4855 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4856 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4857 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4858 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4859 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4860 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4862 2017-11-28 Julia Koval <julia.koval@intel.com>
4864 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4865 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4866 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4867 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4868 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4869 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4870 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4871 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4872 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4873 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4874 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4875 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4876 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4877 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4878 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4879 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4880 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4881 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4882 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4883 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4884 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4885 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4886 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4887 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4888 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4889 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4890 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4891 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4892 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4893 __builtin_ia32_vpshld_v2di_mask): New builtins.
4894 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4895 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4897 2017-11-28 Richard Biener <rguenther@suse.de>
4899 PR tree-optimization/80776
4900 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4902 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4904 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4905 If the incoming edge is an effective fallthru because the other
4906 edge only reaches a __builtin_unreachable () then record ranges
4907 derived from the controlling condition in SSA info.
4908 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4909 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4911 2017-11-28 Olivier Hainque <hainque@adacore.com>
4913 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4914 on mingw build hosts.
4916 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4919 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4921 2017-11-28 Richard Biener <rguenther@suse.de>
4924 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4925 copies generated from memcpy use a character array as reference
4928 2017-11-28 Julia Koval <julia.koval@intel.com>
4929 Sebastian Peryt <sebastian.peryt@intel.com>
4931 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4932 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4933 cilk-common.o, cilk.h, cilk-common.c): Remove.
4935 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4936 * builtins.c (is_builtin_name): Remove cilkplus condition.
4937 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4938 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4939 cilk-builtins.def, cilkplus.def): Remove.
4940 * cif-code.def (CILK_SPAWN): Remove.
4941 * cilk-builtins.def: Delete.
4942 * cilk-common.c: Ditto.
4944 * cilkplus.def: Ditto.
4945 * config/darwin.h (fcilkplus): Delete.
4946 * cppbuiltin.c: Ditto.
4947 * doc/extend.texi: Remove cilkplus doc.
4948 * doc/generic.texi: Ditto.
4949 * doc/invoke.texi: Ditto.
4950 * doc/passes.texi: Ditto.
4951 * gcc.c (fcilkplus): Remove.
4952 * gengtype.c (cilk.h): Remove.
4953 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4955 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4957 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4959 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4960 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4961 cilkplus conditions.
4962 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4963 inline_read_section): Ditto.
4964 * ipa-inline-analysis.c (cilk.h): Remove.
4965 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4966 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4967 append_linker_options): Remove condition for fcilkplus.
4968 * lto/lto-lang.c (cilk.h): Remove.
4969 (lto_init): Remove condition for fcilkplus.
4970 * omp-expand.c (expand_cilk_for_call): Delete.
4971 (expand_omp_taskreg, expand_omp_for_static_chunk,
4972 expand_omp_for): Remove cilkplus
4974 (expand_cilk_for): Delete.
4975 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4976 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4977 execute_lower_omp, diagnose_sb_0): Ditto.
4978 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4979 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4980 * tree-nested.c: Ditto.
4981 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4982 (dump_generic_node): Ditto.
4983 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4984 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4986 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4988 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4990 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4991 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4992 for vector mode and !TARGET_SIMD.
4994 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4996 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4998 * tree.h (SWITCH_LABELS): Remove.
4999 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5000 assert SWITCH_BODY is non-NULL.
5001 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5003 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5005 PR tree-optimization/80788
5006 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5007 has TREE_OVERFLOW set, call drop_tree_overflow.
5009 2017-11-28 Richard Biener <rguenther@suse.de>
5011 PR tree-optimization/83158
5012 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5014 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5017 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5018 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5020 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5022 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5023 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5025 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5028 * varasm.c (compare_constant): Take the mode of the constants into
5029 account when comparing floating point constants.
5031 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5033 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5034 from explicit instantiation of debug_helper.
5035 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5037 2017-11-27 Richard Biener <rguenther@suse.de>
5039 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5042 2017-11-27 Richard Biener <rguenther@suse.de>
5044 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5045 (type_hash_canon): Also clear int_cst_hash_table entry for
5047 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5049 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5051 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5052 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5053 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5055 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5057 * hash-map.h (gt_cleare_cache): Avoid UB.
5059 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5061 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5062 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5063 (decide_unroll_runtime_iterations): Likewise.
5064 (decide_unroll_stupid): Likewise.
5066 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5069 * config/i386/i386.md: Add a loop with incssp.
5071 2017-11-27 Martin Jambor <mjambor@suse.cz>
5073 PR tree-optimization/81248
5074 * tree-sra.c (splice_param_accesses): Remove size check.
5075 (decide_one_param_reduction): Fix size check.
5076 * gimple-pretty-print.c (dump_profile): Silence warning.
5077 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5079 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5082 * dbxout.c (lastlineno): New variable.
5083 (dbx_debug_hooks): Use dbxout_switch_text_section as
5084 switch_text_section debug hook.
5085 (dbxout_function_end): Switch to current_function_section
5086 rather than function_section. If crtl->has_bb_partition,
5087 output just one N_FUN, depending on in_cold_section_p.
5088 (dbxout_source_line): Remember last lineno in lastlineno.
5089 (dbxout_switch_text_section): New function.
5090 (dbxout_function_decl): Adjust dbxout_block caller.
5091 (dbx_block_with_cold_children): New function.
5092 (dbxout_block): Return true if any LBRAC/RBRAC have been
5093 emitted. Use dbx_block_with_cold_children at depth == 0
5094 in second partition. Add PARENT_BLOCKNUM argument, pass
5095 it optionally adjusted to children. Output LBRAC/RBRAC
5096 around recursive call only if the block is in the current
5097 partition, if not and anything was output, emit empty
5099 * final.c (final_scan_insn): Compute cold_function_name
5100 before calling switch_text_section debug hook. Call
5101 that hook even if dwarf2out_do_frame if not emitting
5105 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5106 TREE_READONLY decls.
5108 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5110 PR rtl-optimization/82488
5111 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5113 2017-11-26 Julia Koval <julia.koval@intel.com>
5115 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5117 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5120 2017-11-26 Julia Koval <julia.koval@intel.com>
5122 * config/i386/driver-i386.c (host_detect_local_cpu):
5123 Detect skylake-avx512.
5125 2017-11-26 Julia Koval <julia.koval@intel.com>
5127 * config.gcc: Add -march=cannonlake.
5128 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5129 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5130 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5131 (PTA_CANNONLAKE): New.
5132 (processor_target_table): Add cannonlake.
5133 (ix86_option_override_internal): Ditto.
5134 (fold_builtin_cpu): Ditto.
5135 (get_builtin_code_for_version): Handle cannonlake.
5136 (M_INTEL_COREI7_CANNONLAKE): New.
5137 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5138 * doc/invoke.texi: Add -march=cannonlake.
5140 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5142 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5143 (try_init_one_plugin): Alternative implementation for MinGW.
5144 * Makefile.in (plugin_implib): New.
5145 (gengtype-lex.c): Fix broken AIX workaround.
5146 * configure: Regenerate.
5147 * doc/plugins.texi: Document support for MinGW.
5149 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5151 PR rtl-optimization/81553
5152 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5153 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5154 is a shift where C1 has different mode than the whole shift, use C1's
5155 mode for MULT rather than the shift's mode.
5158 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5159 builtins not enabled in the currently selected ISA.
5161 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5163 PR tree-optimization/71026
5164 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5165 (insert_reciprocals): Change to insert reciprocals before a division
5166 by a square and to insert the square of a reciprocal.
5167 (execute_cse_reciprocals_1): Change to consider division by a square.
5168 (register_division_in): Add importance parameter.
5170 2017-11-24 Richard Biener <rguenther@suse.de>
5172 PR tree-optimization/82402
5173 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5174 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5176 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5178 * match.pd (0-ptr): New transformation.
5180 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5183 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5184 when skipping checking.
5186 2017-11-24 Richard Biener <rguenther@suse.de>
5188 PR tree-optimization/83128
5189 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5190 (vn_reference_lookup_3): Likewise.
5192 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5195 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5196 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5199 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5200 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5201 check for the latter with gimple_call_builtin_p. Do not handle
5202 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5204 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5206 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5209 2017-11-23 Julia Koval <julia.koval@intel.com>
5211 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5212 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5213 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5214 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5215 _mm512_maskz_expandloadu_epi16): New intrinsics.
5216 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5217 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5218 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5219 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5220 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5221 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5222 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5223 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5224 _mm256_maskz_expandloadu_epi8): New intrinsics.
5225 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5226 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5227 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5228 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5229 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5230 * config/i386/sse.md (VI248_VLBW): New iterator.
5231 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5233 2017-11-23 Julia Koval <julia.koval@intel.com>
5235 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5236 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5237 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5238 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5239 _mm512_mask_compressstoreu_epi16): New.
5240 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5241 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5242 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5243 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5244 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5245 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5246 _mm256_mask_compressstoreu_epi8): New.
5247 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5248 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5249 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5250 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5251 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5252 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5253 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5254 __builtin_ia32_compresshi128_mask,
5255 __builtin_ia32_compressstoreuqi512_mask,
5256 __builtin_ia32_compressstoreuhi512_mask,
5257 __builtin_ia32_compressstoreuqi256_mask,
5258 __builtin_ia32_compressstoreuqi128_mask,
5259 __builtin_ia32_compressstoreuhi256_mask,
5260 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5261 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5263 (ix86_expand_special_args_builtin): Handle new types.
5264 (s4fma_expand): Handle new builtin array.
5265 * config/i386/immintrin.h: Include new headers.
5266 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5267 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5269 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5271 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5272 fixes. Declare temp and g variables at the top in order to avoid
5273 {} in most of the cases.
5275 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5277 * match.pd (ptr-0): New transformation.
5279 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5281 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5282 (struct addr_mode_cost_table): New.
5283 (struct tune_params): Add field addr_mode_costs.
5284 * config/arm/arm.c (generic_addr_mode_costs): New.
5285 (arm_slowmul_tune): Initialise addr_mode_costs field.
5286 (arm_fastmul_tune): Likewise.
5287 (arm_strongarm_tune): Likewise.
5288 (arm_xscale_tune): Likewise.
5289 (arm_9e_tune): Likewise.
5290 (arm_marvell_pj4_tune): Likewise.
5291 (arm_v6t2_tune): Likewise.
5292 (arm_cortex_tune): Likewise.
5293 (arm_cortex_a8_tune): Likewise.
5294 (arm_cortex_a7_tune): Likewise.
5295 (arm_cortex_a15_tune): Likewise.
5296 (arm_cortex_a35_tune): Likewise.
5297 (arm_cortex_a53_tune): Likewise.
5298 (arm_cortex_a57_tune): Likewise.
5299 (arm_exynosm1_tune): Likewise.
5300 (arm_xgene1_tune): Likewise.
5301 (arm_cortex_a5_tune): Likewise.
5302 (arm_cortex_a9_tune): Likewise.
5303 (arm_cortex_a12_tune): Likewise.
5304 (arm_cortex_a73_tune): Likewise.
5305 (arm_v7m_tune): Likewise.
5306 (arm_cortex_m7_tune): Likewise.
5307 (arm_v6m_tune): Likewise.
5308 (arm_fa726te_tune): Likewise.
5309 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5311 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5313 * config/arm/arm.c (arm_mem_costs): New function.
5314 (arm_rtx_costs_internal): Use arm_mem_costs.
5316 2017-11-23 Mark Wielaard <mark@klomp.org>
5318 * dwarf2out.c (init_sections_and_labels): Use generation to create
5319 unique ranges_section_label and ranges_base_label. Return generation.
5320 (output_rnglists): Add generation argument. Use generation to create
5321 unique ranges labels.
5322 (dwarf2out_finish): Get generation from init_sections_and_labels
5323 and pass generation to output_rnglists.
5325 2017-11-23 Mike Stump <mikestump@comcast.net>
5326 Eric Botcazou <ebotcazou@adacore.com>
5328 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5329 * cfgloop.h (struct loop): Add unroll field.
5330 * function.h (struct function): Add has_unroll bitfield.
5331 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5332 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5333 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5334 (pass_rtl_unroll_loops::gate): Likewise.
5335 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5336 for which loop->unroll==1.
5337 (decide_unroll_constant_iterations): Use note for consistency and
5338 take loop->unroll into account. Return early if loop->unroll is set.
5339 Fix thinko in existing test.
5340 (decide_unroll_runtime_iterations): Use note for consistency and
5341 take loop->unroll into account.
5342 (decide_unroll_stupid): Likewise.
5343 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5344 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5345 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5347 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5348 (print_loop): Print loop->unroll if set.
5349 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5350 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5351 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5353 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5354 loop->unroll is set and smaller than the trip count. Otherwise bypass
5355 entirely the heuristics if loop->unroll is set. Remove dead note.
5356 Fix off-by-one bug in other note.
5357 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5358 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5360 (tree_unroll_loops_completely): Make static.
5361 (pass_complete_unroll::execute): Use correct type for variable.
5362 (pass_complete_unrolli::execute): Fix formatting.
5363 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5365 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5367 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5368 enable when TARGET_PREFER_AVX128 is set.
5370 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5372 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5374 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5375 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5377 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5379 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5380 frequencies. Fix estimation of aggregate parameters.
5382 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5384 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5387 2017-11-23 Tom de Vries <tom@codesourcery.com>
5389 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5391 2017-11-23 Tom de Vries <tom@codesourcery.com>
5393 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5396 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5399 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5400 sibcall_value_pcrel_fdpic): Use local variable instead of
5402 (calli_tbr_rel): Add missing operand 2.
5403 (call_valuei_tbr_rel): Add missing operand 3.
5405 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5408 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5409 bitpos/bitsize covering the whole destination, use store_expr only if
5410 the complex mode is the same. Otherwise, use expand_normal and if
5411 it returns CONCAT, subreg each part separately instead of trying to
5412 subreg the whole result.
5414 2017-11-23 Richard Biener <rguenther@suse.de>
5416 PR tree-optimization/23094
5417 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5418 come here from walking over backedges in the first iteration.
5419 (vn_reference_lookup_3): Skip clobbers that store the same value.
5421 2017-11-23 Richard Biener <rguenther@suse.de>
5423 PR tree-optimization/81403
5424 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5425 a block we need a leader relative to.
5426 (phi_translate_1): For nary processing require a leader from
5427 get_representative_for given we run expression simplification
5428 using match-and-simplify. Remove previous fix.
5430 2017-11-22 Jeff Law <law@redhat.com>
5432 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5433 Use new method allocate_value_range rather than accessing the
5434 vrp_value_range_pool data member directly.
5435 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5436 to use extract_range_from_stmt method to avoid need for
5437 extract_range_from_assignment method.
5438 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5439 method rather than setting values_propgated data member directly.
5440 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5441 and values propagated data members and extract_range_from_assignment
5442 method. Reorder private data members to conform to standards.
5443 Add new methods set_lattice_propagation_complete and
5444 allocate_value_range.
5446 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5448 PR rtl-optimization/83030
5449 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5450 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5451 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5452 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5453 CROSSING_JUMP_P flag is already set before setting it.
5454 * cfgrtl.c (fixup_partition_crossing): Likewise.
5455 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5458 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5460 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5461 Handle the case where both arguments are using gen_const_vec_series.
5463 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5466 * pretty-print.c (pp_format): Move quoting implementation to
5467 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5468 to pass address of "quote" local.
5469 (pp_begin_quote): New function.
5470 (pp_end_quote): New function.
5471 * pretty-print.h (printer_fn): Convert penultimate param from bool
5473 (pp_begin_quote): New decl.
5474 (pp_end_quote): New decl.
5475 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5476 param from bool to bool *.
5477 * tree-diagnostic.h (default_tree_printer): Likewise.
5479 2017-11-22 Jeff Law <law@redhat.com>
5481 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5482 Initialize vr_values.
5483 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5484 vr_values via vr_values class instance rather than delegators.
5485 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5486 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5487 (evrp_range_analyzer::push_value_range): Likewise.
5488 (evrp_range_analyzer::pop_value_range): Likewise.
5489 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5490 most delegators. Those remaining are exposed as public interfaces.
5491 Make vr_values a pointer and private.
5492 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5494 (evrp_range_analyzer::get_vr_value): New method.
5495 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5496 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5497 (class evrp_dom_walker): Attach evrp_folder class, initialize
5498 it in the ctor. Remove temporary delegators.
5499 (evrp_dom_walker::before_dom_children): Call methods in attached
5500 evrp_range_analyzer class via class instance pointer. Use
5501 free value_range_constant_singleton to remove need for
5502 op_with_constant_singleton_value delegator method. Do not
5503 create a vrp_prop class instance for every call! Narrow
5504 scope of a couple variables.
5505 (evrp_dom_walker::cleanup): Call methods in attached
5506 evrp_range_analyzer class via class instance pointer.
5507 * vr-values.h (class vr_values): Privatize many methods and
5510 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5512 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5513 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5514 * expr.c (expand_expr_real_2): Likewise.
5515 * fold-const.c (const_unop): Likewise.
5516 * optabs-tree.c (optab_for_tree_code): Likewise.
5517 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5518 * tree-inline.c (estimate_operator_cost): Likewise.
5519 * tree-pretty-print.c (dump_generic_node): Likewise.
5520 (op_code_prio): Likewise.
5521 (op_symbol_code): Likewise.
5522 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5523 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5524 * internal-fn.c (direct_internal_fn_optab): New function.
5525 (direct_internal_fn_array, direct_internal_fn_supported_p
5526 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5527 * fold-const-call.c (fold_const_reduction): New function.
5528 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5530 * tree-vect-loop.c: Include internal-fn.h.
5531 (reduction_code_for_scalar_code): Rename to...
5532 (reduction_fn_for_scalar_code): ...this and return an internal
5534 (vect_model_reduction_cost): Take an internal_fn rather than
5536 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5538 (vectorizable_reduction): Use internal functions rather than tree
5539 codes for the reduction operation. Update calls to the functions
5541 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5542 Use calls to internal functions rather than REDUC tree codes.
5543 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5545 2017-11-22 Olivier Hainque <hainque@adacore.com>
5547 * config/vxworks.c (vxworks_override_options): Pick default
5548 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5549 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5550 DWARF_GNAT_ENCODINGS_DEFAULT.
5551 * config/vxworksae.h: Likewise.
5553 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5555 PR tree-optimization/83104
5556 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5559 2017-11-22 Marek Polacek <polacek@redhat.com>
5560 H.J. Lu <hongjiu.lu@intel.com>
5561 Jason Merrill <jason@redhat.com>
5566 * calls.c (initialize_argument_information): Call
5567 warn_parameter_passing_abi target hook.
5568 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5569 argument onto stack.
5570 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5571 * common.opt: Update -fabi-version description.
5572 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5573 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5575 (ix86_is_empty_record): New function.
5576 (ix86_warn_parameter_passing_abi): New function.
5577 (TARGET_EMPTY_RECORD_P): Redefine.
5578 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5579 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5580 * doc/tm.texi: Regenerated.
5581 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5582 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5583 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5584 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5585 instead of int_size_in_bytes.
5586 * expr.c (copy_blkmode_to_reg): Likewise.
5587 * function.c (aggregate_value_p): Return 0 for empty types.
5588 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5590 (locate_and_pad_parm): Call arg size_in_bytes instead
5592 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5593 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5594 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5596 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5597 (std_gimplify_va_arg_expr): Skip empty records. Call
5598 arg_size_in_bytes instead size_in_bytes.
5599 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5600 * tree-core.h (tree_type_common): Add empty_flag.
5601 (tree_decl_common): Update comments.
5602 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5604 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5605 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5607 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5608 * tree.c (default_is_empty_type): New function.
5609 (default_is_empty_record): New function.
5610 (arg_int_size_in_bytes): New function.
5611 (arg_size_in_bytes): New function.
5612 (get_ultimate_context): New function.
5613 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5614 TRANSLATION_UNIT_WARN_EMPTY_P.
5615 (default_is_empty_record, arg_int_size_in_bytes,
5616 arg_size_in_bytes, get_ultimate_context): Declare.
5618 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5620 * config/arm/arm.c (cmse_clear_registers): New function.
5621 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5622 code by call to cmse_clear_registers.
5623 (cmse_nonsecure_entry_clear_before_return): Likewise.
5625 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5627 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5628 (vdot_s32, vdotq_s32): New.
5629 (vdot_lane_u32, vdotq_lane_u32): New.
5630 (vdot_lane_s32, vdotq_lane_s32): New.
5633 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5636 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5637 for unsigned values with fewer HWIs than the precision.
5638 (test_overflow): New function.
5639 (wide_int_cc_tests): Call it.
5641 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5643 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5644 has the same mode class as Pmode.
5646 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5648 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5649 padding_bits_to_clear_ptr.
5650 (cmse_nonsecure_entry_clear_before_return): Likewise.
5652 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5654 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5655 auto_sbitap instead of integer bitfield to control register needing
5658 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5660 PR tree-optimization/83044
5661 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5662 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5663 that need the upper bound. Subtract offset from
5664 get_addr_base_and_unit_offset only if positive and subtract it
5665 before division by eltsize rather than after it.
5668 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5669 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5670 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5674 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5675 expand_binop with *mul_widen_optab, make sure at least one of the
5676 operands doesn't have VOIDmode.
5679 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5681 PR rtl-optimization/82044
5682 PR tree-optimization/82042
5683 * dse.c (record_store): Check for overflow.
5684 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5685 add_wild_read instead of clear_rhs_from_active_local_stores on
5686 overflow and log it into dump_file.
5688 2017-11-22 Richard Biener <rguenther@suse.de>
5690 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5691 fake edges to exit when looking for a place to insert.
5692 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5694 (insert_into_preds_of_block): Commit edge insertion immediately,
5695 assert that doesn't require new BBs.
5696 (fini_pre): Release expressions.
5697 (pass_pre::execute): Shuffle things around a bit, if the fn
5698 is too large do not compute AVAIL either as this is really the
5701 2017-11-22 Richard Biener <rguenther@suse.de>
5703 PR tree-optimization/83089
5704 * tree-if-conv.c (pass_if_conversion::execute): If anything
5705 changed reset SCEV and free the number of iteration estimates.
5707 2017-11-21 Martin Sebor <msebor@redhat.com>
5709 PR tree-optimization/82945
5710 * calls.h (warn_nonstring_bound): Remove unused function.
5712 2017-11-21 Martin Sebor <msebor@redhat.com>
5714 PR tree-optimization/82945
5715 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5716 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5717 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5719 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5720 * calls.h (get_attr_nonstring_decl): Declare new function.
5721 * doc/extend.texi (attribute nonstring): Update.
5722 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5723 get_attr_nonstring_decl and handle it.
5724 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5725 detection of nul-termination.
5726 (strlen_to_stridx): Change to a pointer.
5727 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5728 (pass_strlen::execute): Same.
5730 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5732 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5733 for the new option -mprefer-vector-width=[none|128|256|512].
5734 * config/i386/i386.c (ix86_target_string): remove old style options
5735 -mprefer-avx256 and make -mprefer-avx128 as alias.
5736 (ix86_option_override_internal): Apply defaults for the
5737 -mprefer-vector-width=[128|256] option.
5738 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5739 Implement macros to work with -mprefer-vector-width=.
5740 * config/i386/i386.opt: Implemented option
5741 -mprefer-vector-width=[none|128|256|512].
5742 * doc/invoke.texi: Documentation for
5743 -mprefer-vector-width=[none|128|256|512].
5745 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5747 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5749 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5751 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5753 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5755 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5757 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5758 * expr.c (expand_expr_real_2): Likewise.
5759 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5760 fold_binary_loc): Likewise.
5761 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5762 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5763 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5764 MINUS_EXPR transformations.
5765 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5766 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5767 * tree-inline.c (estimate_operator_cost): Likewise.
5768 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5769 op_symbol_code): Likewise.
5770 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5771 * vr-values.c (extract_range_from_binary_expr): Likewise.
5772 * varasm.c (initializer_constant_valid_p_1): Likewise.
5773 * tree.def: New tree code POINTER_DIFF_EXPR.
5775 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5777 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5778 integer suffix to movbe mnemonic.
5779 (*bswaphi2_movbe): Ditto.
5780 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5782 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5785 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5786 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5787 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5789 2017-11-21 Martin Liska <mliska@suse.cz>
5791 * tree-inline.c (expand_call_inline): Remove not needed
5794 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5795 Jakub Jelinek <jakub@redhat.com>
5798 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5799 Remove static keyword from f variable.
5801 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5803 PR tree-optimization/83086
5804 * gimple-ssa-store-merging.c
5805 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5806 rather than n.base_addr.
5808 2017-11-21 Martin Liska <mliska@suse.cz>
5810 PR rtl-optimization/82044
5811 PR tree-optimization/82042
5812 * dse.c (check_mem_read_rtx): Check for overflow.
5814 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5816 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5819 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5821 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5822 and its corresponding call arg location note.
5824 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5825 Andrew Burgess <andrew.burgess@embecosm.com>
5827 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5829 (arc_return_slot_offset): Likewise.
5830 (arc_eh_return_address_location): New declaration.
5831 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5832 (MUST_SAVE_REGISTER): Add exception handler case.
5833 (MUST_SAVE_RETURN_ADDR): Likewise.
5834 (arc_frame_pointer_required): Likewise.
5835 (arc_frame_pointer_needed): New function.
5836 (arc_compute_frame_size): Changed.
5837 (arc_expand_prologue): Likewise.
5838 (arc_expand_epilogue): Likewise.
5839 (arc_initial_elimination_offset): Likewise.
5840 (arc_return_slot_offset): Delete.
5841 (arc_eh_return_address_location): New function.
5842 (arc_builtin_setjmp_frame_value): Likewise.
5843 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5844 (EH_RETURN_STACKADJ_RTX): Define.
5845 (EH_RETURN_HANDLER_RTX): Likewise.
5846 * config/arc/arc.md (eh_return): Delete.
5848 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5850 * print-tree.h (debug_vec_tree): Remove prototype.
5851 * gdbinit.in (pvt): Remove macro.
5853 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5855 PR tree-optimization/83047
5856 * gimple-ssa-store-merging.c
5857 (imm_store_chain_info::output_merged_store): If the loads with the
5858 same vuse are in different basic blocks, for load_gsi pick a load
5859 location that is dominated by the other loads.
5862 * config/i386/i386.c (ix86_memmodel_check): Start
5863 -Winvalid-memory-model diagnostics with lowercase letter.
5866 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5867 set in_cold_section_p to the partition containing loc_list->first.
5868 When seeing loc_list->last_before_switch node, update secname and
5869 perform range_across_switch second partition handling only after that.
5872 * run-rtl-passes.c: Include debug.h.
5873 (run_rtl_passes): Call debug_hooks->assembly_start.
5874 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5878 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5879 OPTAB_DIRECT in calls to expand_simple_binop.
5881 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5884 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5886 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5889 * spellcheck.h (best_match::blithely_get_best_candidate): New
5892 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5894 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5895 with lower case letter.
5897 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5899 * config/i386/i386.md (bswaphi2): New expander.
5900 (*bswaphi2_movbe): New insn pattern.
5901 (bswaphi -> rorhi pepehole2): New peephole pattern.
5903 2017-11-20 Jeff Law <law@redhat.com>
5905 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5906 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5907 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5908 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5909 gimple-ssa-evrp-analyze.h.
5911 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5912 set BB_VISITED here.
5913 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5915 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5917 * tree-predcom.c: Add general comment on Store-Store chains.
5918 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5920 (get_chain_last_ref_at): Rename into...
5921 (get_chain_last_write_at): ...this.
5922 (get_chain_last_write_before_load): New function.
5923 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5924 CT_STORE_STORE when write reference is added.
5925 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5926 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5927 (initialize_root_vars_store_elim_1): Ditto.
5928 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5929 definition is created.
5930 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5931 chain by replacing it with dominant stored value.
5933 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5935 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5937 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5939 * vr-values.c (extract_range_from_binary_expr): Use a full range
5942 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5944 * config/arm/arm.md (R4_REGNUM): Define constant.
5945 (nonsecure_call_internal): Remove r4 clobber.
5946 (nonsecure_call_value_internal): Likewise.
5947 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5948 clobber and resequence match_operands.
5949 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5950 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5951 (nonsecure_call_value_reg_thumb2): Likewise.
5953 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5955 PR tree-optimization/78821
5956 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5957 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5958 to the address of the base rather than the base itself.
5959 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5960 (find_bswap_or_nop_finalize): New function.
5961 (find_bswap_or_nop): Use it.
5962 (bswap_replace): Return a tree rather than bool, change first
5963 argument from gimple * to gimple_stmt_iterator, allow inserting
5964 into an empty sequence, allow ins_stmt to be NULL - then emit
5965 all stmts into gsi. Fix up MEM_REF address gimplification.
5966 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5967 (struct store_immediate_info): Add N and INS_STMT non-static
5969 (store_immediate_info::store_immediate_info): Initialize them
5970 from newly added ctor args.
5971 (merged_store_group::apply_stores): Formatting fixes. Sort by
5973 (stmts_may_clobber_ref_p): For stores call also
5974 refs_anti_dependent_p.
5975 (gather_bswap_load_refs): New function.
5976 (imm_store_chain_info::try_coalesce_bswap): New method.
5977 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5978 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5979 (imm_store_chain_info::output_merged_store): Fail if number of
5980 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5981 and NOP_EXPR rhs_code.
5982 (pass_store_merging::process_store): Compute n and ins_stmt, if
5983 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5984 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5986 (pass_store_merging::execute): Calculate dominators.
5988 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5989 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5990 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5991 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5992 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5993 class pass_optimize_bswap, bswap_replace,
5994 pass_optimize_bswap::execute): Moved to ...
5995 * gimple-ssa-store-merging.c: ... this file.
5996 Include optabs-tree.h.
5997 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5998 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5999 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6000 anonymous namespace, remove static keywords.
6001 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6002 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6004 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6007 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6009 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6011 * vec.h (debug_helper): New function.
6012 (DEFINE_DEBUG_VEC): New macro.
6013 * hash-set.h (debug_helper): New function.
6014 (DEFINE_DEBUG_HASH_SET): New macro.
6015 * cfg.c (debug_slim (edge)): New function.
6016 Call DEFINE_DEBUG_VEC for edges.
6017 Call DEFINE_DEBUG_HASH_SET for edges.
6018 * cfghooks.c (debug_slim (basic_block)): New function.
6019 Call DEFINE_DEBUG_VEC for basic blocks.
6020 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6021 * print-tree.c (debug_slim): New function to handle trees.
6022 Call DEFINE_DEBUG_VEC for trees.
6023 Call DEFINE_DEBUG_HASH_SET for trees.
6024 (debug (vec<tree, va_gc>) &): Remove.
6025 (debug (<vec<tree, va_gc>) *): Remove.
6026 * print-rtl.c (debug_slim): New function to handle const_rtx.
6027 Call DEFINE_DEBUG_VEC for rtx_def.
6028 Call DEFINE_DEBUG_VEC for rtx_insn.
6029 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6030 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6031 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6032 (debug (vec<rtx_insn *> *ptr): Remove.
6033 (debug_insn_vector): Remove.
6034 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6036 2017-11-20 Tom de Vries <tom@codesourcery.com>
6038 PR rtl-optimization/82020
6039 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6040 IF_THEN_ELSE condition.
6042 2017-11-19 Jeff Law <law@redhat.com>
6044 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6045 of degenerates resulting from ignoring an edge.
6047 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6050 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6052 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6055 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6056 for uninitialied counts.
6058 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6061 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6063 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6066 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6068 2017-11-19 Tom de Vries <tom@codesourcery.com>
6070 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6073 2017-11-19 Tom de Vries <tom@codesourcery.com>
6075 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6078 2017-11-19 Tom de Vries <tom@codesourcery.com>
6080 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6083 2017-11-19 Tom de Vries <tom@codesourcery.com>
6085 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6087 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6089 2017-11-19 Tom de Vries <tom@codesourcery.com>
6091 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6093 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6094 semicolon after MCORE_EXPORT_NAME call.
6096 2017-11-19 Tom de Vries <tom@codesourcery.com>
6099 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6100 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6102 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6104 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6105 (cgraph_edge::clone): Cleanup updating of profile.
6106 * ipa-cp.c (update_profiling_info): Likewise.
6107 * ipa-inline-transform.c (inline_transform): Likewise.
6108 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6109 * ipa-split.c (execute_split_functions): Do not split when function
6111 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6112 * profile-count.c (profile_count::dump): Add global0.
6113 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6115 (profile_count::to_sreal_scale): Likewise.
6116 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6117 (profile_count::combine_with_ipa_count): New function.
6118 * profile-count.h (profile_guessed_global0adjusted): New.
6119 (profile_count::adjusted_zero): New.
6120 (profile_count::global0adjusted): New.
6121 (profile_count::combine_with_ipa_count): New.
6122 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6123 correct profile of return block of split functions.
6124 (copy_cfg_body): Remove unused profile_count.
6125 (copy_body): Likewise.
6126 (expand_call_inline): Update.
6127 (tree_function_versioning): Update.
6129 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6131 * hash-set.h (hash_set::empty): New.
6132 * tree-ssa-threadbackward.h: Delete.
6133 * tree-ssa-threadbackward.c (class thread_jumps): New.
6134 Move max_threaded_paths into class.
6135 (fsm_find_thread_path): Remove arguments that are now in class.
6136 (profitable_jump_thread_path): Rename to...
6137 (thread_jumps::profitable_jump_thread_path): ...this.
6138 (convert_and_register_jump_thread_path): Rename to...
6139 (thread_jumps::convert_and_register_current_path): ...this.
6140 (check_subpath_and_update_thread_path): Rename to...
6141 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6142 (register_jump_thread_path_if_profitable): Rename to...
6143 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6144 (handle_phi): Rename to...
6145 (thread_jumps::handle_phi): ...this.
6146 (handle_assignment): Rename to...
6147 (thread_jumps::handle_assignment): ...this.
6148 (fsm_find_control_statement_thread_paths): Rename to...
6149 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6150 (find_jump_threads_backwards): Rename to...
6151 (thread_jumps::find_jump_threads_backwards): ...this.
6152 Initialize path local data.
6153 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6154 from within thread_jumps class.
6155 (pass_early_thread_jumps::execute): Same.
6157 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6159 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6161 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6163 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6164 to canonical location.
6166 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6169 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6171 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6173 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6175 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6176 rcrt1.o%s/grcrt1.o%s for -static-pie.
6178 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6180 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6181 ix86_shift_rotate_cost): Break out from ...
6182 (ix86_rtx_costs): ... here.
6183 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6186 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6188 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6189 when entry block was promoted unlikely.
6190 (estimate_bb_frequencies): Increase frequency scale.
6191 * profile-count.h (profile_count): Export precision info.
6193 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6195 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6196 disturbin profile of entry block.
6198 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6200 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6203 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6205 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6206 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6208 2017-11-17 Jeff Law <law@redhat.com>
6210 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6211 from evrp_dom_walker class. Various methods moved into new class.
6212 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6213 (evrp_range_analyzer::enter): New method.
6214 (evrp_range_analyzer::leave): New method.
6215 (evrp_dom_walker): Remove delegators no longer needed by this class.
6216 Replace vr_values data member with evrp_range_analyzer
6218 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6219 method extracted from evrp_dom_walker::before_dom_children.
6220 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6221 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6223 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6224 Add private copy constructor and move assignment operators.
6225 Privatize methods and class data where trivially possible.
6226 (evrp_dom_walker::cleanup): New function, extracted from
6227 execute_early_vrp. Simplify access to class data.
6229 * vr-values.h (get_output_for_vrp): Prototype.
6230 * vr-values.c (get_output_for_vrp): New function extracted from
6231 vrp_visit_assignment_or_call and extract_range_from_stmt.
6232 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6234 2017-11-17 Luis Machado <luis.machado@linaro.org>
6236 * config/aarch64/aarch64.c
6237 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6238 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6239 tune_params::AUTOPREFETCHER_WEAK.
6241 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6244 * config/arm/arm.c (arm_valid_target_attribute_rec):
6245 Parse "arch=" and "+<ext>".
6246 (arm_valid_target_attribute_tree): Re-init global options.
6247 (arm_option_override): Make non-static.
6248 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6249 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6250 (__ARM_FEATURE_CRC32): Support undef.
6251 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6252 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6254 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6256 * gdbinit.in (break-on-diagnostic): New command.
6258 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6260 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6261 return to indirect jump for EH return if control-flow
6262 protection is enabled. Change explicit 'false' argument in
6263 pro_epilogue_adjust_stack with a value of
6265 * config/i386/i386.md (simple_return_indirect_internal):
6266 Remove SImode restriction to support 64-bit.
6268 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6270 * combine.c (added_notes_insn): New.
6271 (try_combine): Handle added_notes_insn like added_links_insn.
6272 Rewrite return value code.
6273 (distribute_notes): Set added_notes_insn to the earliest insn we added
6276 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6278 PR rtl-optimization/82621
6279 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6280 dest of one of those SETs is unused.
6282 2017-11-17 Richard Biener <rguenther@suse.de>
6285 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6286 * tree-pretty-print.c (dump_generic_node): Handle
6287 annot_expr_parallel_kind.
6288 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6289 * gimplify.c (gimple_boolify): Likewise.
6291 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6293 * config.gcc (extra_headers): Add cet.h for x86 targets.
6294 * config/i386/cet.h: New file.
6295 * doc/install.texi: Add --enable-cet/--disable-cet.
6297 2017-11-17 Richard Biener <rguenther@suse.de>
6299 PR tree-optimization/83017
6300 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6301 (gen_parallel_loop): Properly count iterations.
6302 (parallelize_loops): Handle loop->can_be_parallel independent
6303 of flag_loop_parallelize_all. Make static profitability test match
6305 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6306 * invoke.texi (parloops-min-per-thread): Document.
6308 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6310 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6311 upstreaming review comments.
6313 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6315 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6316 with fast unaligned access.
6317 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6319 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6321 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6322 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6324 2017-11-17 Richard Biener <rguenther@suse.de>
6326 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6327 folding of references.
6329 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6332 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6333 of replacing call to strncmp with corresponding call to strcmp when
6336 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6338 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6339 option prefer-avx256 for skylake-avx512 configuration.
6340 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6341 (get_builtin_code_for_version): Ditto.
6343 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6344 Monk Chiang <sh.chiang04@gmail.com>
6346 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6347 (FIXED_REGISTERS): Reserve more register numbers.
6348 (CALL_USED_REGISTERS): Likewise.
6349 (REG_ALLOC_ORDER): Likewise.
6350 (REG_CLASS_CONTENTS): Likewise.
6351 (REGISTER_NAMES): Likewise.
6353 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6354 Kito Cheng <kito.cheng@gmail.com>
6356 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6358 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6360 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6362 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6363 parameter list for vec_splats.
6365 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6367 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6369 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6371 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6374 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6375 with the required version of automake 1.11.6.
6377 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6379 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6381 (DU_C2_3_power9): Correct reservation combinations.
6382 (FP_DIV_power9, VEC_DIV_power9): New.
6383 (power9-alu): Split out rotate/shift...
6384 (power9-rot): ...to here, correct dispatch resource.
6385 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6387 (power9-fp): Correct latency.
6388 (power9-sdiv): Add div/sqrt resource.
6389 (power9-ddiv): Correct latency, add div/sqrt resource.
6390 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6391 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6393 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6395 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6397 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6398 switch statement mapping KF built-ins to TF built-ins if we don't
6399 have the proper ISA 3.0 assembler support.
6401 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6403 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6404 (lower_emutls_function_body): Do not compute it.
6406 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6408 * ipa-split.c (split_bb_info): Turn time to sreal.
6409 (split_point): Likewise.
6410 (dump_split_point): Likewise.
6411 (fine_split_points): Likewise.
6412 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6414 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6416 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6419 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6421 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6423 (expensive_function_p): Remove useless assert.
6424 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6426 2017-11-16 Martin Sebor <msebor@redhat.com>
6428 PR tree-optimization/82588
6429 PR tree-optimization/82583
6430 * tree-vrp.c (check_array_ref): Handle flexible array members,
6431 string literals, and inner indices.
6432 (search_for_addr_array): Add detail to diagnostics.
6434 2017-11-16 Nathan Sidwell <nathan@acm.org>
6438 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6439 (SET_DECL_ASSEMBLER_NAME): Forward to
6440 overwrite_decl_assembler_name.
6441 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6442 (overwrite_decl_assembler_name): Declare.
6443 * tree.c (overwrite_decl_assembler_name): New.
6444 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6445 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6446 (LANG_HOOKS_INITIALIZER): Add it.
6447 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6448 * langhooks.c (lhd_set_decl_assembler_name): Use
6449 SET_DECL_ASSEMBLER_NAME.
6450 (lhd_overwrite_decl_assembler_name): Default implementation.
6452 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6453 Jackson Woodruff <jackson.woodruff@arm.com>
6455 PR tree-optimization/71026
6456 * match.pd: Canonicalize constant multiplies in division.
6458 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6460 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6461 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6462 Remove OPT_fomit_frame_pointer entry.
6463 * common/config/arc/arc-common.c: Likewise.
6464 * common/config/arm/arm-common.c: Likewise.
6465 * common/config/avr/avr-common.c: Likewise.
6466 * common/config/c6x/c6x-common.c: Likewise.
6467 * common/config/cr16/cr16-common.c: Likewise.
6468 * common/config/cris/cris-common.c: Likewise.
6469 * common/config/epiphany/epiphany-common.c: Likewise.
6470 * common/config/fr30/fr30-common.c: Likewise.
6471 * common/config/frv/frv-common.c: Likewise.
6472 * common/config/ia64/ia64-common.c: Likewise.
6473 * common/config/iq2000/iq2000-common.c: Likewise.
6474 * common/config/lm32/lm32-common.c: Likewise.
6475 * common/config/m32r/m32r-common.c: Likewise.
6476 * common/config/mcore/mcore-common.c: Likewise.
6477 * common/config/microblaze/microblaze-common.c: Likewise.
6478 * common/config/mips/mips-common.c: Likewise.
6479 * common/config/mmix/mmix-common.c: Likewise.
6480 * common/config/mn10300/mn10300-common.c: Likewise.
6481 * common/config/nios2/nios2-common.c: Likewise.
6482 * common/config/pa/pa-common.c: Likewise.
6483 * common/config/pdp11/pdp11-common.c: Likewise.
6484 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6485 * common/config/riscv/riscv-common.c: Likewise.
6486 * common/config/rs6000/rs6000-common.c: Likewise.
6487 * common/config/rx/rx-common.c: Likewise.
6488 * common/config/s390/s390-common.c: Likewise.
6489 * common/config/sh/sh-common.c: Likewise.
6490 * common/config/sparc/sparc-common.c: Likewise.
6491 * common/config/tilegx/tilegx-common.c: Likewise.
6492 * common/config/tilepro/tilepro-common.c: Likewise.
6493 * common/config/v850/v850-common.c: Likewise.
6494 * common/config/visium/visium-common.c: Likewise.
6495 * common/config/xstormy16/xstormy16-common.c: Likewise.
6496 * common/config/xtensa/xtensa-common.c: Likewise.
6497 * invoke.texi (-fomit-frame-pointer): Update documentation.
6499 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6501 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6503 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6505 * cfg.c (scale_bbs_frequencies_int,
6506 cale_bbs_frequencies_gcov_type): Remove.
6507 * cfg.h (scale_bbs_frequencies_int,
6508 cale_bbs_frequencies_gcov_type): Remove.
6510 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6512 * tree-ssa-loop-manip.c
6513 (scale_dominated_blocks_in_loop): Update to profile counts.
6514 (tree_transform_and_unroll_loop): Likewise.
6516 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6518 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6519 scale_bbs_frequencies_int.
6521 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6523 * final.c (compute_alignments): Use counts rather than frequencies.
6525 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6527 * cfgloopanal.c: Include sreal.h
6528 (average_num_loop_insns): Use counts and sreal for accounting.
6530 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6532 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6535 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6536 Kito Cheng <kito.cheng@gmail.com>
6538 * config/nds32/constraints.md: Provide more constraints.
6539 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6540 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6541 support constraints usage.
6543 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6545 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6547 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6548 Kito Cheng <kito.cheng@gmail.com>
6550 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6551 * config/nds32/nds32.opt: Refine the layout.
6552 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6553 TARGET_EXT_STRING): Support new options.
6554 * config/nds32/nds32.h: Likewise.
6555 * config/nds32/nds32.md: Likewise.
6556 * config/nds32/nds32-predicates.c: Likewise.
6557 * config/nds32/constraints.md: Likewise.
6558 * common/config/nds32/nds32-common.c: Likewise.
6560 2017-11-16 Julia Koval <julia.koval@intel.com>
6563 * config/i386/gfniintrin.h: Add sse check.
6564 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6566 2017-11-16 Julia Koval <julia.koval@intel.com>
6568 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6569 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6570 (ix86_handle_option): Handle -mavx512vbmi2.
6571 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6572 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6573 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6574 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6575 (ix86_valid_target_attribute_inner_p): Ditto.
6576 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6577 * config/i386/i386.opt (mavx512vbmi2): New option.
6578 * doc/invoke.texi: Add new option.
6580 2017-11-16 Julia Koval <julia.koval@intel.com>
6582 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6583 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6584 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6585 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6586 _mm512_gf2p8mul_epi8): New intrinsics.
6587 * config/i386/i386-builtin-types.def
6588 (V64QI_FTYPE_V64QI_V64QI): New type.
6589 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6590 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6591 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6592 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6593 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6594 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6596 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6598 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6599 explicitly as a stream of bytes.
6601 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6603 * config/rs6000/altivec.h (vec_xst_be): New #define.
6604 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6605 and externalize from *altivec_vperm_<mode>_internal.
6606 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6608 (XL_BE_V8HI): Likewise.
6609 (XL_BE_V4SI): Likewise.
6610 (XL_BE_V4SI): Likewise.
6611 (XL_BE_V2DI): Likewise.
6612 (XL_BE_V4SF): Likewise.
6613 (XL_BE_V2DF): Likewise.
6614 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6615 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6616 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6617 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6618 VSX_BUILTIN_VEC_XST_BE.
6619 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6620 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6622 (altivec_init_builtins): Replace conditional calls to def_builtin
6623 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6624 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6625 with unconditional calls. Remove calls to def_builtin for
6626 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6627 __builtin_vec_xst_be.
6628 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6629 to define_expand, and add alternate RTL generation for P8.
6630 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6631 vsx_ld_elemrev_v8hi.
6632 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6633 add alternate RTL generation for P8.
6634 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6635 vsx_ld_elemrev_v16qi.
6636 (vsx_st_elemrev_v8hi): Convert define_insn
6637 to define_expand, and add alternate RTL generation for P8.
6638 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6639 vsx_st_elemrev_v8hi.
6640 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6641 add alternate RTL generation for P8.
6642 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6643 vsx_st_elemrev_v16qi.
6645 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6648 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6649 TARGET_AVX512ER check.
6650 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6651 neither -mzeroupper nor -mno-zeroupper is used and
6652 TARGET_EMIT_VZEROUPPER is set.
6653 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6654 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6656 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6658 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6659 folding of vector compares.
6660 (fold_build_vec_cmp): New helper function.
6661 (fold_compare_helper): New helper function.
6662 (builtin_function_type): Add compare builtins to the list of functions
6663 having unsigned arguments. Cosmetic updates to comment indentation.
6664 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6665 the not+eq combination.
6667 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6669 PR tree-optimization/82726
6670 PR tree-optimization/70754
6671 * tree-predcom.c (order_drefs_by_pos): New function.
6672 (combine_chains): Move code setting has_max_use_after to...
6673 (try_combine_chains): ...here. New parameter. Sort combined chains
6674 according to position information.
6675 (tree_predictive_commoning_loop): Update call to above function.
6676 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6678 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6680 PR tree-optimization/82726
6682 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6684 PR tree-optimization/70754
6685 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6686 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6687 combined stmt before it if not NULL.
6688 (combine_chains): Process refs reversely and compute dominance point
6692 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6694 PR tree-optimization/79663
6695 * tree-predcom.c (combine_chains): Process refs in reverse order
6696 only for ZERO length chains, and add explaining comment.
6698 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6700 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6701 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6702 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6704 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6706 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6708 2017-11-15 Martin Liska <mliska@suse.cz>
6710 * tree-cfg.c (pass_warn_function_return::execute):
6711 Compare warn_return_type for greater than zero.
6713 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6717 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6718 to return true on Xeon and not on Xeon Phi.
6719 (ix86_check_avx256_register): Changed to ...
6720 (ix86_check_avx_upper_register): ... this. Add extra check for
6721 VALID_AVX512F_REG_OR_XI_MODE.
6722 (ix86_avx_u128_mode_needed): Changed
6723 ix86_check_avx256_register to ix86_check_avx_upper_register.
6724 (ix86_check_avx256_stores): Changed to ...
6725 (ix86_check_avx_upper_stores): ... this. Changed
6726 ix86_check_avx256_register to ix86_check_avx_upper_register.
6727 (ix86_avx_u128_mode_after): Changed
6728 avx_reg256_found to avx_upper_reg_found. Changed
6729 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6730 (ix86_avx_u128_mode_entry): Changed
6731 ix86_check_avx256_register to ix86_check_avx_upper_register.
6732 (ix86_avx_u128_mode_exit): Ditto.
6733 * config/i386/i386.h: (host_detect_local_cpu): New define.
6735 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6737 * config/arm/xgene1.md (xgene1): Split into automatons
6738 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6739 (xgene1_f_load): Adjust reservations and/or types.
6740 (xgene1_f_store): Likewise.
6741 (xgene1_load_pair): Likewise.
6742 (xgene1_store_pair): Likewise.
6743 (xgene1_fp_load1): Likewise.
6744 (xgene1_load1): Likewise.
6745 (xgene1_store1): Likewise.
6746 (xgene1_move): Likewise.
6747 (xgene1_alu): Likewise.
6748 (xgene1_simd): Likewise.
6749 (xgene1_bfm): Likewise.
6750 (xgene1_neon_load1): Likewise.
6751 (xgene1_neon_store1): Likewise.
6752 (xgene1_neon_logic): Likewise.
6753 (xgene1_neon_st1): Likewise.
6754 (xgene1_neon_ld1r): Likewise.
6755 (xgene1_alu_cond): Added.
6756 (xgene1_shift_reg): Likwise.
6757 (xgene1_bfx): Likewise.
6758 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6760 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6763 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6765 (can_widen_mult_without_libcall): New function.
6766 (expand_mul_overflow): If only checking unsigned mul overflow,
6767 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6768 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6769 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6770 (expand_DIVMOD): Formatting fix.
6771 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6772 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6773 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6775 PR tree-optimization/82977
6776 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6777 constructed temporary to strlen_to_stridx.put.
6779 2017-11-15 Martin Liska <mliska@suse.cz>
6781 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6782 * configure: Regenerate.
6784 2017-11-15 Martin Liska <mliska@suse.cz>
6787 * config/sh/sh-mem.cc: Use proper probability for
6790 2017-11-14 Jeff Law <law@redhat.com>
6792 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6793 the red zone for stack_clash_protection_final_dynamic_probe targets
6794 when the total dynamic stack size is zero bytes.
6796 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6797 blocks is post order.
6799 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6801 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6802 * final.c (rest_of_clean_state): Set it for the
6803 -fcompare-debug dump.
6804 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6805 class when TDF_COMPARE_DEBUG is set.
6807 * dwarf2out.c (gen_producer_string): Discard
6810 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6813 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6814 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6815 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6816 (__TGMATH_REAL_2_3): Remove macros.
6817 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6818 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6820 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6822 (remquo): Define using __TGMATH_REAL_3.
6824 2017-11-14 Jeff Law <law@redhat.com>
6826 * vr-values.c: New file with contents extracted from tree-vrp.c.
6827 * Makefile.in (OBJS): Add vr-values.o
6828 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6829 (set_value_range, set_and_canonicalize_value_range): Likewise.
6830 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6831 (value_range_constant_singleton, symbolic_range_p): Likewise.
6832 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6833 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6834 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6835 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6836 (find_case_label_range, find_case_label_index): Likewise.
6837 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6838 (range_int_cst_singleton_p, value_inside_range): Likewise.
6839 (get_single_symbol): Likewise.
6840 (switch_update): Move structure definition here.
6841 (to_remove_edges, to_update_switch_stmts): Provide externs.
6842 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6843 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6844 (vrp_val_is_min, set_value_range): Likewise.
6845 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6846 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6847 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6848 (range_is_nonnull, range_int_cst_p): Likewwise.
6849 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6850 (get_single_symbol, operand_less_p): Likewise
6851 (compare_values_warnv, compare_values): Likewise.
6852 (value_inside_range, value_range_constant_singleton): Likewise.
6853 (zero_nonzero_bitgs_from_vr): Likewise.
6854 (extract_range_from_binary_expr_1): Likewise.
6855 (overflow_comparison_p): Likewise.
6856 (to_remove_edges, to_update_switch_stmts): Likewise.
6857 (find_case_label-index, find_case_label_range): Likewise.
6858 (switch_update, set_value_range_to_nonnegative): Remove.
6859 (set_value_range_to_truthvalue): Likewise.
6860 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6861 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6862 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6863 (find_case_label_ranges, test_for_singularity): Likewise.
6864 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6865 (x_vr_values): Move to its remaining use site.
6867 2017-11-10 Jeff Law <law@redhat.com>
6869 * vr-values.h (VR_INITIALIZER): Move #define here.
6870 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6871 * Makefile.in (OBJS): Add tree-evrp.o
6872 * tree-vrp.h (assert_info): Move structure definition here.
6873 (set_value_range_to_varying): Prototype.
6874 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6875 (infer_value_range, register_edge_assert_for): Likewise.
6876 (stmt_interesting_for_vrp): Likewise.
6877 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6878 (set_value_range_to_varying): No longer static.
6879 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6880 (infer_value_range, register_edge_assert_for): Likewise.
6882 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6884 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6885 generate the XXBRD instruction.
6887 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6888 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6889 as being compatible if -mabi=ieeelongdouble.
6890 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6891 to setup float128 built-ins with hardware support.
6892 (BU_FLOAT128_HW_2): Likewise.
6893 (BU_FLOAT128_HW_3): Likewise.
6894 (BU_FLOAT128_HW_VSX_1): Likewise.
6895 (BU_FLOAT128_HW_VSX_2): Likewise.
6896 (scalar_extract_expq): Change float128 built-in functions to
6897 accommodate having both KFmode and TFmode functions. Use the
6898 KFmode variant as the default.
6899 (scalar_extract_sigq): Likewise.
6900 (scalar_test_neg_qp): Likewise.
6901 (scalar_insert_exp_q): Likewise.
6902 (scalar_insert_exp_qp): Likewise.
6903 (scalar_test_data_class_qp): Likewise.
6904 (sqrtf128_round_to_odd): Delete processing the round to odd
6905 built-in functions as special built-in functions, and define them
6906 as float128 built-ins. Use the KFmode variant as the default.
6907 (truncf128_round_to_odd): Likewise.
6908 (addf128_round_to_odd): Likewise.
6909 (subf128_round_to_odd): Likewise.
6910 (mulf128_round_to_odd): Likewise.
6911 (divf128_round_to_odd): Likewise.
6912 (fmaf128_round_to_odd): Likewise.
6913 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6914 support for KFmode and TFmode xststdcqp calls.
6915 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6916 point, switch the built-in handlers for the get/set float128
6917 exponent, get float128 mantissa, float128 test built-ins, and the
6918 float128 round to odd built-in functions. Eliminate creating the
6919 float128 round to odd built-in functions as special built-ins.
6920 (rs6000_init_builtins): Eliminate special creation of the float128
6921 round to odd built-in functions.
6922 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6923 function insns to support both TFmode and KFmode variants.
6924 (xsxsigqp_<mode>): Likewise.
6925 (xsiexpqpf_<mode>): Likewise.
6926 (xsiexpqp_<mode>): Likewise.
6927 (xststdcqp_<mode>): Likewise.
6928 (xststdcnegqp_<mode>): Likewise.
6929 (xststdcqp_<mode>): Likewise.
6931 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6933 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6934 unused path_in_freq_ptr parameter.
6935 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6937 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6939 * ipa-inline.c (edge_badness): Dump sreal frequency.
6940 (compute_inlined_call_time): Match natural implementaiton ...
6941 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6942 forgotten division by CGRAPH_FREQ_BASE.
6944 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6946 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6947 Solaris 11. Update comment.
6948 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6950 * config/sol2.h (STARTFILE_SPEC): Likewise.
6951 * configure: Regenerate.
6953 2017-11-14 Carl Love <cel@us.ibm.com>
6955 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6956 le_ and be_ prefixes to swap* variables. Remove
6957 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6960 2017-11-14 Jason Merrill <jason@redhat.com>
6962 Support GTY((cache)) on hash_map.
6963 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6964 (ggc_cache_remove): Override it instead of ggc_mx.
6965 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6966 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6967 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6968 (simple_cache_map_traits): Override maybe_mx.
6969 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6970 (hash_map): Friend gt_cleare_cache.
6971 (gt_cleare_cache): New.
6972 * tree.h (tree_cache_traits): New hash_map traits class.
6973 (tree_cache_map): New typedef.
6975 2017-11-14 Richard Biener <rguenther@suse.de>
6977 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6978 paramter and handling.
6979 (cleanup_control_flow_bb): Likewise.
6980 (cleanup_control_flow_pre): New helper performing a DFS walk
6981 to call cleanup_control_flow_bb in PRE order.
6982 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6983 via cleanup_control_flow_pre.
6985 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6987 * config/aarch64/aarch64-simd.md
6988 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6989 (*aarch64_simd_bsl<mode>_alt): Likewise.
6990 (aarch64_simd_bsldi_internal): New.
6991 (aarch64_simd_bsldi_alt): Likewise.
6993 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6995 * tracer.c (better_p): Do not compare frequencies.
6996 * reg-stack.c (better_edge): Likewise.
6997 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7000 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7002 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7003 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7004 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7005 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7006 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7007 * predict.c (maybe_hot_frequency_p): Inline to ...
7008 (maybe_hot_count_p): ... here; rewrite to counts.
7009 (counts_to_freqs): Rename to ...
7010 (update_max_bb_count): ... this one.
7011 (expensive_function_p): Use counts.
7012 (estimate_bb_frequencies): Update.
7013 (rebuild_frequencies): Update.
7014 * predict.h (counts_to_freqs): Rename to ...
7015 (update_max_bb_count): ... this one.
7016 * profile.c (compute_branch_probabilities): Add debug info
7017 * tree-inline.c (expand_call_inline): Update debug info.
7018 (optimize_inline_calls): Use update_max_bb_count..
7019 (tree_function_versioning): Use update_max_bb_count..
7020 * value-prof.c (gimple_value_profile_transformations):
7021 Do not use update_max_bb_count.
7023 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7025 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7026 always use frequencies.
7028 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7030 * bb-reorder.c: Remove frequencies from comments.
7031 (better_edge_p): Use profile counts.
7032 (find_traces): Dump profile counts.
7033 (rotate_loop): Use profile counts.
7034 (find_traces_1_round): Likewise.
7035 (connect_better_edge_p): Use counts instead of probabilities for
7037 (copy_bb_p): Drop early check for non-0 frequency.
7038 (sanitize_hot_paths): Update comments.
7040 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7042 * ipa-split.c (struct split_point): Add count.
7043 (consider_split): Do not compute incoming frequency; compute incoming
7044 count and store it to split_point.
7045 (split_function): Set count of the call to split part correctly.
7047 2017-11-13 Carl Love <cel@us.ibm.com>
7049 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7051 2017-11-13 Tom Tromey <tom@tromey.com>
7053 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7055 2017-11-13 Carl Love <cel@us.ibm.com>
7057 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7058 Add support for builtins:
7059 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7060 vector {un,}signed {char,int,short},
7061 vector {un,}signed {char,int,short}) arguments.
7062 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7063 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7064 Add BU_P9V_AV_2 expansions for the builtins.
7065 * config/rs6000/altivec.h (vec_first_match_index,
7066 vec_first_mismatch_index, vec_first_match_or_eos_index,
7067 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7068 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7069 new extern declaration.
7070 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7071 * config/rs6000/vsx.md (first_match_index_<mode>,
7072 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7073 first_mismatch_or_eos_index_<mode>): Add define expand.
7074 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7075 * doc/extend.texi: Update the built-in documenation file for the new
7078 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7080 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7081 into the min/max operations for _Float<N> and _Float<N>X types.
7083 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7086 * dwarf2out.c (do_eh_frame): New static variable.
7087 (dwarf2out_begin_prologue): Set it.
7088 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7090 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7092 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7094 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7096 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7099 * debug.h (dwarf2out_do_eh_frame): Declare.
7100 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7101 (dwarf2out_do_frame): Use it.
7102 (dwarf2out_do_cfi_asm): Likewise.
7103 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7104 (dwarf2out_assembly_start): Likewise.
7105 (dwarf2out_begin_prologue): Fix comment.
7106 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7107 if the target needs either debug or unwind DWARF2 info.
7108 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7109 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7110 -ftrapping-math, -ftrapv and -fwrapv.
7112 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7114 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7115 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7116 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7117 (estimate_edge_size_and_time): Likewise.
7118 (ipa_merge_fn_summary_after_inlining): Likewise.
7119 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7120 (compute_uninlined_call_time): Use sreal_frequency.
7121 (compute_inlined_call_time): Likewise.
7122 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7123 * profile-count.c: Include sreal.h.
7124 (profile_count::to_sreal_scale): New.
7125 * profile-count.h: Forward declare sreal.
7126 (profile_count::to_sreal_scale): Declare.
7128 2017-11-13 Nathan Sidwell <nathan@acm.org>
7130 * diagnostic.c (maybe_line_and_column): New.
7131 (diagnostic_get_location_text): Use it.
7132 (diagnostic_report_current_module): Likewise.
7133 (test_diagnostic_get_location_text): Add tests.
7135 2017-11-13 Luis Machado <luis.machado@linaro.org>
7137 * doc/md.texi (Specifying processor pipeline description): Fix
7138 incorrect latency for the div instruction example.
7140 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7142 PR tree-optimization/78821
7143 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7144 that bit_not_p is the same.
7145 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7146 (split_group): Count precisely bit_not_p bits in each statement.
7147 (invert_op): New function.
7148 (imm_store_chain_info::output_merged_store): Use invert_op to
7149 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7150 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7152 2017-11-13 Martin Liska <mliska@suse.cz>
7154 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7155 (struct source_info): Likewise.
7156 (add_branch_counts): Likewise.
7157 (add_line_counts): Likewise.
7158 (function_summary): Likewise.
7159 (output_intermediate_line): Likewise.
7160 (generate_results): Likewise.
7162 2017-11-13 Martin Liska <mliska@suse.cz>
7164 * gcov.c (struct block_info): Remove typedef for block_t.
7165 (struct line_info): Likewise.
7166 (line_info::has_block): Likewise.
7167 (EXIT_BLOCK): Likewise.
7168 (unblock): Likewise.
7169 (circuit): Likewise.
7170 (get_cycles_count): Likewise.
7171 (process_file): Likewise.
7172 (read_graph_file): Likewise.
7173 (solve_flow_graph): Likewise.
7174 (find_exception_blocks): Likewise.
7175 (add_line_counts): Likewise.
7176 (accumulate_line_info): Likewise.
7177 (output_line_details): Likewise.
7179 2017-11-13 Martin Liska <mliska@suse.cz>
7181 * gcov.c (struct arc_info): Remove typedef for arc_t.
7182 (struct line_info): Likewise.
7183 (add_branch_counts): Likewise.
7184 (output_branch_count): Likewise.
7185 (function_info::~function_info): Likewise.
7186 (circuit): Likewise.
7187 (output_intermediate_line): Likewise.
7188 (read_graph_file): Likewise.
7189 (solve_flow_graph): Likewise.
7190 (find_exception_blocks): Likewise.
7191 (add_line_counts): Likewise.
7192 (accumulate_line_info): Likewise.
7193 (output_line_details): Likewise.
7194 (output_function_details): Likewise.
7196 2017-11-13 Martin Liska <mliska@suse.cz>
7198 * gcov.c (struct function_info): Remove typedef for function_t.
7199 (struct source_info): Likewise.
7200 (source_info::get_functions_at_location): Likewise.
7201 (solve_flow_graph): Likewise.
7202 (find_exception_blocks): Likewise.
7203 (add_line_counts): Likewise.
7204 (output_intermediate_file): Likewise.
7205 (process_file): Likewise.
7206 (generate_results): Likewise.
7207 (release_structures): Likewise.
7208 (read_graph_file): Likewise.
7209 (read_count_file): Likewise.
7210 (accumulate_line_counts): Likewise.
7211 (output_lines): Likewise.
7213 2017-11-13 Martin Liska <mliska@suse.cz>
7215 * gcov.c (function_info::function_info): Remove num_counts
7216 and add vector<gcov_type>.
7217 (function_info::~function_info): Use the vector.
7218 (process_file): Likewise.
7219 (read_graph_file): Likewise.
7220 (read_count_file): Likewise.
7221 (solve_flow_graph): Likewise.
7223 2017-11-13 Martin Liska <mliska@suse.cz>
7225 * gcov.c (function_info::is_artificial): New function.
7226 (process_file): Erase all artificial early.
7227 (generate_results): Skip as all artificial are already
7230 2017-11-13 Martin Liska <mliska@suse.cz>
7232 * gcov.c (read_graph_file): Store to global vector of functions.
7233 (read_count_file): Iterate the vector.
7234 (process_file): Likewise.
7235 (generate_results): Likewise.
7236 (release_structures): Likewise.
7238 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7240 PR tree-optimization/82954
7241 * gimple-ssa-store-merging.c
7242 (imm_store_chain_info::coalesce_immediate_stores): If
7243 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7245 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7247 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7248 Upddate call to ENDIAN_LANE_N.
7249 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7250 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7251 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7252 and use aarch64_endian_lane_rtx.
7253 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7255 2017-11-12 Tom de Vries <tom@codesourcery.com>
7257 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7259 2017-11-12 Tom de Vries <tom@codesourcery.com>
7261 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7263 2017-11-12 Tom de Vries <tom@codesourcery.com>
7265 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7267 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7268 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7269 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7271 2017-11-11 Martin Sebor <msebor@redhat.com>
7274 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7277 * prefic.c (translate_name): Replace strncpy with memcpy to
7278 avoid -Wstringop-truncation.
7280 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7282 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7284 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7286 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7288 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7290 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7293 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7295 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7296 enable generating XXBRH if the value is in a vector register.
7297 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7298 value is in a vector register.
7299 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7300 register to register bswap64's instead of doing the GPR sequence
7301 used on previous machines.
7302 (bswapdi2_xxbrd): New insn.
7303 (bswapdi2_reg): Disallow on ISA 3.0.
7304 (register to register bswap64 splitter): Do not split the insn on
7305 ISA 3.0 systems that use XXBRD.
7307 2017-11-10 Martin Sebor <msebor@redhat.com>
7310 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7312 (find_subframework_file): Same.
7314 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7316 * auto-profile.c (afdo_indirect_call): Drop frequency.
7317 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7318 (cgraph_node::create_edge): Drop frequency argument.
7319 (cgraph_node::create_indirect_edge): Drop frequency argument.
7320 (cgraph_edge::make_speculative): Drop frequency arguments.
7321 (cgraph_edge::resolve_speculation): Do not update frequencies
7322 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7323 (cgraph_node::dump): Check consistency in IPA mode.
7324 (cgraph_edge::maybe_hot_p): Use IPA counter.
7325 (cgraph_edge::verify_count_and_frequency): Rename to ...
7326 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7327 (cgraph_node::verify_node): Update.
7328 * cgraph.h (struct cgraph_edge): Drop frequency.
7329 (cgraph_edge::frequency): New function.
7330 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7332 (cgraph_edge::rebuild_edges): Likewise.
7333 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7334 (duplicate_thunk_for_node): Do not pass frequency.
7335 (cgraph_node::create_clone): Scale only counts.
7336 (cgraph_node::create_virtual_clone): Do not pass frequency.
7337 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7338 (cgraph_node::create_version_clone): Do not pass frequency.
7339 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7340 (cgraph_node::expand_thunk): Do not pass frequency.
7341 (cgraph_node::create_wrapper): Do not pass frequency.
7342 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7344 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7345 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7346 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7347 (gather_caller_stats): Use frequency function.
7348 (ipcp_cloning_candidate_p): Use frequency function.
7349 (ipcp_propagate_stage): Use frequency function.
7350 (get_info_about_necessary_edges): Use frequency function.
7351 (update_profiling_info): Update only IPA profile.
7352 (update_specialized_profile): Use frequency functoin.
7353 (perhaps_add_new_callers): Update only IPA profile.
7354 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7355 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7356 (dump_ipa_call_summary): Use frequency function.
7357 (estimate_edge_size_and_time): Use frequency function.
7358 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7359 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7360 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7361 (update_noncloned_counts): ... ths one; scale counts only.
7362 (clone_inlined_nodes): Do not scale frequency.
7363 (inline_call): Do not pass frequency.
7364 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7365 (compute_inlined_call_time): Use IPA profile.
7366 (want_inline_small_function_p): Use IPA profile.
7367 (want_inline_self_recursive_call_p): Use IPA profile.
7368 (edge_badness): Use IPA profile.
7369 (lookup_recursive_calls): Use IPA profile.
7370 (recursive_inlining): Do not pass frequency.
7371 (resolve_noninline_speculation): Do not update frequency.
7372 (inline_small_functions): Collect max of IPA profile.
7373 (dump_overall_stats): Dump IPA porfile.
7374 (dump_inline_stats): Dump IPA porfile.
7375 (ipa_inline): Collect IPA stats.
7376 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7377 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7378 (ipa_propagate_frequency): Use frequency function.
7379 (ipa_profile): Cleanup.
7380 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7381 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7382 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7383 (input_node): Do not stream frequency.
7384 (input_edge): Do not stream frequency.
7385 (merge_profile_summaries): Scale only IPA profiles.
7386 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7387 * predict.c (drop_profile): Do not recompute frequency.
7388 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7389 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7390 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7391 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7392 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7393 * tree-inline.c (copy_bb): Do not scale frequency.
7394 (expand_call_inline): Do not scale frequency.
7395 (tree_function_versioning): Do not scale frequency.
7396 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7398 2017-11-10 Julia Koval <julia.koval@intel.com>
7400 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7401 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7402 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7403 (_mm256_maskz_gf2p8affine_epi64_epi8)
7404 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7405 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7406 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7407 (__builtin_ia32_vgf2p8affineqb_v32qi)
7408 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7409 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7411 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7415 (arm_option_override): Refactor.
7416 (arm_option_reconfigure_globals): New.
7417 (arm_options_perform_arch_sanity_checks): New.
7418 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7420 (arm_options_perform_arch_sanity_checks): Likewise
7422 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7424 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7425 (power9-qpmul): New.
7426 * rs6000/rs6000.md ("type" attr): Add qmul.
7427 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7428 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7429 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7431 2017-11-10 Martin Sebor <msebor@redhat.com>
7434 * builtins.c (compute_objsize): Handle arrays that
7435 compute_builtin_object_size likes to fail for. Make extern.
7436 * builtins.h (compute_objsize): Declare.
7437 (check_strncpy_sizes): New function.
7438 (expand_builtin_strncpy): Call check_strncpy_sizes.
7439 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7440 -Wstringop-truncation.
7441 (gimple_fold_builtin_strncat): Same.
7442 * gimple.c (gimple_build_call_from_tree): Set call location.
7443 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7444 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7445 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7446 (handle_builtin_strlen): Use strlen_to_stridx.
7447 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7449 Use strlen_to_stridx.
7450 (pass_strlen::execute): Release strlen_to_stridx.
7451 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7452 (-Wstringop-truncation): Document new option.
7454 2017-11-10 Martin Liska <mliska@suse.cz>
7456 PR gcov-profile/82702
7457 * gcov.c (main): Handle intermediate files in a different
7459 (get_gcov_intermediate_filename): New function.
7460 (output_gcov_file): Remove support of intermediate files.
7461 (generate_results): Allocate intermediate file.
7462 (release_structures): Clean-up properly fn_end.
7463 (output_intermediate_file): Start iterating with line 1.
7465 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7467 PR tree-optimization/82929
7468 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7469 ops_swapped_p non-static data member.
7470 (store_immediate_info::store_immediate_info): Clear it.
7471 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7472 ops set ops_swapped_p.
7473 (count_multiple_uses): Handle ops_swapped_p.
7475 2017-11-10 Martin Liska <mliska@suse.cz>
7477 * coverage.c (coverage_init): Stream information about
7478 support of has_unexecuted_blocks.
7479 * doc/gcov.texi: Document that.
7480 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7481 * gcov.c (read_graph_file): Likewise.
7482 (output_line_beginning): Fix a small issue with
7485 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7487 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7488 reference of trivial component.
7490 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7493 * gimple-ssa-store-merging.c
7494 (pass_store_merging::terminate_all_aliasing_chains): For
7495 gimple_store_p stmts also call refs_output_dependent_p.
7497 PR rtl-optimization/82913
7498 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7501 2017-11-09 Jeff Law <law@redhat.com>
7503 * vr-values.h: New file with vr_values class.
7504 * tree-vrp.c: Include vr-values.h
7505 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7506 data objects into the vr_values class.
7507 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7508 (get_value_range): Make it a member function within vr_values class.
7509 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7510 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7511 (op_with_constant_singleton_value_range): Likewise.
7512 (extract_range_for_var_from_comparison_expr): Likewise.
7513 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7514 (extract_range_from_binary_expr): Likewise.
7515 (extract_range_from_unary_expr): Likewise.
7516 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7517 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7518 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7519 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7520 (compare_name_with_value, compare_names): Likewise.
7521 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7522 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7523 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7524 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7525 (extract_range_from_phi_node): Likewise.
7526 (simplify_truth_ops_using_ranges): Likewise.
7527 (simplify_div_or_mod_using_ranges): Likewise.
7528 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7529 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7530 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7531 (simplify_float_conversion_using_ranges): Likewise.
7532 (simplify_internal_call_using_ranges): Likewise.
7533 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7534 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7535 poitner via x_vr_values for calls into gimple folder.
7536 (vrp_initialize_lattice): Make this the vr_values ctor.
7537 (vrp_free_lattice): Make this the vr_values dtor.
7538 (set_vr_value): New function.
7539 (class vrp_prop): Add vr_values data member. Add various member
7540 functions as well as member functions that delegate to vr_values.
7541 (check_array_ref): Make a member function within vrp_prop class.
7542 (search_for_addr_array, vrp_initialize): Likewise.
7543 (vrp_finalize): Likewise. Revamp to avoid direct access to
7544 vr_value, values_propagated, etc.
7545 (check_array_bounds): Extract vrp_prop class instance pointer from
7546 walk info structure. Use it to call member functions.
7547 (check_all_array_refs): Make a member function within vrp_prop class.
7548 Smuggle class instance pointer via walk info structure.
7549 (x_vr_values): New local static.
7550 (vrp_valueize): Use x_vr_values to get class instance.
7551 (vr_valueize_1): Likewise.
7552 (class vrp_folder): Add vr_values data member. Add various member
7553 functions as well as member functions that delegate to vr_values.
7554 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7555 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7556 class instance from vr_values. Use it to call member functions.
7557 (vrp_dom_walker): Add vr_values data member.
7558 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7559 instance via x_vr_values.
7560 (identify_jump_threads): Accept vr_values as argument. Store
7561 it into the walker structure.
7562 (evrp_dom_walker): Add vr_values class data member. Add various
7564 (evrp_dom_walker::try_find_new_range): Use vr_values data
7565 member to access the memory allocator.
7566 (evrp_dom_walker::before_dom_children): Store vr_values class
7567 instance into the vrp_folder class.
7568 (evrp_dom_walker::push_value_range): Rework to avoid direct
7569 access to num_vr_values and vr_value.
7570 (evrp_dom_walker::pop_value_range): Likewise.
7571 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7572 Use vr_values to get to dump_all_value_ranges member function.
7573 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7574 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7575 Pass vr_values class instance down to identify_jump_threads.
7576 Remove call to vrp_free_lattice.
7577 (debug_all_value_ranges): Remove.
7579 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7580 (vrp_folder): Likewise.
7582 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7583 Get it from the existing bitmap instead.
7584 (vrp_intersect_ranges_1): Likewise.
7586 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7588 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7590 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7591 set bit_not_p to it.
7592 (imm_store_chain_info::coalesce_immediate_stores): Break group
7593 if bit_not_p is different.
7594 (count_multiple_uses, split_group,
7595 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7596 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7597 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7598 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7599 to store_immediate_info ctor.
7601 2017-11-09 Jim Wilson <jimw@sifive.com>
7603 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7604 (scan_prog_file): Likewise.
7606 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7608 * bb-reorder.c (max_entry_frequency): Remove.
7609 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7610 connect_traces, push_to_next_round_p): Remove prototypes.
7611 (find_traces_1_round): Use counts only.
7612 (push_to_next_round_p): Likewise.
7613 (find_traces): Likewise.
7614 (rotate_loop): Likewise.
7615 (find_traces_1_round): Likewise.
7616 (connect_traces): Likewise.
7617 (edge_order): Likewise.
7619 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7621 * config/arm/arm.c (output_return_instruction): Add comments to
7622 indicate requirement for cmse_nonsecure_entry return to account
7623 for the size of clearing instruction output here.
7624 (thumb_exit): Likewise.
7625 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7626 return in hardfloat mode.
7628 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7630 * config/rs6000/rs6000.c (machine_function): Add a bool,
7631 "toc_is_wrapped_separately".
7632 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7633 if it wasn't explicitly set or unset, we are optimizing for speed, and
7634 doing separate shrink-wrapping.
7635 (rs6000_get_separate_components): Enable the TOC component if
7636 saving the TOC register in the prologue.
7637 (rs6000_components_for_bb): Handle the TOC component.
7638 (rs6000_emit_prologue_components): Store the TOC register where needed.
7639 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7640 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7642 2017-11-09 Martin Jambor <mjambor@suse.cz>
7644 * ipa-param-manipulation.c: New file.
7645 * ipa-param-manipulation.h: Likewise.
7646 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7647 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7648 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7649 (ipa_parm_adjustment): Likewise.
7650 (ipa_parm_adjustment_vec): Likewise.
7651 (ipa_get_vector_of_formal_parms): Moved declaration to
7652 ipa-param-manipulation.h.
7653 (ipa_get_vector_of_formal_parm_types): Likewise.
7654 (ipa_modify_formal_parameters): Likewise.
7655 (ipa_modify_call_arguments): Likewise.
7656 (ipa_combine_adjustments): Likewise.
7657 (ipa_dump_param_adjustments): Likewise.
7658 (ipa_modify_expr): Likewise.
7659 (ipa_get_adjustment_candidate): Likewise.
7660 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7661 ipa-param-manipulation.c.
7662 (ipa_get_vector_of_formal_parm_types): Likewise.
7663 (ipa_modify_formal_parameters): Likewise.
7664 (ipa_modify_call_arguments): Likewise.
7665 (ipa_modify_expr): Likewise.
7666 (get_ssa_base_param): Likewise.
7667 (ipa_get_adjustment_candidate): Likewise.
7668 (index_in_adjustments_multiple_times_p): Likewise.
7669 (ipa_combine_adjustments): Likewise.
7670 (ipa_dump_param_adjustments): Likewise.
7671 * tree-sra.c: Also include ipa-param-manipulation.h
7672 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7675 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7676 Alan Hayward <alan.hayward@arm.com>
7677 David Sherwood <david.sherwood@arm.com>
7679 * doc/sourcebuild.texi (vect_masked_store): Document.
7681 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7682 Alan Hayward <alan.hayward@arm.com>
7683 David Sherwood <david.sherwood@arm.com>
7685 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7687 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7688 Alan Hayward <alan.hayward@arm.com>
7689 David Sherwood <david.sherwood@arm.com>
7691 * doc/sourcebuild.texi (vect_variable_length): Document.
7693 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7694 Alan Hayward <alan.hayward@arm.com>
7695 David Sherwood <david.sherwood@arm.com>
7697 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7699 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7700 Alan Hayward <alan.hayward@arm.com>
7701 David Sherwood <david.sherwood@arm.com>
7703 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7705 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7706 Alan Hayward <alan.hayward@arm.com>
7707 David Sherwood <david.sherwood@arm.com>
7709 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7710 previously undocumented selectors.
7711 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7713 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7715 * doc/rtl.texi (const_vector): Say that elements can be
7716 const_wide_ints too.
7717 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7718 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7719 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7720 * optabs.c (expand_vector_broadcast): Likewise.
7722 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7723 Alan Hayward <alan.hayward@arm.com>
7724 David Sherwood <david.sherwood@arm.com>
7726 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7727 scaled index even if the unscaled address was invalid.
7728 Don't increase the complexity of using a scale in that case.
7730 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7731 Alan Hayward <alan.hayward@arm.com>
7732 David Sherwood <david.sherwood@arm.com>
7734 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7735 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7736 UNITS_PER_WORD bytes.
7737 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7738 based on REGMODE_NATURAL_SIZE of the inner mode.
7739 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7740 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7741 * expmed.c (lowpart_bit_field_p): Divide the value up into
7742 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7743 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7744 whether something is likely to occupy more than one register.
7746 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7749 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7750 profile_count::adjust_for_ipa_scaling.
7751 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7752 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7754 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7756 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7758 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7759 (split_group): Add total_orig and total_new arguments, estimate the
7760 number of statements related to the store group without store merging
7761 and with store merging.
7762 (imm_store_chain_info::output_merged_store): Adjust split_group
7763 callers, punt if estimated number of statements with store merging
7764 is not smaller than estimated number of statements without it.
7766 (handled_load): Remove has_single_use checks.
7767 (pass_store_merging::process_store): Likewise.
7769 2017-11-09 Richard Biener <rguenther@suse.de>
7771 PR tree-optimization/82902
7772 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7774 2017-11-09 Martin Liska <mliska@suse.cz>
7777 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7780 2017-11-09 Martin Liska <mliska@suse.cz>
7782 PR tree-optimization/82669
7783 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7785 2017-11-09 Martin Liska <mliska@suse.cz>
7787 PR gcov-profile/48463
7788 * coverage.c (coverage_begin_function): Output also end locus
7789 of a function and information whether the function is
7791 * gcov-dump.c (tag_function): Parse and print the information.
7792 * gcov.c (INCLUDE_MAP): Add include.
7793 (INCLUDE_SET): Likewise.
7794 (struct line_info): Move earlier in the source file because
7795 of vector<line_info> in function_info structure.
7796 (line_info::line_info): Likewise.
7797 (line_info::has_block): Likewise.
7798 (struct source_info): Add new member index.
7799 (source_info::get_functions_at_location): New function.
7800 (function_info::group_line_p): New function.
7801 (output_intermediate_line): New function.
7802 (output_intermediate_file): Use the mentioned function.
7803 (struct function_start): New.
7804 (struct function_start_pair_hash): Likewise.
7805 (process_file): Add code that identifies group functions.
7806 Assign lines either to global or function scope.
7807 (generate_results): Skip artificial functions.
7808 (find_source): Assign index for each source file.
7809 (read_graph_file): Read new flag artificial and end_line.
7810 (add_line_counts): Assign it either to global of function scope.
7811 (accumulate_line_counts): Isolate core of the function to
7812 accumulate_line_info and call it for both function and global
7814 (accumulate_line_info): New function.
7815 (output_line_beginning): Fix GNU coding style.
7816 (print_source_line): New function.
7817 (output_line_details): Likewise.
7818 (output_function_details): Likewise.
7819 (output_lines): Iterate both source (global) scope and function
7821 (struct function_line_start_cmp): New class.
7822 * doc/gcov.texi: Reflect changes in documentation.
7824 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7827 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7828 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7829 and similarly for not instead of neg.
7831 2017-11-08 Andi Kleen <ak@linux.intel.com>
7833 * config/i386/i386.opt: Add -mforce-indirect-call.
7834 * config/i386/predicates.md: Check for flag_force_indirect_call.
7835 * doc/invoke.texi: Document -mforce-indirect-call
7837 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7839 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7841 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7842 config/riscv/riscv.c (predict.h): New include.
7843 (riscv_slow_unaligned_access_p): No longer static.
7844 (riscv_block_move_straight): Add require.
7845 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7847 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7850 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7851 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7852 nonimmediate_operand predicate for operand 1 instead of
7855 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7857 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7859 * config/aarch64/constraints.md (Uml): New constraint.
7860 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7863 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7865 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7866 of two vec_duplicates into a vec_concat.
7868 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7870 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7871 Simplify vec_merge of vec_duplicate and vec_concat.
7872 * config/aarch64/constraints.md (Utq): New constraint.
7873 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7876 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7878 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7879 Simplify vec_merge of vec_duplicate and const_vector.
7880 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7882 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7883 mode iterator. Update predicate on operand 1 to
7884 handle non-const_vec constants. Delete constraints.
7885 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7887 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7889 PR tree-optimization/78821
7890 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7892 (store_operand_info::store_operand_info): Initialize it to false.
7893 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7894 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7895 store in the group, and if chain_info is non-NULL, to ignore altogether
7897 (compatible_load_p): Fail if bit_not_p does not match.
7898 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7899 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7900 (pass_store_merging::process_store): Adjust
7901 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7902 call terminate_all_aliasing_chains newly when adding a store into
7903 a chain with non-NULL chain_info.
7905 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7907 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7909 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7911 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7913 (aarch64_layout_frame): Initialise emit_frame_chain.
7914 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7915 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7917 2017-11-08 Martin Liska <mliska@suse.cz>
7919 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7920 of gimple_call_internal_p.
7922 2017-11-07 Tom de Vries <tom@codesourcery.com>
7924 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7926 2017-11-07 Tom de Vries <tom@codesourcery.com>
7928 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7931 2017-11-08 Martin Liska <mliska@suse.cz>
7934 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7936 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7938 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7940 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7943 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7944 (*cmp<mode>_ccz_1): New insn with $k alternative.
7947 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7948 RTX_COMM_COMPARE as commutative as well.
7949 (ix86_binary_operator_ok): Formatting fix.
7950 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7951 *<code><mode>3<mask_name><round_saeonly_name>,
7952 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7953 *<s>mul<mode>3_highpart<mask_name>,
7954 *vec_widen_umult_even_v16si<mask_name>,
7955 *vec_widen_umult_even_v8si<mask_name>,
7956 *vec_widen_umult_even_v4si<mask_name>,
7957 *vec_widen_smult_even_v16si<mask_name>,
7958 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7959 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7960 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7961 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7962 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7963 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7964 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7965 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7966 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7967 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7968 ix86_binary_operator_ok. Formatting fixes.
7969 (*<plusminus_insn><mode>3<mask_name><round_name>,
7970 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7973 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7975 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7976 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7977 eq and ne if TARGET_ISEL.
7978 (cmp): New code_iterator.
7979 (UNS, UNSU_, UNSIK): New code_attrs.
7980 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7981 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7983 ("eq<mode>3"): ... this.
7984 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7986 ("ne<mode>3"): ... this.
7988 2017-11-07 Julia Koval <julia.koval@intel.com>
7991 * common/config/i386/i386-common.c
7992 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7993 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7994 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7995 * config/i386/i386.opt: Ditto.
7996 * config/i386/i386.c (ix86_target_string): Ditto.
7997 (ix86_option_override_internal): Ditto.
7998 (ix86_init_mpx_builtins): Move MPX to args2.
7999 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8000 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8001 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8002 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8003 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8004 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8005 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8007 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8010 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8011 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8012 (zero-extendsidi peephole2): Remove peephole.
8014 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8017 * stor-layout.c: Include attribs.h.
8018 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8019 explicit lookup of "aligned" attribute.
8021 2017-11-07 Andrew Waterman <andrew@sifive.com>
8023 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8024 (riscv_expand_block_move): Likewise.
8025 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8027 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8028 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8029 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8030 (riscv_adjust_block_mem): Likewise.
8031 (riscv_block_move_loop): Likewise.
8032 (riscv_expand_block_move): Likewise.
8033 * config/riscv/riscv.md (movmemsi): New pattern.
8035 2017-11-07 Michael Clark <michaeljclark@mac.com>
8037 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8038 (MUSL_DYNAMIC_LINKER): Likewise.
8040 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8042 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8043 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8046 2017-11-07 Richard Biener <rguenther@suse.de>
8048 * match.pd: Fix build.
8050 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8051 Jackson Woodruff <jackson.woodruff@arm.com>
8053 PR tree-optimization/71026
8054 * match.pd: Canonicalize negate in division.
8056 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8059 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8061 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8063 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8064 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8066 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8068 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8069 non-scalar integral types.
8070 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8071 (-(A-B), -(~A)): New transformations.
8073 2017-11-07 Tom de Vries <tom@codesourcery.com>
8075 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8076 semicolon after "do {} while (0)".
8077 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8078 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8079 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8080 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8081 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8083 2017-11-07 Tom de Vries <tom@codesourcery.com>
8085 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8086 after "do {} while (0)".
8087 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8088 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8089 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8090 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8091 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8093 2017-11-07 Tom de Vries <tom@codesourcery.com>
8096 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8098 (arm_rtx_costs_internal): Add missing semicolon after
8099 HANDLE_NARROW_SHIFT_ARITH call.
8101 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8103 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8104 disable isel if it was not set explicitly.
8106 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8108 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8109 (add_type_attribute) likewise.
8111 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8113 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8114 of struct ix86_frame.
8115 (ix86_initial_elimination_offset): Likewise.
8116 (ix86_expand_split_stack_prologue): Likewise.
8118 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8120 * tree-vrp.h (enum value_range_type): Update stale comment.
8122 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8123 Alan Hayward <alan.hayward@arm.com>
8124 David Sherwood <david.sherwood@arm.com>
8126 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8127 (aarch64_expand_vec_perm_const): Take the number of units too.
8128 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8129 (aarch64_expand_vec_perm_const): Likewise.
8130 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8131 (vec_perm<mode>): Update accordingly.
8133 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8134 Alan Hayward <alan.hayward@arm.com>
8135 David Sherwood <david.sherwood@arm.com>
8137 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8138 Take the number of units too.
8139 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8140 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8141 but check for a vector mode before rather than after the call.
8142 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8143 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8144 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8145 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8146 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8147 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8148 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8149 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8150 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8151 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8152 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8153 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8154 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8155 (aarch64_sqdmull2_n<mode>): Update accordingly.
8157 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8158 Alan Hayward <alan.hayward@arm.com>
8159 David Sherwood <david.sherwood@arm.com>
8161 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8162 the number of units too.
8163 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8164 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8165 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8166 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8167 (vec_store_lanesxi<mode>): Update accordingly.
8169 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8170 Alan Hayward <alan.hayward@arm.com>
8171 David Sherwood <david.sherwood@arm.com>
8173 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8174 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8175 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8176 of units rather than the mode.
8177 * config/aarch64/iterators.md (nunits): New mode attribute.
8178 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8179 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8180 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8181 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8182 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8183 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8184 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8185 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8186 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8187 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8188 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8189 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8190 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8191 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8192 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8193 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8194 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8195 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8196 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8197 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8198 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8199 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8200 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8201 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8202 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8203 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8204 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8205 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8206 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8207 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8208 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8209 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8210 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8211 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8212 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8213 (aarch64_simd_vec_setv2di): Likewise.
8215 2017-11-06 Carl Love <cel@us.ibm.com>
8217 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8219 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8220 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8222 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8224 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8225 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8226 Add power 8 macro expansions.
8227 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8228 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8229 power 8 instructions. (VSX_XXBR): Add iterator.
8231 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8233 * config/arm/arm.md (predicable_short_it): Change default to "no",
8234 improve documentation, remove uses that are identical to the default.
8235 (enabled_for_depr_it): Rename to enabled_for_short_it.
8236 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8237 Remove default uses.
8238 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8239 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8240 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8241 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8243 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8246 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8247 float128 helper macros, which are no longer used after deleting
8248 the old 'q' built-in functions, and moving the round to odd
8249 built-in functions to being special built-in functions.
8250 (BU_FLOAT128_2): Likewise.
8251 (BU_FLOAT128_1_HW): Likewise.
8252 (BU_FLOAT128_2_HW): Likewise.
8253 (BU_FLOAT128_3_HW): Likewise.
8254 (FABSQ): Delete old 'q' built-in functions.
8255 (COPYSIGNQ): Likewise.
8256 (SQRTF128_ODD): Move round to odd built-in functions to be
8257 special built-in functions, so that we can handle
8258 -mabi=ieeelongdouble.
8259 (TRUNCF128_ODD): Likewise.
8260 (ADDF128_ODD): Likewise.
8261 (SUBF128_ODD): Likewise.
8262 (MULF128_ODD): Likewise.
8263 (DIVF128_ODD): Likewise.
8264 (FMAF128_ODD): Likewise.
8265 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8266 built-in names to 'f128'.
8267 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8268 old 'q' built-in functions, as the machine independent code for
8269 'f128' built-in functions handles this.
8270 (rs6000_expand_builtin): Add expansion for float128 round to odd
8271 functions, keying off on -mabi=ieeelongdouble of whether to use
8272 the KFmode or TFmode variant.
8273 (rs6000_init_builtins): Initialize the _Float128 round to odd
8275 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8276 _Float128 'q' built-in functions are now mapped into the new
8277 'f128' built-in functions.
8279 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8281 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8282 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8284 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8287 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8288 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8289 * toplev.c: Include "ipa-fnsummary.h".
8290 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8292 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8294 PR tree-optimization/82838
8295 * gimple-ssa-store-merging.c
8296 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8297 on a separate gimple_seq which is then appended to seq.
8299 2017-11-06 Jeff Law <law@redhat.com>
8302 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8303 (get_probe_interval): New functions.
8304 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8305 (ix86_adjust_stack_and_probe): Likewise.
8306 (output_adjust_stack_and_probe): Likewise.
8307 (ix86_emit_probe_stack_range): Likewise.
8308 (ix86_expand_prologue): Likewise.
8310 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8312 PR tree-optimization/82816
8313 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8314 if the modes of the two types are the same.
8315 (convert_plusminus_to_widen): Likewise.
8317 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8319 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8321 (usadv16qi): New define_expand.
8322 (usadv8hi): New define_expand.
8324 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8327 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8329 (inline_transform): Likewise.
8330 * predict.c (counts_to_freqs): Remove useless conditional.
8331 * profile-count.h (profile_count::apply_scale): Move sanity check.
8332 * tree-inline.c (copy_bb): Always scale.
8333 (copy_cfg_body): Likewise.
8335 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8338 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8340 (*cmp_ite0): Add enabled_for_depr_it attribute.
8341 (*cmp_ite1): Likewise.
8343 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8345 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8348 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8350 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8351 return the value by pointer.
8352 (extract_range_from_multiplicative_op_1): Update accordingly.
8353 Return as soon as an operation fails.
8355 2017-11-05 Tom de Vries <tom@codesourcery.com>
8358 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8359 (DEF_SANITIZER_BUILTIN): ... here.
8360 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8361 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8363 2017-11-05 Tom de Vries <tom@codesourcery.com>
8366 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8368 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8369 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8370 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8372 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8373 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8374 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8376 2017-11-05 Tom de Vries <tom@codesourcery.com>
8379 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8382 2017-11-04 Michael Clark <michaeljclark@mac.com>
8384 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8385 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8387 (*addsi3_extended): Likewise.
8388 (*addsi3_extended2): Likewise.
8389 (<optab>si3): Likewise.
8390 (<optab>di3): Likewise.
8391 (<optab><mode>3): Likewise.
8392 (<*optabe>si3_internal): Likewise.
8393 (zero_extendqi<SUPERQI:mode>2): Likewise.
8394 (*add<mode>hi3): Likewise.
8395 (*xor<mode>hi3): Likewise.
8396 (<optab>di3): Likewise.
8397 (*<optab>si3_extend): Likewise.
8398 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8399 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8400 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8402 2017-11-04 Andrew Waterman <andrew@sifive.com>
8404 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8405 TARGET_STRICT_ALIGN based upon -mtune argument.
8407 2017-11-04 Andrew Waterman <andrew@sifive.com>
8409 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8411 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8413 * config/i386/i386.c (choose_basereg): Use optional scratch
8414 register and add assertion.
8415 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8416 needed, and don't allocate stack.
8417 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8418 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8419 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8421 2017-11-03 Jeff Law <law@redhat.com>
8423 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8424 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8425 to probe at the start of a noreturn function.
8427 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8429 PR tree-optimization/78821
8430 * gimple-ssa-store-merging.c: Update the file comment.
8431 (MAX_STORE_ALIAS_CHECKS): Define.
8432 (struct store_operand_info): New type.
8433 (store_operand_info::store_operand_info): New constructor.
8434 (struct store_immediate_info): Add rhs_code and ops data members.
8435 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8436 and op1r arguments to the ctor, initialize corresponding data members.
8437 (struct merged_store_group): Add load_align_base and load_align
8439 (merged_store_group::merged_store_group): Initialize them.
8440 (merged_store_group::do_merge): Update them.
8441 (merged_store_group::apply_stores): Pick the constant for
8442 encode_tree_to_bitpos from one of the two operands, or skip
8443 encode_tree_to_bitpos if neither operand is a constant.
8444 (class pass_store_merging): Add process_store method decl. Remove
8445 bool argument from terminate_all_aliasing_chains method decl.
8446 (pass_store_merging::terminate_all_aliasing_chains): Remove
8447 var_offset_p argument and corresponding handling.
8448 (stmts_may_clobber_ref_p): New function.
8449 (compatible_load_p): New function.
8450 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8451 if there is overlap and rhs_code is not INTEGER_CST. For
8452 non-overlapping stores terminate group if rhs is not mergeable.
8453 (get_alias_type_for_stmts): Change first argument from
8454 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8455 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8456 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8458 (get_location_for_stmts): Change first argument from
8459 auto_vec<gimple *> & to vec<gimple *> &.
8460 (struct split_store): Remove orig_stmts data member, add orig_stores.
8461 (split_store::split_store): Create orig_stores rather than orig_stmts.
8462 (find_constituent_stmts): Renamed to ...
8463 (find_constituent_stores): ... this. Change second argument from
8464 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8465 to info structures rather than the statements.
8466 (split_group): Rename ALLOW_UNALIGNED argument to
8467 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8468 it. Adjust find_constituent_stores caller.
8469 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8470 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8471 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8472 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8473 (mem_valid_for_store_merging): New function.
8474 (handled_load): New function.
8475 (pass_store_merging::process_store): New method.
8476 (pass_store_merging::execute): Use process_store method. Adjust
8477 terminate_all_aliasing_chains caller.
8479 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8481 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8482 Return true for more constants, symbols and label references.
8483 (aarch64_valid_floating_const): Remove unused function.
8485 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8488 * config/aarch64/aarch64.c (aarch64_layout_frame):
8489 Undo forcing of LR at bottom of frame.
8491 2017-11-03 Jeff Law <law@redhat.com>
8494 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8495 for int_registers_saved.
8497 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8498 extracted from tree-ssa-dom.c.
8499 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8500 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8501 (record_equivalences_from_incoming_edge): Add additional argument
8502 to single_pred_edge_ignoring_loop_edges call.
8503 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8504 (uncprop_dom_walker::before_dom_children): Add additional argument
8505 to single_pred_edge_ignoring_loop_edges call.
8506 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8507 single_pred_edge_ignoring_loop_edges rather than open coding.
8508 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8510 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8512 * match.pd (-(-A)): Rewrite.
8514 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8516 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8517 (rs6000_emit_int_cmove): New declaration.
8518 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8519 (rs6000_emit_sISEL): Delete.
8520 (rs6000_emit_int_cmove): Make non-static.
8521 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8522 instead of rs6000_emit_sISEL.
8524 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8526 * asan.c (create_cond_insert_point): Maintain profile.
8527 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8529 * basic-block.h (struct basic_block_def): Remove frequency.
8530 (EDGE_FREQUENCY): Use to_frequency
8531 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8533 (find_traces): Update to use to_frequency.
8534 (find_traces_1_round): Likewise; use only IPA counts.
8535 (bb_to_key): Likewise.
8536 (connect_traces): Use IPA counts only.
8537 (copy_bb_p): Update to use to_frequency.
8538 (fix_up_crossing_landing_pad): Likewise.
8539 (sanitize_hot_paths): Likewise.
8540 * bt-load.c (basic_block_freq): Likewise.
8541 * cfg.c (init_flow): Set count_max to uninitialized.
8542 (check_bb_profile): Remove frequencies; check counts.
8543 (dump_bb_info): Do not dump frequencies.
8544 (update_bb_profile_for_threading): Update counts only.
8545 (scale_bbs_frequencies_int): Likewise.
8546 (MAX_SAFE_MULTIPLIER): Remove.
8547 (scale_bbs_frequencies_gcov_type): Update counts only.
8548 (scale_bbs_frequencies_profile_count): Update counts only.
8549 (scale_bbs_frequencies): Update counts only.
8550 * cfg.h (struct control_flow_graph): Add count-max.
8551 (update_bb_profile_for_threading): Update prototype.
8552 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8553 (find_many_sub_basic_blocks): Likewise.
8554 * cfgcleanup.c (try_forward_edges): Likewise.
8555 (try_crossjump_to_edge): Likewise.
8556 * cfgexpand.c (expand_gimple_cond): Likewise.
8557 (expand_gimple_tailcall): Likewise.
8558 (construct_init_block): Likewise.
8559 (construct_exit_block): Likewise.
8560 * cfghooks.c (verify_flow_info): Check consistency of counts.
8561 (dump_bb_for_graph): Do not dump frequencies.
8562 (split_block_1): Do not update frequencies.
8563 (split_edge): Do not update frequencies.
8564 (make_forwarder_block): Do not update frequencies.
8565 (duplicate_block): Do not update frequencies.
8566 (account_profile_record): Do not update frequencies.
8567 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8568 for global heuristics.
8569 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8570 (expected_loop_iterations_unbounded): Use counts only.
8571 * cfgloopmanip.c (scale_loop_profile): Simplify.
8572 (create_empty_loop_on_edge): Simplify
8574 (duplicate_loop_to_header_edge): Simplify
8575 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8576 (update_br_prob_note): Take care of removing note when profile
8578 (relink_block_chain): Do not dump frequency.
8579 (rtl_account_profile_record): Use to_frequency.
8580 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8581 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8582 (cgraph_update_edges_for_call_stmt_node): Likewise.
8583 (cgraph_edge::verify_count_and_frequency): Update.
8584 (cgraph_node::verify_node): Temporarily disable frequency verification.
8585 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8586 to_cgraph_frequency.
8587 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8588 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8590 (cgraph_node::expand_thunk): Update profile.
8591 * except.c (dw2_build_landing_pads): Do not update frequency.
8592 * final.c (compute_alignments): Use to_frequency.
8593 (dump_basic_block_info): Do not dump frequency.
8594 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8595 (dump_gimple_bb_header): Do not dump frequency.
8596 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8598 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8599 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8600 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8601 (init_before_recovery): Do not update frequency.
8602 (sched_create_recovery_edges): Do not update frequency.
8603 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8604 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8605 (ipa_cp_c_finalize): Set max_count to uninitialized.
8606 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8607 (param_change_prob): Use counts.
8608 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8610 * ipa-split.c (consider_split): Use to_frequency.
8611 (split_function): Use to_frequency.
8612 * ira-build.c (loop_compare_func): Likewise.
8613 (mark_loops_for_removal): Likewise.
8614 (mark_all_loops_for_removal): Likewise.
8615 * loop-doloop.c (doloop_modify): Do not update frequency.
8616 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8618 * lto-streamer-in.c (input_function): Update count_max.
8619 * omp-expand.c (expand_omp_taskreg): Update count_max.
8620 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8621 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8622 (maybe_hot_count_p): Use ipa counts only.
8623 (maybe_hot_bb_p): Simplify.
8624 (maybe_hot_edge_p): Simplify.
8625 (probably_never_executed): Do not take frequency argument.
8626 (probably_never_executed_bb_p): Do not pass frequency.
8627 (probably_never_executed_edge_p): Likewise.
8628 (combine_predictions_for_bb): Check that profile is nonzero.
8629 (propagate_freq): Do not set frequency.
8630 (drop_profile): Simplify.
8631 (counts_to_freqs): Simplify.
8632 (expensive_function_p): Use to_frequency.
8633 (propagate_unlikely_bbs_forward): Simplify.
8634 (determine_unlikely_bbs): Simplify.
8635 (estimate_bb_frequencies): Add hack to silence graphite issues.
8636 (compute_function_frequency): Use ipa counts.
8637 (pass_profile::execute): Update.
8638 (rebuild_frequencies): Use counts only.
8639 (force_edge_cold): Use counts only.
8640 * profile-count.c (profile_count::dump): Dump new count types.
8641 (profile_count::differs_from_p): Check compatiblity.
8642 (profile_count::to_frequency): New function.
8643 (profile_count::to_cgraph_frequency): New function.
8644 * profile-count.h (struct function): Declare.
8645 (enum profile_quality): Add profile_guessed_local and
8646 profile_guessed_global0.
8647 (class profile_proability): Decrease number of bits to 29;
8648 update from_reg_br_prob_note and to_reg_br_prob_note.
8649 (class profile_count: Update comment; decrease number of bits
8650 to 61. Check compatibility.
8651 (profile_count::compatible_p): New private member function.
8652 (profile_count::ipa_p): New member function.
8653 (profile_count::operator<): Handle global zero correctly.
8654 (profile_count::operator>): Handle global zero correctly.
8655 (profile_count::operator<=): Handle global zero correctly.
8656 (profile_count::operator>=): Handle global zero correctly.
8657 (profile_count::nonzero_p): New member function.
8658 (profile_count::force_nonzero): New member function.
8659 (profile_count::max): New member function.
8660 (profile_count::apply_scale): Handle IPA scalling.
8661 (profile_count::guessed_local): New member function.
8662 (profile_count::global0): New member function.
8663 (profile_count::ipa): New member function.
8664 (profile_count::to_frequency): Declare.
8665 (profile_count::to_cgraph_frequency): Declare.
8666 * profile.c (OVERLAP_BASE): Delete.
8667 (compute_frequency_overlap): Delete.
8668 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8669 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8670 * sched-ebb.c (rank): Use counts only.
8671 * shrink-wrap.c (handle_simple_exit): Use counts only.
8672 (try_shrink_wrapping): Use counts only.
8673 (place_prologue_for_one_component): Use counts only.
8674 * tracer.c (find_best_predecessor): Use to_frequency.
8675 (find_trace): Use to_frequency.
8676 (tail_duplicate): Use to_frequency.
8677 * trans-mem.c (expand_transaction): Do not update frequency.
8678 * tree-call-cdce.c: Do not update frequency.
8679 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8680 (gimple_merge_blocks): Likewise.
8681 (gimple_split_edge): Likewise.
8682 (gimple_duplicate_sese_region): Likewise.
8683 (gimple_duplicate_sese_tail): Likewise.
8684 (move_sese_region_to_fn): Likewise.
8685 (gimple_account_profile_record): Likewise.
8686 (insert_cond_bb): Likewise.
8687 * tree-complex.c (expand_complex_div_wide): Likewise.
8688 * tree-eh.c (lower_resx): Update profile.
8689 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8691 (initialize_cfun): Do not initialize frequencies
8692 (freqs_to_counts): Delete.
8693 (copy_cfg_body): Ignore count parameter.
8694 (copy_body): Update.
8695 (expand_call_inline): Update count_max.
8696 (optimize_inline_calls): Update count_max.
8697 (tree_function_versioning): Update count_max.
8698 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8699 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8701 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8702 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8703 (try_peel_loop): Likewise.
8704 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8706 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8707 (tree_transform_and_unroll_loop): Do not use frequencies
8708 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8709 Use reliable prediction only.
8710 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8711 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8712 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8713 probability scaling.
8714 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8715 not update frequency
8716 (any_remaining_duplicated_blocks): Likewise.
8717 (update_profile): Likewise.
8718 (estimated_freqs_path): Delete.
8719 (freqs_to_counts_path): Delete.
8720 (clear_counts_path): Delete.
8721 (ssa_fix_duplicate_block_edges): Likewise.
8722 (duplicate_thread_path): Likewise.
8723 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8724 * tree-tailcall.c (decrease_profile): Do not update frequency.
8725 (eliminate_tail_call): Likewise.
8726 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8727 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8728 (optimize_mask_stores): Likewise.
8729 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8730 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8731 (ubsan_expand_ptr_ifn): Update profile.
8732 * value-prof.c (gimple_ic): Simplify.
8733 * value-prof.h (gimple_ic): Update prototype.
8734 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8735 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8737 (want_inline_self_recursive_call_p): Likewise.
8738 (resolve_noninline_speculation): Only cummulate defined counts.
8739 (inline_small_functions): Use nonzero_p.
8740 (ipa_inline): Do not access freed node.
8742 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8744 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8745 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8747 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8749 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8750 non-legitimate address.
8752 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8754 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8755 (*lt0_<mode>di, *lt0_<mode>si): New.
8757 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8759 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8760 TARGET_PAIRED_FLOAT.
8762 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8763 Jim Wilson <jim.wilson@linaro.org>
8765 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8766 * config/aarch64/aarch64-tune.md: Regenerated.
8767 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8768 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8770 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8772 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8773 (arc_expand_prologue): Restore blink for millicode.
8774 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8776 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8779 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8780 gen_vec_duplicate after forcing the scalar into a register.
8782 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8784 * combine (try_combine): Print the insns input to try_combine to the
8787 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8790 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8791 Remove second argument from aarch64_process_target_attr call.
8792 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8794 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8796 (aarch64_handle_attr_arch): Remove second argument.
8797 (aarch64_handle_attr_cpu): Ditto.
8798 (aarch64_handle_attr_tune): Ditto.
8799 (aarch64_handle_attr_isa_flags): Ditto.
8800 (aarch64_process_one_target_attr): Ditto.
8801 (aarch64_process_target_attr): Ditto.
8802 (aarch64_option_valid_attribute_p): Remove second argument.
8803 on aarch64_process_target_attr call.
8805 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8807 * diagnostic.c: Include "selftest-diagnostic.h".
8808 (selftest::assert_location_text): New function.
8809 (selftest::test_diagnostic_get_location_text): New function.
8810 (selftest::diagnostic_c_tests): Call it.
8812 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8814 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8815 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8816 (class selftest::test_diagnostic_context): Move to...
8817 * selftest-diagnostic.c: New file.
8818 * selftest-diagnostic.h: New file.
8820 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8822 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8823 offset range for FT32B.
8824 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8825 * config/ft32/ft32.md: Add TARGET_NOPM.
8826 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8827 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8829 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8831 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8833 2017-11-02 Jeff Law <law@redhat.com>
8835 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8836 virtual keyword on FINAL OVERRIDE members.
8838 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8839 virtuals together. Add virtual destructor.
8840 (substitute_and_fold_engine): Similarly.
8842 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8844 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8846 2017-11-02 Richard Biener <rguenther@suse.de>
8848 PR tree-optimization/82795
8849 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8851 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8853 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8855 (gcc_GAS_CHECK_FEATURE): Remove.
8856 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8857 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8858 * configure: Regenerate.
8860 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8862 * config/arc/arc.c (hwloop_optimize): Account for empty
8865 2017-11-02 Richard Biener <rguenther@suse.de>
8868 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8869 Truncate ARRAY_REF index and element size.
8871 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8873 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8875 2017-11-01 Jeff Law <law@redhat.com>
8877 * tree-ssa-ccp.c (ccp_folder): New class derived from
8878 substitute_and_fold_engine.
8879 (ccp_folder::get_value): New member function.
8880 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8881 (ccp_fold_stmt): Remove prototype.
8882 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8883 * tree-ssa-copy.c (copy_folder): New class derived from
8884 substitute_and_fold_engine.
8885 (copy_folder::get_value): Renamed from get_value.
8886 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8887 * tree-vrp.c (vrp_folder): New class derived from
8888 substitute_and_fold_engine.
8889 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8890 (vrp_folder::get_value): New member function.
8891 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8892 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8893 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8894 provide a class interface to folder/substitute routines.
8895 (ssa_prop_fold_stmt_fn): Remove typedef.
8896 (ssa_prop_get_value_fn): Likewise.
8897 (subsitute_and_fold): Remove prototype.
8898 (replace_uses_in): Likewise.
8899 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8900 Renamed from replace_uses_in. Call the virtual member function
8901 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8902 (substitute_and_fold_dom_walker): Remove initialization of
8903 data member entries for calbacks. Add substitute_and_fold_engine
8904 member and initialize it.
8905 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8906 member functions for get_value, replace_phi_args_in c
8907 replace_uses_in, and fold_stmt calls.
8908 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8909 substitute_and_fold. Remove assert. Update ctor call.
8911 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8912 (ssa_prop_visit_phi_fn): Likewise.
8913 (class ssa_propagation_engine): New class to provide an interface
8915 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8917 (ssa_prop_visit_phi): Likewise.
8918 (ssa_propagation_engine::simulate_stmt): Moved into class.
8919 Call visit_phi/visit_stmt from the class rather than via
8920 file scoped static variables.
8921 (ssa_propagation_engine::simulate_block): Moved into class.
8922 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8923 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8924 set file scoped statics for the visit_stmt/visit_phi callbacks.
8925 * tree-complex.c (complex_propagate): New class derived from
8926 ssa_propagation_engine.
8927 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8928 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8929 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8931 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8932 ssa_propagation_engine.
8933 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8934 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8935 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8936 * tree-ssa-copy.c (copy_prop): New class derived from
8937 ssa_propagation_engine.
8938 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8939 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8940 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8941 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8942 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8943 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8944 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8946 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8948 PR rtl-optimization/82778
8949 PR rtl-optimization/82597
8950 * compare-elim.c (struct comparison): Add in_a_setter field.
8951 (find_comparison_dom_walker::before_dom_children): Remove killed
8952 bitmap and df_simulate_find_defs call, instead walk the defs.
8953 Compute last_setter and initialize in_a_setter. Merge definitions
8954 with first initialization for a few variables.
8955 (try_validate_parallel): Use insn_invalid_p instead of
8956 recog_memoized. Return insn rather than just the pattern.
8957 (try_merge_compare): Fix up comment. Don't uselessly test if
8958 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8960 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8963 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8964 Alan Hayward <alan.hayward@arm.com>
8965 David Sherwood <david.sherwood@arm.com>
8967 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8968 aarch64_hard_regno_nregs to get the number of registers
8971 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8972 Alan Hayward <alan.hayward@arm.com>
8973 David Sherwood <david.sherwood@arm.com>
8975 * config/aarch64/constraints.md (Upl): Rename to...
8977 * config/aarch64/aarch64.md
8978 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8981 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8982 Alan Hayward <alan.hayward@arm.com>
8983 David Sherwood <david.sherwood@arm.com>
8985 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8986 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8989 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8990 Alan Hayward <alan.hayward@arm.com>
8991 David Sherwood <david.sherwood@arm.com>
8993 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8994 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8995 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8997 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8998 of a single element.
8999 * config/aarch64/iterators.md: Add a comment above the permute
9000 unspecs to say that they are generated directly by
9001 aarch64_expand_vec_perm_const.
9002 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9004 2017-11-01 Nathan Sidwell <nathan@acm.org>
9006 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9008 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9010 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9011 and medany code models, and describe what they do.
9013 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9015 Revert accidental duplicate:
9017 * combine.c (can_change_dest_mode): Reject changes in
9018 REGMODE_NATURAL_SIZE.
9020 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9022 PR rtl-optimization/64682
9023 PR rtl-optimization/69567
9024 PR rtl-optimization/69737
9025 PR rtl-optimization/82683
9026 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9027 register mentioned in the note, drop the note, unless it came from I3,
9028 in which case it should go to I3 again.
9030 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9032 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9033 and return false if not.
9034 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9036 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9038 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9039 range is known to be empty.
9041 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9042 Alan Hayward <alan.hayward@arm.com>
9043 David Sherwood <david.sherwood@arm.com>
9045 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9046 and CONST_VECTOR_NUNITS instead of computing the number of units from
9047 the byte sizes of the vector and element.
9048 (simplify_binary_operation_1): Likewise.
9049 (simplify_const_binary_operation): Likewise.
9050 (simplify_ternary_operation): Likewise.
9052 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9053 Alan Hayward <alan.hayward@arm.com>
9054 David Sherwood <david.sherwood@arm.com>
9056 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9057 (int_mem_offset): ...this new function.
9058 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9059 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9060 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9063 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9064 Alan Hayward <alan.hayward@arm.com>
9065 David Sherwood <david.sherwood@arm.com>
9067 * lower-subreg.c (interesting_mode_p): New function.
9068 (compute_costs, find_decomposable_subregs, decompose_register)
9069 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9070 (resolve_clobber, dump_choices): Use it.
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 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9078 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9079 Alan Hayward <alan.hayward@arm.com>
9080 David Sherwood <david.sherwood@arm.com>
9082 * alias.c (find_base_value, find_base_term): Only process integer
9083 truncations. Check the precision rather than the size.
9085 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9086 Alan Hayward <alan.hayward@arm.com>
9087 David Sherwood <david.sherwood@arm.com>
9089 * machmode.h (is_narrower_int_mode): New function
9090 * optabs.c (expand_float, expand_fix): Use it.
9091 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9093 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9094 Alan Hayward <alan.hayward@arm.com>
9095 David Sherwood <david.sherwood@arm.com>
9097 * rtl.h (narrower_subreg_mode): New function.
9098 * ira-color.c (update_costs_from_allocno): Use it.
9100 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9101 Alan Hayward <alan.hayward@arm.com>
9102 David Sherwood <david.sherwood@arm.com>
9104 * optabs-query.h (convert_optab_p): New function, split out from...
9105 (convert_optab_handler): ...here.
9106 (widening_optab_handler): Delete.
9107 (find_widening_optab_handler): Remove permit_non_widening parameter.
9108 (find_widening_optab_handler_and_mode): Likewise. Provide an
9109 override that operates on mode class wrappers.
9110 * optabs-query.c (widening_optab_handler): Delete.
9111 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9112 parameter. Assert that the two modes are the same class and that
9113 the "from" mode is narrower than the "to" mode. Use
9114 convert_optab_handler instead of widening_optab_handler.
9115 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9116 instead of widening_optab_handler.
9117 * expr.c (expand_expr_real_2): Update calls to
9118 find_widening_optab_handler.
9119 * optabs.c (expand_widen_pattern_expr): Likewise.
9120 (expand_binop_directly): Take the insn_code as a parameter.
9121 (expand_binop): Only call find_widening_optab_handler for
9122 conversion optabs; use optab_handler otherwise. Update calls
9123 to find_widening_optab_handler and expand_binop_directly.
9124 Use convert_optab_handler instead of widening_optab_handler.
9125 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9126 find_widening_optab_handler and use scalar_mode rather than
9128 (convert_plusminus_to_widen): Likewise.
9130 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9131 Alan Hayward <alan.hayward@arm.com>
9132 David Sherwood <david.sherwood@arm.com>
9134 * machmode.h (fixed_size_mode): New class.
9135 * rtl.h (get_pool_mode): Return fixed_size_mode.
9136 * gengtype.c (main): Add fixed_size_mode.
9137 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9138 (get_raw_arg_mode): Likewise.
9139 * doc/tm.texi: Regenerate.
9140 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9141 * targhooks.c (default_get_reg_raw_mode): Likewise.
9142 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9143 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9144 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9145 (msp430_get_raw_result_mode): Likewise.
9146 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9147 * dbxout.c (dbxout_parms): Require fixed-size modes.
9148 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9149 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9150 * omp-low.c (lower_oacc_reductions): Likewise.
9151 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9152 (simplify_subreg): Update accordingly.
9153 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9154 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9155 that aren't fixed-size.
9156 (get_pool_mode): Return a fixed_size_mode.
9157 (output_constant_pool_2): Take a fixed_size_mode.
9159 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9160 Alan Hayward <alan.hayward@arm.com>
9161 David Sherwood <david.sherwood@arm.com>
9163 * doc/rtl.texi (vec_series): Document.
9164 (const): Say that the operand can be a vec_series.
9165 * rtl.def (VEC_SERIES): New rtx code.
9166 * rtl.h (const_vec_series_p_1): Declare.
9167 (const_vec_series_p): New function.
9168 * emit-rtl.h (gen_const_vec_series): Declare.
9169 (gen_vec_series): Likewise.
9170 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9171 (gen_vec_series): Likewise.
9172 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9173 * simplify-rtx.c (simplify_unary_operation): Handle negations
9175 (simplify_binary_operation_series): New function.
9176 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9177 (test_vector_ops_series): New function.
9178 (test_vector_ops): Call it.
9179 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9180 gen_const_vec_series.
9181 (altivec_lvsr): Likewise.
9182 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9184 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9185 Alan Hayward <alan.hayward@arm.com>
9186 David Sherwood <david.sherwood@arm.com>
9188 * doc/rtl.texi (const): Update description of address constants.
9189 Say that vector constants are allowed too.
9190 * common.md (E, F): Use CONSTANT_P instead of checking for
9192 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9193 checking for CONST_VECTOR.
9194 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9196 * expr.c (expand_expr_real_2): Check for vector modes instead
9197 of checking for CONST_VECTOR.
9198 * rtl.h (const_vec_p): New function.
9199 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9200 (unwrap_const_vec_duplicate): Handle them here too.
9202 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9203 David Malcolm <dmalcolm@redhat.com>
9204 Alan Hayward <alan.hayward@arm.com>
9205 David Sherwood <david.sherwood@arm.com>
9207 * rtl.h (vec_duplicate_p): New function.
9208 * selftest-rtl.c (assert_rtx_eq_at): New function.
9209 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9210 (assert_rtx_eq_at): Declare.
9211 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9212 * selftest-run-tests.c (selftest::run_tests): Call it.
9213 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9214 (simplify_unary_operation_1): Recursively handle vector duplicates.
9215 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9217 (simplify_subreg): Handle subregs of vector duplicates.
9218 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9219 (selftest::simplify_rtx_c_tests): New functions.
9221 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9222 Alan Hayward <alan.hayward@arm.com>
9223 David Sherwood <david.sherwood@arm.com>
9225 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9226 (gen_vec_duplicate): Likewise.
9227 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9229 (gen_const_vector): ...here.
9230 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9231 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9232 whose elements are all equal.
9233 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9234 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9235 (simplify_relational_operation): Likewise.
9236 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9238 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9239 (aarch64_expand_vector_init): Likewise.
9240 * config/arm/arm.c (neon_vdup_constant): Likewise.
9241 (neon_expand_vector_init): Likewise.
9242 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9243 (arm_block_set_unaligned_vect): Likewise.
9244 (arm_block_set_aligned_vect): Likewise.
9245 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9246 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9247 (expand_vec_perm_even_odd_pack): Likewise.
9248 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9249 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9250 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9251 gen_const_vec_duplicate.
9252 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9253 * config/mips/mips.c (mips_gen_const_int_vector): Use
9254 gen_const_vec_duplicate.
9255 (mips_expand_vector_init): Use CONST0_RTX.
9256 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9257 (define_split): Use gen_const_vec_duplicate.
9258 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9259 (define_split): Use gen_const_vec_duplicate.
9260 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9261 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9262 * config/spu/spu.c (spu_const): Likewise.
9264 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9265 Alan Hayward <alan.hayward@arm.com>
9266 David Sherwood <david.sherwood@arm.com>
9268 * combine.c (can_change_dest_mode): Reject changes in
9269 REGMODE_NATURAL_SIZE.
9271 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9273 * configure.ac (--enable-libssp): New.
9274 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9275 trying to determine target-specific default. Adjust indentation.
9276 * configure: Regenerated.
9277 * doc/install.texi (Configuration): Expand --disable-libssp
9280 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9282 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9285 2017-10-31 Martin Jambor <mjambor@suse.cz>
9288 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9290 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9292 * auto-profile.c (autofdo_source_profile::read): Use
9293 UNKNOWN_LOCATION rather than 0.
9294 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9295 (warning_at): ...this overload.
9296 (warning_at_rich_loc_n): Rename to...
9297 (warning_n): ...this overload.
9298 (error_at_rich_loc): Rename to...
9299 (error_at): ...this overload.
9300 (pedwarn_at_rich_loc): Rename to...
9301 (pedwarn): ...this overload.
9302 (permerror_at_rich_loc): Rename to...
9303 (permerror): ...this overload.
9304 (inform_at_rich_loc): Rename to...
9305 (inform): ...this overload.
9306 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9307 (diagnostic_n_impl_richloc): Rename to...
9308 (diagnostic_n_impl): ...this rich_location *-based decl.
9309 (inform_at_rich_loc): Rename to...
9310 (inform): ...this, and add an assertion.
9311 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9312 (warning_at_rich_loc): Rename to...
9313 (warning_at): ...this, and add an assertion.
9314 (warning_at_rich_loc_n): Rename to...
9315 (warning_n): ...this, and add an assertion.
9316 (warning_n): Update location_t-based implementation for removal of
9317 location_t-based diagnostic_n_impl.
9318 (pedwarn_at_rich_loc): Rename to...
9319 (pedwarn): ...this, and add an assertion.
9320 (permerror_at_rich_loc): Rename to...
9321 (permerror): ...this, and add an assertion.
9322 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9323 (error_at_rich_loc): Rename to...
9324 (error_at): ...this, and add an assertion.
9325 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9326 (driver::do_spec_on_infiles): Likewise.
9327 * substring-locations.c (format_warning_va): Update for renaming
9328 of inform_at_rich_loc.
9330 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9332 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9333 _Float<N>X built-in functions so that the variant without the
9334 "__builtin_" prefix is only enabled for the GNU C and Objective C
9335 languages when they are in non-strict ANSI/ISO mode.
9336 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9337 * target.def (floatn_builtin_p): Add a target hook to control
9338 whether _Float<N> and _Float<N>X built-in functions without the
9339 "__builtin_" prefix are enabled, and return true for C and
9340 Objective C in the default hook. Include langhooks.h in
9342 * targhooks.h (default_floatn_builtin_p): Likewise.
9343 * targhooks.c (default_floatn_builtin_p): Likewise.
9344 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9345 floatn_builtin_p target hook.
9346 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9348 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9349 Eric Botcazou <ebotcazou@adacore.com>
9351 PR rtl-optimization/81803
9352 * lra-constraints.c (curr_insn_transform): Also reload the whole
9353 register for a strict subreg no wider than a word if this is for
9354 a WORD_REGISTER_OPERATIONS target.
9356 2017-10-31 Jason Merrill <jason@redhat.com>
9358 * gdbinit.in: Skip over inlines from timevar.h.
9360 2017-10-31 Martin Liska <mliska@suse.cz>
9362 * doc/gcov.texi: Document new option.
9363 * gcov.c (print_usage): Likewise print it.
9364 (process_args): Support the argument.
9365 (format_count): New function.
9366 (format_gcov): Use the function.
9368 2017-10-31 Martin Liska <mliska@suse.cz>
9370 * gcov.c (struct name_map): do not use typedef.
9371 Define operator== and operator<.
9372 (name_search): Remove.
9373 (name_sort): Remove.
9374 (main): Do not allocate names.
9375 (process_file): Add vertical space.
9376 (generate_results): Use std::find.
9377 (release_structures): Do not release memory.
9378 (find_source): Use std::find.
9380 2017-10-31 Martin Liska <mliska@suse.cz>
9382 * gcov.c (struct line_info): Remove it's typedef.
9383 (line_info::line_info): Add proper ctor.
9384 (line_info::has_block): Do not use a typedef.
9385 (struct source_info): Do not use typedef.
9386 (circuit): Likewise.
9387 (get_cycles_count): Likewise.
9388 (output_intermediate_file): Iterate via vector iterator.
9389 (add_line_counts): Use std::vector methods.
9390 (accumulate_line_counts): Likewise.
9391 (output_lines): Likewise.
9393 2017-10-31 Martin Liska <mliska@suse.cz>
9395 * gcov.c (struct source_info): Remove typedef.
9396 (source_info::source_info): Add proper ctor.
9397 (accumulate_line_counts): Use struct, not it's typedef.
9398 (output_gcov_file): Likewise.
9399 (output_lines): Likewise.
9400 (main): Do not allocate an array.
9401 (output_intermediate_file): Use size of vector container.
9402 (process_file): Resize the vector.
9403 (generate_results): Do not preallocate, use newly added vector
9405 (release_structures): Do not release sources.
9406 (find_source): Use vector methods.
9407 (add_line_counts): Do not use typedef.
9409 2017-10-31 Martin Liska <mliska@suse.cz>
9411 * doc/gcov.texi: Document that.
9412 * gcov.c (add_line_counts): Mark lines with a non-executed
9414 (output_line_beginning): Handle such lines.
9415 (output_lines): Pass new argument.
9416 (output_intermediate_file): Print it in intermediate format.
9418 2017-10-31 Martin Liska <mliska@suse.cz>
9420 * color-macros.h: New file.
9421 * diagnostic-color.c: Factor out color related to macros to
9423 * doc/gcov.texi: Document -k option.
9424 * gcov.c (INCLUDE_STRING): Include string.h.
9425 (print_usage): Add -k option.
9426 (process_args): Parse it.
9427 (pad_count_string): New function.
9428 (output_line_beginning): Likewise.
9429 (DEFAULT_LINE_START): New macro.
9430 (output_lines): Support color output.
9432 2017-10-31 Martin Liska <mliska@suse.cz>
9434 PR gcov-profile/82633
9435 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9436 their interaction with GCOV infrastructure.
9437 * configure.ac: Add -fkeep-{inline,static}-functions to
9439 * configure: Regenerate.
9441 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9444 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9446 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9449 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9450 into a register if it does not fit into an immediate offset field.
9452 2017-10-31 Olivier Hainque <hainque@adacore.com>
9454 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9456 2017-10-31 Julia Koval <julia.koval@intel.com>
9458 * config.gcc: Add gfniintrin.h.
9459 * config/i386/gfniintrin.h: New.
9460 * config/i386/i386-builtin-types.def
9461 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9462 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9463 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9464 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9465 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9466 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9467 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9468 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9469 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9470 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9471 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9472 * config/i386/immintrin.h: Include gfniintrin.h.
9473 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9475 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9477 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9479 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9481 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9482 (arm_ashldi3_1bit): Remove pattern.
9483 (ashrdi3): Remove shift by 1 expansion.
9484 (arm_ashrdi3_1bit): Remove pattern.
9485 (lshrdi3): Remove shift by 1 expansion.
9486 (arm_lshrdi3_1bit): Remove pattern.
9487 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9488 cost of ashldi3 by 1.
9489 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9490 (<shift>di3_neon): Likewise.
9492 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9494 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9495 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9496 and (*aarch64_simd_mov<VQ:mode>).
9497 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9498 pattern alternative.
9499 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9500 attributes to match pattern alternative.
9502 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9504 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9505 * config/rs6000/emmintrin.h: New file.
9506 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9508 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9510 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9511 * (movdi_vfp_cortexa8): Remove pattern.
9513 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9515 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9517 (Specific, alpha*-dec-osf5.1): Remove.
9518 (Specific, mips-sgi-irix5): Remove.
9519 (Specific, mips-sgi-irix6): Remove.
9521 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9524 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9525 arguments to clear_bit_region_be.
9527 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9529 * gimplify.c: Include memmodel.h.
9531 2017-10-30 Martin Jambor <mjambor@suse.cz>
9533 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9534 condition whether loop should be executed at all.
9536 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9538 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9539 gimple folding of vec_madd() intrinsics.
9540 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9541 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9542 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9544 2017-10-30 Richard Biener <rguenther@suse.de>
9546 PR tree-optimization/82762
9548 2017-10-23 Richard Biener <rguenther@suse.de>
9550 PR tree-optimization/82129
9552 2017-08-01 Richard Biener <rguenther@suse.de>
9554 PR tree-optimization/81181
9555 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9556 (compute_antic): ... end of iteration here.
9558 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9560 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9561 -std=iso9899:2017 and -std=gnu17.
9562 * doc/standards.texi (C Language): Document C17 support.
9563 * doc/cpp.texi (Overview): Mention -std=c17.
9564 (Standard Predefined Macros): Document C11 and C17 values of
9565 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9566 * doc/extend.texi (Inline): Do not list individual options for
9567 standards newer than C99.
9568 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9570 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9573 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9575 * asan.c (asan_finish_file): Align asan globals array by shadow
9578 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9581 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9582 (struct store_immediate_info): Add bitregion_start and bitregion_end
9584 (store_immediate_info::store_immediate_info): Add brs and bre
9585 arguments and initialize bitregion_{start,end} from those.
9586 (struct merged_store_group): Add bitregion_start, bitregion_end,
9587 align_base and mask fields. Drop unnecessary struct keyword from
9588 struct store_immediate_info. Add do_merge method.
9589 (clear_bit_region_be): Use memset instead of loop storing zeros.
9590 (merged_store_group::do_merge): New method.
9591 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9592 stores as long as the surrounding bitregions have no gaps.
9593 (merged_store_group::merge_overlapping): Use do_merge.
9594 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9595 is byte aligned, rather than requiring that start and width are
9596 byte aligned. Drop unnecessary struct keyword from
9597 struct store_immediate_info. Allocate and populate also mask array.
9598 Make start of the arrays relative to bitregion_start rather than
9599 start and size them according to bitregion_{end,start} difference.
9600 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9601 struct store_immediate_info.
9602 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9603 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9604 struct keyword from struct store_immediate_info.
9605 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9606 between stores as long as the surrounding bitregions have no gaps.
9608 (struct split_store): Add orig non-static data member.
9609 (split_store::split_store): Initialize orig to false.
9610 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9611 if there is exactly a single original stmt. Change stmts argument
9612 to pointer from reference, if NULL, don't push anything to it. Add
9613 first argument, use it to optimize skipping over orig stmts that
9614 are known to be before bitpos already. Simplify.
9615 (split_group): Return unsigned int count how many stores are or
9616 would be needed rather than a bool. Add allow_unaligned argument.
9617 Change split_stores argument from reference to pointer, if NULL,
9618 only do a dry run computing how many stores would be produced.
9619 Rewritten algorithm to use both alignment and misalign if
9620 !allow_unaligned and handle bitfield stores with gaps.
9621 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9622 from bitregion_start instead of start. Compute allow_unaligned
9623 here, if true, do 2 split_group dry runs to compute which one
9624 produces fewer stores and prefer aligned if equal. Punt if
9625 new count is bigger or equal than original before emitting any
9626 statements, rather than during that. Remove no longer needed
9627 new_ssa_names tracking. Replace num_stmts with
9628 split_stores.length (). Use 32-bit stack allocated entries
9629 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9630 if possible. Handle bitfields with gaps.
9631 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9632 Compute bitregion_{start,end} for the stores and construct
9633 store_immediate_info with that. Formatting fixes.
9635 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9638 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9639 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9641 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9643 * gimplify.c: Include tm_p.h.
9645 * common.opt (gcoff): Re-add as ignored option.
9646 (gcoff1, gcoff2, gcoff3): Likewise.
9648 * Makefile.in (OBJS): Delete sdbout.o.
9649 (GTFILES): Delete $(srcdir)/sdbout.c.
9650 * debug.h: Delete sdb_debug_hooks.
9651 * final.c: Delete sdbout.h include.
9652 (final_scan_insn): Delete SDB_DEBUG check.
9653 (rest_of_clean_state): Likewise.
9654 * output.h: Delete sdb_begin_function_line.
9657 * toplev.c: Delete sdbout.h include.
9658 (process_options): Delete SDB_DEBUG check.
9659 * tree-core.h (tree_type_common): Delete pointer field of
9661 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9662 TYPE_SYMTAB_POINTER.
9663 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9664 (TYPE_SYMTAB_IS_POINTER): Delete.
9665 (TYPE_SYMTAB_IS_DIE): Renumber.
9666 * xcoffout.c: Refer to former sdbout.c file.
9667 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9669 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9670 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9671 (Debugging Options): Delete -gcoff.
9672 (-gstabs): Delete SDB reference.
9674 (-gcoff@var{level}): Delete.
9675 * doc/passes.texi (Debugging information output): Delete SDB and
9676 sdbout.c references.
9677 * doc/tm.texi: Regenerate.
9678 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9679 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9681 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9682 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9683 (SDB_DEBUGGING_INFO): Delete.
9684 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9685 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9686 * target.def (output_source_filename): Delete COFF reference.
9688 * common.opt (gcoff): Delete.
9689 (gxcoff+): Update Negative chain.
9690 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9692 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9693 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9694 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9696 (expand_function_start): Change sdb reference to past tense.
9697 (expand_function_end): Change sdb reference to past tense.
9698 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9699 * opts.c (debug_type_names): Delete coff entry.
9700 (common_handle_option): Delete OPT_gcoff case.
9701 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9703 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9704 * config/cris/cris.h: Delete SDB reference in comment.
9705 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9706 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9707 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9708 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9710 (ix86_expand_prologue): Likewise.
9711 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9712 * config/ia64/ia64.h: Likewise.
9713 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9714 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9716 * config/mmix/mmix.h: Likewise.
9717 * config/nds32/nds32.c: Likewise.
9718 * config/stormy/storym16.h: Likewise.
9719 * config/visium/visium.h: Likewise.
9720 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9722 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9724 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9725 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9726 FRAME_POINTER_REGNUM point at high end of local var area.
9728 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9730 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9731 Move comment around. Do not reset best_edge for a copiable
9732 destination if the copy would cause a partition change.
9733 (better_edge_p): Remove redundant check.
9735 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9737 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9739 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9741 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9742 for math functions that have _Float<N> and _Float<N>X variants.
9743 (mathfn_built_in_2): Add support for math functions that have
9744 _Float<N> and _Float<N>X variants.
9745 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9746 (expand_builtin_mathfn_ternary): Add support for fma with
9747 _Float<N> and _Float<N>X variants.
9748 (expand_builtin): Likewise.
9749 (fold_builtin_3): Likewise.
9750 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9751 create math function _Float<N> and _Float<N>X variants as external
9753 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9754 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9755 the __builtin_ prefix and if not strict ansi, without the prefix.
9756 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9757 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9758 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9759 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9760 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9761 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9762 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9763 function signatures for fma _Float<N> and _Float<N>X variants.
9764 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9765 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9766 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9767 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9768 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9769 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9770 * gencfn-macros.c (print_case_cfn): Add support for math functions
9771 that have _Float<N> and _Float<N>X variants.
9772 (print_define_operator_list): Likewise.
9773 (fltfn_suffixes): Likewise.
9775 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9776 for math functions that have _Float<N> and _Float<N>X variants.
9777 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9778 and _Float<N>X variants.
9779 (COPYSIGN): Likewise.
9782 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9783 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9785 (integer_valued_read_call_p): Likewise.
9786 * fold-const-call.c (fold_const_call_ss): Likewise.
9787 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9788 _Float<N> and _Float<N>X variants.
9789 (fold_const_call_ssss): Add support for fma _Float<N> and
9790 _Float<N>X variants.
9791 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9792 support for copysign and fma _Float<N> and _Float<N>X variants.
9793 (backprop::process_builtin_call_use): Likewise.
9794 * tree-call-cdce.c (can_test_argument_range); Add support for
9795 sqrt _Float<N> and _Float<N>X variants.
9796 (edom_only_function): Likewise.
9797 (get_no_error_domain): Likewise.
9798 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9799 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9800 copysign _Float<N> and _Float<N>X variants.
9801 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9802 handled by machine independent code.
9803 (FMAF128): Likewise.
9804 * doc/cpp.texi (Common Predefined Macros): Document defining
9805 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9806 fma _Float<N> and _Float<N>X variants.
9808 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9811 * config/i386/i386-modes.def (CCFPU): Remove definition.
9812 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9813 (ix86_cc_modes_compatible): Ditto.
9814 (ix86_expand_carry_flag_compare): Ditto.
9815 (ix86_expand_int_movcc): Ditto.
9816 (ix86_expand_int_addcc): Ditto.
9817 (ix86_reverse_condition): Ditto.
9818 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9819 Return true/false for unordered/ordered fp comparisons.
9820 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9821 (ix86_prepare_fp_compare_args): Update for rename.
9822 (ix86_expand_fp_compare): Update for rename. Generate unordered
9823 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9824 (ix86_expand_sse_compare_and_jump): Ditto.
9825 * config/i386/predicates.md (fcmov_comparison_operator):
9826 Remove CCFPU mode handling.
9827 (ix86_comparison_operator): Ditto.
9828 (ix86_carry_flag_operator): Ditto.
9829 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9830 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9831 (*cmpu<mode>_cc_i387): Ditto.
9832 (FPCMP): Remove mode iterator.
9833 (unord): Remove mode attribute.
9834 (unord_subst): New define_subst transformation
9835 (unord): New define_subst attribute.
9837 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9838 (*cmpi<unord>xf_i387): Ditto.
9839 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9840 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9841 using unord_subst transformation.
9842 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9843 (round_saeonly): Also handle CCFP mode.
9844 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9845 Remove UNSPEC_SAHF unspec handling.
9847 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9849 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9851 2017-10-27 Jeff Law <law@redhat.com>
9853 * gimple-ssa-sprintf.c: Include domwalk.h.
9854 (class sprintf_dom_walker): New class, derived from dom_walker.
9855 (sprintf_dom_walker::before_dom_children): New function.
9856 (struct call_info): Moved into sprintf_dom_walker class
9857 (compute_formath_length, handle_gimple_call): Likewise.
9858 (sprintf_length::execute): Call the dominator walker rather
9859 than walking the statements.
9861 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9862 gimple statement locations.
9863 (check_array_bounds): Corresponding changes. Get the statement's
9864 location directly from wi->stmt.
9866 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9869 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9871 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9873 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9874 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9876 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9879 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9880 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9881 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9882 maybe_get_pool_constant.
9883 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9886 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9888 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9889 2.26 caveat. Update gas and gld versions.
9890 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9893 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9895 * cgraph.h (set_malloc_flag): Declare.
9896 * cgraph.c (set_malloc_flag_1): New function.
9897 (set_malloc_flag): Likewise.
9898 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9899 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9901 (read_ipa_call_summary): Add support for reading is_return_callee.
9902 (write_ipa_call_summary): Stream is_return_callee.
9903 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9904 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9905 ipa-prop.h, ipa-fnsummary.h.
9906 (pure_const_names): Change to static.
9907 (malloc_state_e): Define.
9908 (malloc_state_names): Define.
9909 (funct_state_d): Add field malloc_state.
9910 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9911 (check_retval_uses): New function.
9912 (malloc_candidate_p): Likewise.
9913 (analyze_function): Add support for malloc attribute.
9914 (pure_const_write_summary): Stream malloc_state.
9915 (pure_const_read_summary): Add support for reading malloc_state.
9916 (dump_malloc_lattice): New function.
9917 (propagate_malloc): New function.
9918 (warn_function_malloc): New function.
9919 (ipa_pure_const::execute): Call propagate_malloc and
9920 ipa_free_fn_summary.
9921 (pass_local_pure_const::execute): Add support for malloc attribute.
9922 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9923 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9925 2017-10-27 Martin Liska <mliska@suse.cz>
9927 PR gcov-profile/82457
9928 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9929 for fork-like functions to be properly instrumented.
9931 2017-10-27 Richard Biener <rguenther@suse.de>
9934 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9935 info when we redirected EH.
9937 2017-10-26 Michael Collison <michael.collison@arm.com>
9939 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9941 (<optab>_trunchf<GPI:mode>2: New pattern.
9942 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9943 * config/aarch64/iterators.md (wv): New mode attribute.
9944 (vf, VF): New mode attributes.
9945 (vgp, VGP): New mode attributes.
9946 (s): Update attribute with SImode and DImode prefixes.
9948 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9950 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9951 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9952 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9953 (nios2_option_overide): Initialize it. Don't allow R0-relative
9954 addressing with PIC.
9955 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9956 (nios2_symbolic_constant_p): Likewise.
9957 (nios2_legitimate_address_p): Likewise.
9958 (nios2_r0rel_section_name_p): New.
9959 (nios2_symbol_ref_in_r0rel_data_p): New.
9960 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9961 (r0rel_constant_p): New.
9962 (nios2_print_operand_address): Handle r0rel_constant_p.
9963 (nios2_cdx_narrow_form_p): Likewise.
9964 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9965 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9966 (Nios II Options): Document -mr0rel-sec.
9968 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9970 * config/nios2/nios2.c: Include xregex.h.
9971 (nios2_gprel_sec_regex): New.
9972 (nios2_option_overide): Initialize it. Don't allow GP-relative
9973 addressing with PIC.
9974 (nios2_small_section_name_p): Check for regex match.
9975 * config/nios2/nios2.opt (mgprel-sec=): New option.
9976 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9977 (Nios II Options): Document -mgprel-sec.
9979 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9981 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9983 2017-10-26 Tom de Vries <tom@codesourcery.com>
9985 PR tree-optimization/82707
9986 * gimple.c (gimple_copy): Fix unsharing of
9987 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9989 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9991 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9992 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9993 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9994 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9995 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9996 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9997 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9998 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9999 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10000 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10001 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10002 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10003 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10004 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10005 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10006 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10007 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10009 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10011 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10013 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10014 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10015 warning to rs6000.c. Remove the Undocumented flag, since it has
10017 (-mabi=ibmlongdouble): Likewise.
10018 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10019 already set, set the default format for long double.
10020 (rs6000_debug_reg_global): Print whether long double is IBM or
10022 (rs6000_option_override_internal): Rework setting long double
10023 format. Only warn if the user is changing the long double default
10024 and they did not use -Wno-psabi.
10025 * doc/invoke.texi (PowerPC options): Update the documentation for
10026 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10028 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10029 Alan Hayward <alan.hayward@arm.com>
10030 David Sherwood <david.sherwood@arm.com>
10032 * rtl.h (wider_subreg_mode): New function.
10033 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10034 rather than an unsigned int *.
10035 * ira-color.c (regno_max_ref_width): Replace with...
10036 (regno_max_ref_mode): ...this new variable.
10037 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10038 Use wider_subreg_mode.
10039 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10040 rather than an unsigned int *.
10041 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10042 (process_alt_operands): Likewise.
10043 (invariant_p): Likewise.
10044 * lra-spills.c (assign_mem_slot): Likewise.
10045 (add_pseudo_to_slot): Likewise.
10046 * lra.c (collect_non_operand_hard_regs): Likewise.
10047 (add_regs_to_insn_regno_info): Likewise.
10048 * reload1.c (regno_max_ref_width): Replace with...
10049 (regno_max_ref_mode): ...this new variable.
10050 (reload): Update accordingly. Update call to
10051 ira_sort_regnos_for_alter_reg.
10052 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10053 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10054 (scan_paradoxical_subregs): Likewise.
10056 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10058 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10059 (aarch64_frame): Add emit_frame_chain boolean.
10060 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10061 Move eh_return case to aarch64_layout_frame.
10062 (aarch64_layout_frame): Initialize emit_frame_chain.
10063 (aarch64_expand_prologue): Use emit_frame_chain.
10065 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10067 * config/aarch64/aarch64.c (aarch64_layout_frame):
10068 Ensure LR is always stored at the bottom of the callee-saves.
10069 Remove rarely used frame layout which saves callee-saves at top of
10070 frame, so the store of LR can be used as a valid probe in all cases.
10072 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10074 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10075 Improve unaligned TImode/TFmode base/offset split.
10077 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10078 Alan Hayward <alan.hayward@arm.com>
10079 David Sherwood <david.sherwood@arm.com>
10081 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10082 * combine.c (find_single_use_1): Likewise.
10083 (expand_field_assignment): Likewise.
10084 (move_deaths): Likewise.
10085 * lra-constraints.c (simplify_operand_subreg): Likewise.
10086 (curr_insn_transform): Likewise.
10087 * lra.c (collect_non_operand_hard_regs): Likewise.
10088 (add_regs_to_insn_regno_info): Likewise.
10089 * rtlanal.c (reg_referenced_p): Likewise.
10090 (covers_regno_no_parallel_p): Likewise.
10092 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10094 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10095 Don't print any bits outside the precision of the value.
10096 * wide-int.cc (test_printing): Add some new tests.
10098 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10100 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10101 supports -xbrace_comment option.
10102 * configure: Regenerate.
10103 * config.in: Regenerate.
10104 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10105 (ASM_CPU_SPEC): Use it.
10107 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10109 * target.def (static_rtx_alignment): New hook.
10110 * targhooks.h (default_static_rtx_alignment): Declare.
10111 * targhooks.c (default_static_rtx_alignment): New function.
10112 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10113 * doc/tm.texi: Regenerate.
10114 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10115 instead of targetm.constant_alignment. Remove call to
10116 set_mem_attributes.
10117 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10118 (cris_preferred_mininum_alignment): New function, split out from...
10119 (cris_constant_alignment): ...here.
10120 (cris_static_rtx_alignment): New function.
10121 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10123 (ix86_constant_alignment): ...here.
10124 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10125 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10126 (mmix_static_rtx_alignment): New function.
10127 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10128 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10130 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10133 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10134 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10136 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10138 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10139 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10140 mark as a sign-extending load.
10141 (local_pic_load_u): Define.
10143 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10145 PR middle-end/82062
10146 * fold-const.c (operand_equal_for_comparison_p): Also return true
10147 if ARG0 is a simple variant of ARG1 with narrower precision.
10148 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10150 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10152 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10154 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10155 scatter_static, scatter_per_elt.
10156 * x86-tune-costs.h: Add new cost entries.
10158 2017-10-25 Richard Biener <rguenther@suse.de>
10160 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10161 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10162 class pass_fre): Move to ...
10163 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10164 class pass_fre): ... here and adjust for statistics.
10166 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10169 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10170 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10171 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10172 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10175 2017-10-25 Richard Biener <rguenther@suse.de>
10177 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10178 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10179 eliminate_push_avail, eliminate_insert): Move inside...
10180 (class eliminate_dom_walker): ... this class in preparation
10182 (fini_eliminate): Remove by merging with ...
10183 (eliminate): ... this function. Adjust for class changes.
10184 (pass_pre::execute): Remove fini_eliminate call.
10185 (pass_fre::execute): Likewise.
10187 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10190 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10191 (VPERMI2, VPERMI2I): New mode iterators.
10192 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10193 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10195 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10196 mode iterator. Remove 3 old define_insn patterns.
10197 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10198 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10199 VPERMI2 mode iterator, remove the other two expanders.
10200 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10201 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10202 instructions, remove the other two patterns.
10203 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10204 mode iterator, remove the other two patterns.
10205 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10206 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10207 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10208 and adjust argument order accordingly.
10209 (ix86_expand_vec_perm): Adjust caller.
10210 (expand_vec_perm_1): Likewise.
10211 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10212 (expand_vec_perm_vpermt2_vpshub2): ... this.
10213 (ix86_expand_vec_perm_const_1): Adjust caller.
10214 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10217 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10218 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10219 (vec_shl_<mode>): Remove unused expander.
10220 (avx512bw_<shift_insn><mode>3): New define_insn.
10221 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10222 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10224 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10227 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10230 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10232 PR rtl-optimization/82396
10233 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10234 (autopref_multipass_init): Simplify initialization.
10235 (autopref_rank_data): Simplify sort order.
10236 * gcc/sched-int.h (autopref_multipass_data_): Remove
10237 multi_mem_insn_p, min_offset and max_offset.
10239 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10241 PR middle-end/60580
10242 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10243 Check special value of flag_omit_frame_pointer.
10244 (aarch64_can_eliminate): Likewise.
10245 (aarch64_override_options_after_change_1): Simplify handling of
10246 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10248 2017-10-24 Richard Biener <rguenther@suse.de>
10250 PR tree-optimization/82697
10251 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10252 zero for conditional load and unconditional store.
10254 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10256 * doc/install.texi: Document bootstrap-cet.
10258 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10261 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10262 ENDBR instruction at function entrance if function is only
10265 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10268 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10269 patterns to better describe from which operation the CF is computed.
10270 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10271 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10272 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10273 is 0, use _0 suffixed expanders instead of emitting a comparison
10276 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10278 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10279 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10281 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10283 PR middle-end/82569
10284 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10285 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10286 * loop-iv.c (iv_get_reaching_def): Likewise.
10287 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10288 variable is promoted and the partition contains undefined values.
10290 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10292 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10294 (nios2_address_cost): Define.
10295 (nios2_legitimize_address): Recognize (exp + constant) directly.
10296 (TARGET_ADDRESS_COST): Define.
10298 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10300 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10301 (nios2_symbolic_memory_operand_p): Declare.
10302 (nios2_split_large_constant): Declare.
10303 (nios2_split_symbolic_memory_operand): Declare.
10304 * config/nios2/nios2.c: Adjust includes.
10305 (nios2_symbolic_constant_allowed): New.
10306 (nios2_symbolic_constant_p): New.
10307 (nios2_plus_symbolic_constant_p): New.
10308 (nios2_valid_addr_expr_p): Recognize addresses involving
10309 symbolic constants.
10310 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10311 (nios2_symbolic_memory_operand_p): New.
10312 (nios2_large_constant_p): New.
10313 (nios2_split_large_constant): New.
10314 (nios2_split_plus_large_constant): New.
10315 (nios2_split_symbolic_memory_operand): New.
10316 (nios2_legitimize_address): Code refactoring. Handle addresses
10317 involving symbolic constants.
10318 (nios2_emit_move_sequence): Likewise.
10319 (nios2_print_operand): Improve error output.
10320 (nios2_print_operand_address): Handle LO_SUM.
10321 (nios2_cdx_narrow_form_p): Likewise.
10322 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10323 operands involving symbolic constants.
10324 (movhi_internal, movsi_internal): Likewise.
10325 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10326 (extendhisi2, extendqi<mode>2): Likewise.
10328 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10330 * tree-pass.h (PROP_rtl_split_insns): Define.
10331 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10333 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10335 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10337 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10340 * target.def (const_not_ok_for_debug_p): Default to
10341 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10342 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10343 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10344 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10345 which targetm.const_not_ok_for_debug_p returned true.
10346 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10348 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10350 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10351 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10352 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10354 (ix86_const_not_ok_for_debug_p): New function.
10355 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10356 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10358 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10361 * system.h: Conditionally include "unique-ptr.h" if
10362 INCLUDE_UNIQUE_PTR is defined.
10363 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10364 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10366 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10368 * config/rl78/rl78.md: New define_expand "subdi3".
10370 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10373 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10374 DF_REF_INSN if DF_REF_INSN_INFO is false.
10376 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10378 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10379 xmm_move instead of sse_move.
10380 (sse_store_index): New function.
10381 (ix86_register_move_cost): Be more sensible about mismatch stall;
10382 model AVX moves correctly; make difference between sse->integer and
10384 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10385 moves; make difference between SSE and AVX.
10386 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10387 and zmm_move. Increase size of sse load and store tables;
10388 add unaligned load and store tables; add ssemmx_to_integer.
10389 * x86-tune-costs.h: Update all entries according to real
10390 move latencies from Agner Fog's manual and chip documentation.
10392 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10395 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10396 * config/i386/constraints.md (Wf): New constraint.
10397 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10398 (cmp<dwi>_doubleword): Removed.
10399 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10400 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10401 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10402 expand with cmp<dwi>_doubleword. For LTU and GEU use
10403 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10405 * common.opt (gcolumn-info): Enable by default.
10406 * doc/invoke.texi (gcolumn-info): Document new default.
10408 2017-10-23 Richard Biener <rguenther@suse.de>
10410 PR tree-optimization/82672
10411 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10412 Fold the stmt if we propagated into it.
10414 2017-10-23 Richard Biener <rguenther@suse.de>
10416 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10417 (bitmap_remove_expr_from_set): ... this. All callers call this
10418 for non-constant values.
10419 (bitmap_set_subtract): Rename to...
10420 (bitmap_set_subtract_expressions): ... this. Adjust and
10422 (bitmap_set_contains_value): Remove superfluous check.
10423 (bitmap_set_replace_value): Inline into single caller ...
10424 (bitmap_value_replace_in_set): ... here and simplify.
10425 (dependent_clean): Merge into ...
10426 (clean): ... this using an overload. Adjust.
10427 (prune_clobbered_mems): Adjust.
10428 (compute_antic_aux): Likewise.
10429 (compute_partial_antic_aux): Likewise.
10431 2017-10-23 Richard Biener <rguenther@suse.de>
10433 PR tree-optimization/82129
10435 2017-08-01 Richard Biener <rguenther@suse.de>
10437 PR tree-optimization/81181
10438 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10439 (compute_antic): ... end of iteration here.
10441 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10443 * target.def (starting_frame_offset): New hook.
10444 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10445 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10446 * doc/tm.texi.in: Regenerate.
10447 * hooks.h (hook_hwi_void_0): Declare.
10448 * hooks.c (hook_hwi_void_0): New function.
10449 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10450 STARTING_FRAME_OFFSET.
10451 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10452 * reload1.c (reload): Likewise.
10453 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10454 instead of STARTING_FRAME_OFFSET.
10455 * function.c (try_fit_stack_local): Likewise.
10456 (assign_stack_local_1): Likewise
10457 (instantiate_virtual_regs): Likewise.
10458 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10459 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10460 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10461 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10462 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10463 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10464 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10465 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10466 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10467 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10468 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10469 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10470 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10471 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10472 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10473 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10474 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10475 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10476 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10477 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10478 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10479 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10480 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10481 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10482 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10483 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10484 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10485 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10486 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10487 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10488 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10489 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10490 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10491 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10492 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10493 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10494 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10495 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10496 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10497 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10498 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10499 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10500 return a HOST_WIDE_INT.
10501 (avr_builtin_setjmp_frame_value): Use it instead of
10502 STARTING_FRAME_OFFSET.
10503 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10504 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10505 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10507 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10508 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10509 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10510 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10511 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10512 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10513 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10514 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10515 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10516 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10517 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10518 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10520 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10521 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10522 * config/mips/mips.c (mips_compute_frame_info): Refer to
10523 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10524 (mips_starting_frame_offset): New function.
10525 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10526 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10527 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10528 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10529 and return a HOST_WIDE_INT.
10530 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10531 (mmix_initial_elimination_offset): Refer to
10532 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10533 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10534 * config/pa/pa.c (pa_starting_frame_offset): New function.
10535 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10536 (pa_expand_prologue): Likewise.
10537 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10538 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10539 !FRAME_GROWS_DOWNWARD handling to...
10540 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10541 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10542 !FRAME_GROWS_DOWNWARD handling to...
10543 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10544 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10545 !FRAME_GROWS_DOWNWARD handling to...
10546 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10547 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10549 (rs6000_starting_frame_offset): New function.
10550 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10551 !FRAME_GROWS_DOWNWARD handling to...
10552 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10553 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10554 !FRAME_GROWS_DOWNWARD handling to...
10555 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10556 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10557 !FRAME_GROWS_DOWNWARD handling to...
10558 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10559 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10560 (rs6000_starting_frame_offset): New function.
10561 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10562 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10563 * config/vax/vax.c (vax_starting_frame_offset): New function.
10564 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10565 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10566 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10567 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10568 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10569 * system.h (STARTING_FRAME_OFFSET): Poison.
10571 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10573 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10574 SCALAR_TYPE_MODE instead of TYPE_MODE.
10576 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10577 Alan Hayward <alan.hayward@arm.com>
10578 David Sherwood <david.sherwood@arm.com>
10580 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10582 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10583 Alan Hayward <alan.hayward@arm.com>
10584 David Sherwood <david.sherwood@arm.com>
10586 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10588 2017-10-23 Richard Biener <rguenther@suse.de>
10590 PR tree-optimization/82129
10591 * tree-ssa-pre.c (bitmap_set_and): Remove.
10592 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10593 canonicalizing expressions in the set to those with lowest
10594 ID rather than taking that from the first edge.
10596 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10598 * combine.c (rtx_equal_for_field_assignment_p): Use
10599 byte_lowpart_offset.
10601 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10602 Alan Hayward <alan.hayward@arm.com>
10603 David Sherwood <david.sherwood@arm.com>
10605 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10606 to a SUBREG_PROMOTED_VAR.
10608 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10609 Alan Hayward <alan.hayward@arm.com>
10610 David Sherwood <david.sherwood@arm.com>
10612 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10613 (expand_debug_source_expr): Likewise.
10614 * combine.c (combine_simplify_rtx): Likewise.
10615 * cse.c (fold_rtx): Likewise.
10616 * optabs.c (expand_float): Likewise.
10617 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10618 (simplify_binary_operation_1): Likewise.
10620 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10621 Alan Hayward <alan.hayward@arm.com>
10622 David Sherwood <david.sherwood@arm.com>
10624 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10625 (record_promoted_value): Likewise.
10626 * expr.c (expand_expr_real_2): Likewise.
10627 * ree.c (update_reg_equal_equiv_notes): Likewise.
10628 (combine_set_extension): Likewise.
10629 * rtlanal.c (low_bitmask_len): Likewise.
10630 * simplify-rtx.c (neg_const_int): Likewise.
10631 (simplify_binary_operation_1): Likewise.
10633 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10634 Alan Hayward <alan.hayward@arm.com>
10635 David Sherwood <david.sherwood@arm.com>
10637 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10638 * regcprop.c (maybe_mode_change): Likewise.
10639 * reload1.c (alter_reg): Likewise.
10641 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10643 * inchash.h (inchash::hash::add_wide_int): New function.
10644 * lto-streamer-out.c (hash_tree): Use it.
10646 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10648 * inchash.h (inchash::hash::add_wide_int): Rename to...
10649 (inchash::hash::add_hwi): ...this.
10650 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10651 (polymorphic_call_target_hasher::hash): Likewise.
10652 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10653 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10654 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10655 * lto-streamer-out.c (hash_tree): Likewise.
10656 * optc-save-gen.awk: Likewise.
10657 * tree.c (add_expr): Likewise.
10659 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10662 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10663 for ordered inequality comparisons even with TARGET_IEEE_FP.
10665 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10668 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10669 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10670 Expand with cmp<dwi>_doubleword.
10672 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10674 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10675 List CET intrinsics.
10676 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10677 specific to -fcf-protection option.
10679 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10681 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10682 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10683 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10684 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10685 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10686 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10687 (extra_objs): Add cet.o for Linux/x86 targets.
10688 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10689 * config/i386/cet.c: New file.
10690 * config/i386/cetintrin.h: Likewise.
10691 * config/i386/t-cet: Likewise.
10692 * config/i386/cpuid.h (bit_SHSTK): New.
10693 (bit_IBT): Likewise.
10694 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10695 pass IBT and SHSTK bits.
10696 * config/i386/i386-builtin-types.def
10697 (VOID_FTYPE_UNSIGNED_PVOID): New.
10698 (VOID_FTYPE_UINT64_PVOID): Likewise.
10699 * config/i386/i386-builtin.def: Add CET intrinsics.
10700 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10701 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10702 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10703 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10705 * config/i386/i386.c (rest_of_insert_endbranch): New.
10706 (pass_data_insert_endbranch): Likewise.
10707 (pass_insert_endbranch): Likewise.
10708 (make_pass_insert_endbranch): Likewise.
10709 (ix86_notrack_prefixed_insn_p): Likewise.
10710 (ix86_target_string): Add -mibt, -mshstk flags.
10711 (ix86_option_override_internal): Add flag_cf_protection
10713 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10714 (ix86_print_operand): Add 'notrack' prefix output.
10715 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10716 (ix86_expand_builtin): Expand CET intrinsics.
10717 (x86_output_mi_thunk): Add 'endbranch' instruction.
10718 * config/i386/i386.h (TARGET_IBT): New.
10719 (TARGET_IBT_P): Likewise.
10720 (TARGET_SHSTK): Likewise.
10721 (TARGET_SHSTK_P): Likewise.
10722 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10723 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10724 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10725 (builtin_setjmp_setup): New pattern.
10726 (builtin_longjmp): Likewise.
10727 (rdssp<mode>): Likewise.
10728 (incssp<mode>): Likewise.
10729 (saveprevssp): Likewise.
10730 (rstorssp): Likewise.
10731 (wrss<mode>): Likewise.
10732 (wruss<mode>): Likewise.
10733 (setssbsy): Likewise.
10734 (clrssbsy): Likewise.
10735 (nop_endbr): Likewise.
10736 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10738 * config/i386/immintrin.h: Include <cetintrin.h>.
10739 * config/i386/linux-common.h
10740 (file_end_indicate_exec_stack_and_cet): New prototype.
10741 (TARGET_ASM_FILE_END): New.
10743 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10745 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10746 latencies instead of having separate table; make difference between
10747 integer and float costs.
10748 * i386.h (processor_costs): Remove scalar_stmt_cost,
10749 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10750 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10752 * x86-tune-costs.h: Remove entries which has been removed in
10753 procesor_costs from all tables; make cond_taken_branch_cost
10754 and cond_not_taken_branch_cost COST_N_INSNS based.
10756 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10758 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10760 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10762 * config/i386/i386.md (isa): Remove fma_avx512f.
10763 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10764 <avx512>_fmadd_<mode>_mask3<round_name>,
10765 <avx512>_fmsub_<mode>_mask<round_name>,
10766 <avx512>_fmsub_<mode>_mask3<round_name>,
10767 <avx512>_fnmadd_<mode>_mask<round_name>,
10768 <avx512>_fnmadd_<mode>_mask3<round_name>,
10769 <avx512>_fnmsub_<mode>_mask<round_name>,
10770 <avx512>_fnmsub_<mode>_mask3<round_name>,
10771 <avx512>_fmaddsub_<mode>_mask<round_name>,
10772 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10773 <avx512>_fmsubadd_<mode>_mask<round_name>,
10774 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10775 (*vec_widen_umult_even_v16si<mask_name>,
10776 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10777 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10779 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10781 * extend.texi: Add 'nocf_check' documentation.
10782 * gimple.texi: Add second parameter to
10783 gimple_build_call_from_tree.
10784 * invoke.texi: Add -fcf-protection documentation.
10785 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10787 2017-10-20 Richard Biener <rguenther@suse.de>
10789 PR tree-optimization/82473
10790 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10791 the largest input type.
10793 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10795 * c-attribs.c (handle_nocf_check_attribute): New function.
10796 (c_common_attribute_table): Add 'nocf_check' handling.
10797 * gimple-parser.c: Add second argument NULL to
10798 gimple_build_call_from_tree.
10799 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10800 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10802 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10803 * common.opt: Add fcf-protection flag.
10804 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10805 * flag-types.h: Add enum cf_protection_level.
10806 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10807 Add 'nocf_check' attribute propagation to gimple call.
10808 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10809 (gimple_build_call_from_tree): Update prototype.
10810 (gimple_call_nocf_check_p): New function.
10811 (gimple_call_set_nocf_check): Likewise.
10812 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10813 * ipa-icf.c: Add nocf_check attribute in statement hash.
10814 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10815 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10816 * toplev.c (process_options): Add flag_cf_protection handling.
10818 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10820 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10822 2017-10-20 Richard Biener <rguenther@suse.de>
10824 PR tree-optimization/82603
10825 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10826 remove false predicated stores.
10828 2017-10-20 Richard Biener <rguenther@suse.de>
10830 * graphite-isl-ast-to-gimple.c
10831 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10832 Remove return value and simplify, dump copied stmt after lhs
10834 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10835 Reduce dump verbosity.
10836 (gsi_insert_earliest): Likewise.
10837 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10838 * graphite.c (print_global_statistics): Adjust dumping.
10839 (print_graphite_scop_statistics): Likewise.
10840 (print_graphite_statistics): Do not dump loops here.
10841 (graphite_transform_loops): But here.
10843 2017-10-20 Nicolas Roche <roche@adacore.com>
10845 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10846 * configure: Regenerate.
10848 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10851 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10852 functions when optimizing replace GIMPLE_RETURN stmts with
10853 calls to __builtin_unreachable ().
10856 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10857 for -fsanitize=thread link of executables.
10858 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10859 link of executables.
10862 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10863 New mode iterators.
10864 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10865 define_insns for logical vector shifts to use VI248_AVX512BW
10866 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10867 condition, useless isa and prefix attributes. Change the first
10868 2 of these define_insns to ...
10869 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10870 define_insn for avx512vl.
10871 (<shift_insn><mode>3): ... and this, new define_insn without
10872 masking for non-avx512vl.
10875 * config/i386/sse.md (*andnot<mode>3,
10876 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10877 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10878 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10879 not applied use empty suffix even for TARGET_AVX512VL.
10880 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10881 is applied, supply evex,evex or evex,evex,evex instead of just
10884 2017-10-20 Julia Koval <julia.koval@intel.com>
10886 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10887 (OPTION_MASK_ISA_GFNI_UNSET): New.
10888 (ix86_handle_option): Handle OPT_mgfni.
10889 * config/i386/cpuid.h (bit_GFNI): New.
10890 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10891 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10892 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10893 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10894 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10895 * config/i386/i386.opt: Add mgfni.
10897 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10899 * config/msp430/msp430.c (msp430_option_override): Disable
10900 -fdelete-null-pointer-checks.
10901 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10903 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10905 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10906 of x87 and SSE instructions.
10908 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10910 * asan.c (create_cond_insert_point): Do not update edge count.
10911 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10912 (afdo_propagate_circuit): Likewise.
10913 (afdo_calculate_branch_prob): Likewise.
10914 (afdo_annotate_cfg): Likewise.
10915 * basic-block.h (struct edge_def): Remove count.
10916 (edge_def::count): New accessor.
10917 * bb-reorder.c (rotate_loop): Update.
10918 (find_traces_1_round): Update.
10919 (connect_traces): Update.
10920 (sanitize_hot_paths): Update.
10921 * cfg.c (unchecked_make_edge): Update.
10922 (make_single_succ_edge): Update.
10923 (check_bb_profile): Update.
10924 (dump_edge_info): Update.
10925 (update_bb_profile_for_threading): Update.
10926 (scale_bbs_frequencies_int): Update.
10927 (scale_bbs_frequencies_gcov_type): Update.
10928 (scale_bbs_frequencies_profile_count): Update.
10929 (scale_bbs_frequencies): Update.
10930 * cfganal.c (connect_infinite_loops_to_exit): Update.
10931 * cfgbuild.c (compute_outgoing_frequencies): Update.
10932 (find_many_sub_basic_blocks): Update.
10933 * cfgcleanup.c (try_forward_edges): Update.
10934 (try_crossjump_to_edge): Update
10935 * cfgexpand.c (expand_gimple_cond): Update
10936 (expand_gimple_tailcall): Update
10937 (construct_exit_block): Update
10938 * cfghooks.c (verify_flow_info): Update
10939 (redirect_edge_succ_nodup): Update
10940 (split_edge): Update
10941 (make_forwarder_block): Update
10942 (duplicate_block): Update
10943 (account_profile_record): Update
10944 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10945 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10946 * cfgloopmanip.c (scale_loop_profile): Update.
10948 (lv_adjust_loop_entry_edge): Update.
10949 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10950 (force_nonfallthru_and_redirect): Update.
10951 (purge_dead_edges): Update.
10952 (rtl_flow_call_edges_add): Update.
10953 * cgraphunit.c (init_lowered_empty_function): Update.
10954 (cgraph_node::expand_thunk): Update.
10955 * gimple-pretty-print.c (dump_probability): Update.
10956 (dump_edge_probability): Update.
10957 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10958 * haifa-sched.c (sched_create_recovery_edges): Update.
10959 * hsa-gen.c (convert_switch_statements): Update.
10960 * ifcvt.c (dead_or_predicable): Update.
10961 * ipa-inline-transform.c (inline_transform): Update.
10962 * ipa-split.c (split_function): Update.
10963 * ipa-utils.c (ipa_merge_profiles): Update.
10964 * loop-doloop.c (add_test): Update.
10965 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10966 * lto-streamer-in.c (input_cfg): Update.
10967 (input_function): Update.
10968 * lto-streamer-out.c (output_cfg): Update.
10969 * modulo-sched.c (sms_schedule): Update.
10970 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10971 * predict.c (maybe_hot_edge_p): Update.
10972 (unlikely_executed_edge_p): Update.
10973 (probably_never_executed_edge_p): Update.
10974 (dump_prediction): Update.
10975 (drop_profile): Update.
10976 (propagate_unlikely_bbs_forward): Update.
10977 (determine_unlikely_bbs): Update.
10978 (force_edge_cold): Update.
10979 * profile.c (compute_branch_probabilities): Update.
10980 * reg-stack.c (better_edge): Update.
10981 * shrink-wrap.c (handle_simple_exit): Update.
10982 * tracer.c (better_p): Update.
10983 * trans-mem.c (expand_transaction): Update.
10984 (split_bb_make_tm_edge): Update.
10985 * tree-call-cdce.c: Update.
10986 * tree-cfg.c (gimple_find_sub_bbs): Update.
10987 (gimple_split_edge): Update.
10988 (gimple_duplicate_sese_region): Update.
10989 (gimple_duplicate_sese_tail): Update.
10990 (gimple_flow_call_edges_add): Update.
10991 (insert_cond_bb): Update.
10992 (execute_fixup_cfg): Update.
10993 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10994 * tree-complex.c (expand_complex_div_wide): Update.
10995 * tree-eh.c (lower_resx): Update.
10996 (unsplit_eh): Update.
10997 (cleanup_empty_eh_move_lp): Update.
10998 * tree-inline.c (copy_edges_for_bb): Update.
10999 (freqs_to_counts): Update.
11000 (copy_cfg_body): Update.
11001 * tree-ssa-dce.c (remove_dead_stmt): Update.
11002 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11003 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11004 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11005 (unloop_loops): Update.
11006 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11007 * tree-ssa-loop-split.c (connect_loops): Update.
11008 (split_loop): Update.
11009 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11010 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11011 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11012 * tree-ssa-reassoc.c (branch_fixup): Update.
11013 * tree-ssa-tail-merge.c (replace_block_by): Update.
11014 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11015 (compute_path_counts): Update.
11016 (update_profile): Update.
11017 (recompute_probabilities): Update.
11018 (update_joiner_offpath_counts): Update.
11019 (estimated_freqs_path): Update.
11020 (freqs_to_counts_path): Update.
11021 (clear_counts_path): Update.
11022 (ssa_fix_duplicate_block_edges): Update.
11023 (duplicate_thread_path): Update.
11024 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11025 (case_bit_test_cmp): Update.
11026 (collect_switch_conv_info): Update.
11027 (gen_inbound_check): Update.
11028 (do_jump_if_equal): Update.
11029 (emit_cmp_and_jump_insns): Update.
11030 * tree-tailcall.c (decrease_profile): Update.
11031 (eliminate_tail_call): Update.
11032 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11033 (vect_do_peeling): Update.
11034 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11035 * ubsan.c (ubsan_expand_null_ifn): Update.
11036 (ubsan_expand_ptr_ifn): Update.
11037 * value-prof.c (gimple_divmod_fixed_value): Update.
11038 (gimple_mod_pow2): Update.
11039 (gimple_mod_subtract): Update.
11040 (gimple_ic): Update.
11041 (gimple_stringop_fixed_value): Update.
11043 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11046 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11048 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11050 PR rtl-optimization/82395
11051 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11052 based on non_spilled_static_chain_regno_p.
11054 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11056 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11057 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11058 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11060 2017-10-19 Martin Sebor <msebor@redhat.com>
11062 PR tree-optimization/82596
11063 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11065 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11067 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11068 (get_mem_refs_of_builtin_call): Likewise.
11069 * builtins.c (expand_builtin_apply): Adjust call to
11070 allocate_dynamic_stack_space.
11071 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11072 the third argument to allocate_dynamic_stack_space, otherwise -1.
11073 (expand_builtin): Deal with all alloca variants.
11074 (is_inexpensive_builtin): Likewise.
11075 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11076 * calls.c (special_function_p): Deal with all alloca variants.
11077 (initialize_argument_information): Adjust call to
11078 allocate_dynamic_stack_space.
11079 (expand_call): Likewise.
11080 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11081 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11082 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11083 use it for the stack usage computation.
11084 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11085 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11086 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11087 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11088 (in_loop_p): Remove first argument and useless check.
11089 (pass_walloca::execute): Remove useless test and adjust call to above.
11090 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11091 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11092 (gimplify_call_expr): Deal with all alloca variants.
11093 * hsa-gen.c (gen_hsa_alloca): Likewise.
11094 (gen_hsa_insns_for_call): Likewise.
11095 * ipa-pure-const.c (special_builtin_state): Likewise.
11096 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11097 * tree-object-size.c (alloc_object_size): Likewise.
11098 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11099 (call_may_clobber_ref_p_1): Likewise.
11100 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11101 (ccp_fold_stmt): Likewise.
11102 (optimize_stack_restore): Likewise.
11103 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11104 (mark_all_reaching_defs_necessary_1): Likewise.
11105 (propagate_necessity): Likewise.
11106 (eliminate_unnecessary_stmts): Likewise.
11107 * tree.c (build_common_builtin_nodes): Build
11108 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11109 (build_alloca_call_expr): New function.
11110 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11111 (CASE_BUILT_IN_ALLOCA): Likewise.
11112 (build_alloca_call_expr): Declare.
11113 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11115 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11118 * dwarf2out.c (new_die_raw): New static inline function.
11119 (new_die): Use it to create the DIE.
11120 (add_AT_external_die_ref): Likewise.
11121 (clone_die): Likewise.
11122 (clone_as_declaration): Likewise.
11123 (dwarf2out_vms_debug_main_pointer): Likewise.
11124 (base_type_die): Likewise. Remove early return for corner cases.
11125 Do not call add_pubtype on the DIE here.
11126 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11127 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11128 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11129 native order exists for base types, attach the DIE manually and call
11130 add_pubtype on it. Do not equate a reverse order DIE to the type.
11132 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11134 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11135 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11137 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11139 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11141 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11142 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11143 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11144 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11145 * opts.c (sanitizer_opts): Add builtin.
11146 * ubsan.c (instrument_builtin): New function.
11147 (pass_ubsan::execute): Call it.
11148 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11149 * doc/invoke.texi: Document -fsanitize=builtin.
11151 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11152 builtins, store max (log2 (align), 0) into uchar field instead of
11153 align into uptr field.
11154 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11155 store uchar 0 field instead of uptr 0 field.
11156 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11157 instead of passing one address of struct with 2 locations pass
11158 two addresses of structs with 1 location each.
11159 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11160 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11161 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11162 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11163 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11164 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11165 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11166 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11168 2017-10-19 Martin Liska <mliska@suse.cz>
11171 * file-find.c (remove_prefix): Remove.
11172 * file-find.h (remove_prefix): Likewise.
11173 * gcc-ar.c: Remove smartness of lookup.
11175 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11177 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11178 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11179 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11181 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11184 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11185 (setcc + and to xor + setcc): New peephole2.
11187 2017-10-19 Tom de Vries <tom@codesourcery.com>
11189 * doc/sourcebuild.texi (Test Directives, Variants of
11190 dg-require-support): Add dg-require-stack-size.
11192 2017-10-19 Martin Liska <mliska@suse.cz>
11195 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11196 that have a large alignment.
11197 (gimplify_target_expr): Likewise.
11199 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11201 PR rtl-optimization/82602
11202 * ira.c (rtx_moveable_p): Return false for volatile asm.
11204 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11207 * config/i386/i386-modes.def (CCGZ): New CC mode.
11208 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11209 * config/i386/predicates.md (ix86_comparison_operator):
11211 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11212 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11213 with double-word subtraction.
11214 (put_condition_code): Handle CCGZmode.
11216 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11218 * wide-int.cc (debug (const wide_int &)): New.
11219 (debug (const wide_int *)): New.
11220 (debug (const widest_int &)): New.
11221 (debug (const widest_int *)): New.
11223 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11225 PR middle-end/82556
11226 * lra-constraints.c (curr_insn_transform): Use non-input operand
11227 instead of output one for matched reload.
11229 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11231 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11232 (tree-ssa-loop-ivopts.h): New header file.
11233 (struct builtin_info): New fields.
11234 (classify_builtin_1): Compute and record base and offset parts for
11235 memset builtin partition by calling strip_offset.
11236 (offset_cmp, fuse_memset_builtins): New functions.
11237 (finalize_partitions): Fuse adjacent memset partitions by calling
11239 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11240 Expose the interface.
11241 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11243 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11245 PR tree-optimization/82574
11246 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11247 that data reference must be executed exactly once per iteration
11248 against the outermost loop in nest.
11249 (classify_partition): Update call to above function.
11251 2017-10-18 Richard Biener <rguenther@suse.de>
11253 PR tree-optimization/82591
11254 * graphite.c (graphite_transform_loops): Move code gen message
11256 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11257 Here. Handle scop_to_isl_ast failing.
11258 (scop_to_isl_ast): Limit the number of ISL operations.
11260 2017-10-18 Richard Biener <rguenther@suse.de>
11262 * graphite-isl-ast-to-gimple.c
11263 (translate_isl_ast_to_gimple::set_rename): Simplify.
11264 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11265 (graphite_copy_stmts_from_block): ... here.
11266 (copy_bb_and_scalar_dependences): Simplify.
11267 (add_parameters_to_ivs_params): Canonicalize.
11268 (generate_entry_out_of_ssa_copies): Simplify.
11269 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11270 by passing in ISL dimension.
11271 (parameter_index_in_region_1): Rename to ...
11272 (parameter_index_in_region): ... this.
11273 (extract_affine): Adjust assert, pass down parameter index.
11274 (add_param_constraints): Use range-info when available.
11275 (build_scop_context): Adjust.
11276 * sese.c (new_sese_info): Adjust.
11277 (free_sese_info): Likewise.
11278 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11279 Remove unused typedefs.
11280 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11282 2017-10-18 Martin Liska <mliska@suse.cz>
11284 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11286 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11288 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11289 (s390_sched_init): Do not reset s390_sched_state if we entered the
11290 current basic block via a fallthru edge and all others are unlikely.
11292 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11294 * config/s390/s390.c (NUM_SIDES): New variable.
11295 (LONGRUNNING_THRESHOLD): New variable.
11296 (LATENCY_FACTOR): New variable.
11297 (s390_sched_score): Decrease score for long-running instructions on
11299 (s390_sched_variable_issue): Perform bookkeeping for long-running
11302 2017-10-18 Richard Biener <rguenther@suse.de>
11304 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11305 Simplify with removal of the parameter rename map.
11306 (set_rename): Likewise.
11307 (should_copy_to_new_region): Likewise.
11308 (graphite_copy_stmts_from_block): Likewise.
11309 (copy_bb_and_scalar_dependences): Remove initialization of
11310 unused copied_bb_map.
11311 (copy_def): Remove.
11312 (copy_internal_parameters): Likewise.
11313 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11314 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11315 Use INTEGRAL_TYPE_P.
11316 (parameter_index_in_region_1): Rename to ...
11317 (assign_parameter_index_in_region): ... this. Assert we have
11318 a parameter we handle.
11319 (scan_tree_for_params): Adjust.
11320 * sese.h (parameter_rename_map_t): Remove.
11321 (struct sese_info_t): Remove unused parameter_rename_map and
11322 copied_bb_map members.
11323 * sese.c (new_sese_info): Adjust.
11324 (free_sese_info): Likewise.
11326 2017-10-18 Martin Liska <mliska@suse.cz>
11329 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11330 on an abnormal edge.
11332 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11334 * doc/invoke.texi (ffunction-sections and fdata-sections):
11337 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11339 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11340 the use statement can throw internally.
11342 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11344 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11345 any RTX present on the RHS of a SET.
11346 * compare-elim.c (try_eliminate_compare): Restore comment.
11348 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11350 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11351 may be also called on tcc_type nodes.
11352 * langhooks.c (lhd_tree_size): Likewise.
11354 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11356 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11357 format_warning_at_substring.
11358 (maybe_warn): Convert source_range * param to a location_t. Pass
11359 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11360 (format_directive): Remove code to extract source_ranges and
11361 source_range * in favor of just a location_t.
11362 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11364 * substring-locations.c (format_warning_va): Convert
11365 source_range * param to a location_t.
11366 (format_warning_at_substring): Likewise.
11367 * substring-locations.h (format_warning_va): Likewise.
11368 (format_warning_at_substring): Likewise.
11370 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11372 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11374 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11375 and scatter/gather ops.
11377 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11378 vec_gather_load and vec_scatter_store.
11379 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11380 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11381 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11382 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11383 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11384 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11386 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11388 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11389 Detect FTST instruction and handle its register pops. Only pop
11390 second operand if can_pop_second_op is true.
11391 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11392 set can_pop_second_op to false in the compare_for_stack_reg call.
11394 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11395 output_fp_compare for stack register operands.
11396 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11397 instructions here. Do not emit stack register pops here. Assert
11398 that FCOMPP pops next to top stack register. Rewrite function.
11400 2017-10-17 Nathan Sidwell <nathan@acm.org>
11402 PR middle-end/82577
11403 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11404 use DECL_ASSEMBLER_NAME_RAW.
11406 PR middle-end/82546
11407 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11410 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11411 Wilco Dijkstra <wilco.dijkstra@arm.com>
11413 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11414 converstion to Pmode from the buf_addr.
11416 2017-10-17 Richard Biener <rguenther@suse.de>
11418 * graphite-dependences.c (scop_get_reads_and_writes): Change
11419 output parameters to references.
11421 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11423 PR 71026/tree-optimization
11424 * fold-const.c (distribute_real_division): Removed.
11425 (fold_binary_loc): Remove calls to distribute_real_divison.
11427 2017-10-17 Richard Biener <rguenther@suse.de>
11429 * graphite-scop-detection.c
11430 (scop_detection::stmt_has_simple_data_refs_p): Always use
11431 the full nest as region.
11432 (try_generate_gimple_bb): Likewise.
11433 * sese.c (scalar_evolution_in_region): Simplify now that
11434 SCEV can handle instantiation in regions.
11435 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11436 in the non-loop part of a function if requested.
11438 2017-10-17 Richard Biener <rguenther@suse.de>
11440 PR tree-optimization/82563
11441 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11443 (graphite_regenerate_ast_isl): Call it.
11444 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11446 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11448 PR tree-optimization/82549
11449 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11450 Formatting fixes. Instead of calling make_bit_field_ref with negative
11453 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11455 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11456 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11457 _mm_maskz_reduce_ss): New.
11458 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11459 __builtin_ia32_reducess_mask): Ditto..
11460 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11461 * config/i386/sse.md (reduces<mode>): Renamed to ...
11462 (reduces<mode><mask_scalar_name>): ... this.
11463 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11465 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11466 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11468 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11470 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11471 * selftest-run-tests.c (selftest::run_tests): Call
11472 selftest::unique_ptr_tests_cc_tests.
11473 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11474 * unique-ptr-tests.cc: New file.
11476 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11479 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11481 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11482 (make_hard_regno_born, make_hard_regno_dead): Update
11483 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11485 2017-10-16 Jeff Law <law@redhat.com>
11487 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11489 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11491 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11493 2017-10-16 Olivier Hainque <hainque@adacore.com>
11495 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11496 with_cpu if we were configured for an e500v2 target cpu name.
11498 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11500 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11501 * doc/invoke.texi: Document +nodsp as a valid extension for
11504 2017-10-16 Martin Liska <mliska@suse.cz>
11506 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11507 (test_set_range): Likewise.
11508 (test_range_functions): Rename to ...
11509 (test_bit_in_range): ... this.
11510 (sbitmap_c_tests): Add new test.
11512 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11514 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11516 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11517 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11519 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11521 * config/aarch64/aarch64-builtins.c
11522 (aarch64_types_quadopu_lane_qualifiers): New.
11523 (TYPES_QUADOPU_LANE): New.
11524 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11525 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11526 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11527 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11528 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11529 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11530 (Vdottype, DOTPROD): New.
11531 (sur): Add SDOT and UDOT.
11533 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11535 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11536 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11537 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11538 Add TARGET_DOTPROD.
11539 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11540 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11541 Enable TARGET_DOTPROD.
11542 (cortex-a75.cortex-a55): Likewise.
11543 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11545 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11547 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11548 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11550 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11552 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11553 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11554 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11555 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11556 * config/arm/types.md (neon_dot, neon_dot_q): New.
11557 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11559 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11561 * config/arm/arm.h (TARGET_DOTPROD): New.
11562 * config/arm/arm.c (arm_arch_dotprod): New.
11563 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11564 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11565 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11566 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11567 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11568 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11569 * doc/invoke.texi (armv8.2-a): Document dotprod
11571 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11573 * i386.c (ix86_vec_cost): New function.
11574 (ix86_rtx_costs): Handle vector operations better.
11575 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11576 * x86-tune-costs.h: Add new costs to all tables.
11578 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11580 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11582 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11583 divsd, sqrtss and sqrtsd
11584 * x86-tune-costs.h: Add new entries to all costs.
11585 (znver1_cost): Fix to match real instruction latencies.
11587 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11588 Michael Collison <michael.collison@arm.com>
11590 * compare-elim.c: Include emit-rtl.h.
11591 (can_merge_compare_into_arith): New function.
11592 (try_validate_parallel): Likewise.
11593 (try_merge_compare): Likewise.
11594 (try_eliminate_compare): Call the above when no previous clobber
11596 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11599 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11601 PR middle-end/62263
11602 PR middle-end/82498
11603 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11604 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11606 PR middle-end/62263
11607 PR middle-end/82498
11608 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11609 to be any operand_equal_p operands. For & (B - 1) require
11610 B to be power of 2. Recognize
11611 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11613 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11616 * optabs.c (expand_memory_blockage): Fix call of
11617 targetm.have_memory_blockage.
11619 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11622 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11623 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11624 objects to extra_objs instead of overwriting it.
11626 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11628 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11629 Use any_fp_register_operand as operand[3] predicate. Simplify
11630 equality test for operands[2] and operands[4] memory location.
11631 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11632 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11633 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11634 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11635 any_fp_register_operand as operand[1] predicate. Simplify
11636 equality test for operands[0] and operands[3] memory location.
11637 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11638 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11639 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11641 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11643 * target-insns.def: Add memory_blockage.
11644 * optabs.c (expand_memory_blockage): New function.
11645 (expand_asm_memory_barrier): Rename ...
11646 (expand_asm_memory_blockage): ... to this.
11647 (expand_mem_thread_fence): Call expand_memory_blockage
11648 instead of expand_asm_memory_barrier.
11649 (expand_mem_singnal_fence): Ditto.
11650 (expand_atomic_load): Ditto.
11651 (expand_atomic_store): Ditto.
11652 * doc/md.texi (Standard Pattern Names For Generation):
11653 Document memory_blockage instruction pattern.
11655 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11657 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11658 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11659 * config/rl78/rl78.md: New define_expand "adddi3".
11661 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11663 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11666 2017-10-13 Jeff Law <law@redhat.com>
11668 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11670 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11673 * internal-fn.c (expand_mul_overflow): If both operands have
11674 the same highpart of -1 or 0 and the topmost bit of lowpart
11675 is different, overflow is if res <= 0 rather than res < 0.
11677 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11679 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11680 TARGET_P9_VECTOR code for unaligned_load case.
11682 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11684 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11686 2017-10-13 Nathan Sidwell <nathan@acm.org>
11688 * tree-core.h (tree_contains_struct): Make bool.
11689 * tree.c (tree_contains_struct): Likewise.
11690 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11691 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11692 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11693 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11694 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11696 2017-10-13 Richard Biener <rguenther@suse.de>
11698 * graphite-isl-ast-to-gimple.c
11699 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11700 parameters and dominance check.
11701 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11702 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11703 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11704 Do not update SSA form here or do intermediate IL verification.
11705 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11706 (graphite_initialize): Remove check on the number of loops in
11707 the function and inline into graphite_transform_loops.
11708 (graphite_finalize): Inline into graphite_transform_loops.
11709 (graphite_transform_loops): Perform SSA update and IL verification
11711 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11713 2017-10-13 Richard Biener <rguenther@suse.de>
11715 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11716 graphite_expression_type_precision): Avoid global constructor
11718 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11719 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11720 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11721 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11722 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11723 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11725 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11728 * config/i386/i386.h (ix86_red_zone_size): New.
11729 * config/i386/i386.md (push peephole2s): Replace
11730 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11732 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11733 Alan Hayward <alan.hayward@arm.com>
11734 David Sherwood <david.sherwood@arm.com>
11736 * combine.c (can_change_dest_mode): Reject changes in
11737 REGMODE_NATURAL_SIZE.
11739 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11740 Alan Hayward <alan.hayward@arm.com>
11741 David Sherwood <david.sherwood@arm.com>
11743 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11744 (expand_debug_source_expr): Likewise.
11745 * combine.c (combine_simplify_rtx): Likewise.
11746 * cse.c (fold_rtx): Likewise.
11747 * fwprop.c (canonicalize_address): Likewise.
11748 * targhooks.c (default_shift_truncation_mask): Likewise.
11750 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11751 Alan Hayward <alan.hayward@arm.com>
11752 David Sherwood <david.sherwood@arm.com>
11754 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11755 (widened_mode): Likewise.
11756 (expand_unop): Likewise.
11757 * ree.c (transform_ifelse): Likewise.
11758 (merge_def_and_ext): Likewise.
11759 (combine_reaching_defs): Likewise.
11760 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11762 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11763 Alan Hayward <alan.hayward@arm.com>
11764 David Sherwood <david.sherwood@arm.com>
11766 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11767 * combine.c (gen_lowpart_for_combine): Likewise.
11768 * dwarf2out.c (rtl_for_decl_location): Likewise.
11769 * final.c (alter_subreg): Likewise.
11770 * rtlhooks.c (gen_lowpart_general): Likewise.
11771 (gen_lowpart_if_possible): Likewise.
11773 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11774 Alan Hayward <alan.hayward@arm.com>
11775 David Sherwood <david.sherwood@arm.com>
11777 * calls.c (expand_call): Use subreg_lowpart_offset.
11778 * cse.c (cse_insn): Likewise.
11779 * regcprop.c (copy_value): Likewise.
11780 (copyprop_hardreg_forward_1): Likewise.
11782 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11785 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11786 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11787 =Q constraints to +Q and into insn condition add check
11788 that operands[0] and operands[1] are equal.
11789 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11790 =Q constraints to +Q and into insn condition add check
11791 that operands[0] is equal to either operands[1] or operands[2].
11794 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11795 instead of handling MINUS_EXPR twice (once for each argument),
11796 canonicalize operand order and handle just once, use rtype where
11797 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11800 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11801 any values of __C while still being pattern recognizable as a simple
11802 rotate instruction.
11804 2017-10-13 Richard Biener <rguenther@suse.de>
11806 PR tree-optimization/82451
11808 2017-10-02 Richard Biener <rguenther@suse.de>
11810 PR tree-optimization/82355
11811 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11812 a mapping for the enclosing loop but avoid generating one for
11813 the loop tree root.
11814 (copy_bb_and_scalar_dependences): Remove premature codegen
11815 error on PHIs in blocks duplicated into multiple places.
11816 * graphite-scop-detection.c
11817 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11818 in the region use it as loop and nest to analyze the DR in.
11819 (try_generate_gimple_bb): Likewise.
11820 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11821 (add_loop_constraints): For blocks in a loop not in the region
11822 create a dimension with a single iteration.
11823 * sese.h (gbb_loop_at_index): Remove assert.
11825 * cfgloop.c (loop_preheader_edge): For the loop tree root
11826 return the single successor of the entry block.
11827 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11828 Reset the SCEV hashtable and niters.
11829 * graphite-scop-detection.c
11830 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11831 assert that we only have POLYNOMIAL_CHREC that vary in loops
11832 contained in the region.
11833 (scop_detection::graphite_can_represent_expr): Adjust.
11834 (scop_detection::stmt_has_simple_data_refs_p): For loops
11835 not in the region set loop to NULL. The nest is now the
11836 entry edge to the region.
11837 (try_generate_gimple_bb): Likewise.
11838 * sese.c (scalar_evolution_in_region): Adjust for
11839 instantiate_scev change.
11840 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11841 Make nest parameter the edge into the region.
11842 (create_data_ref): Likewise.
11843 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11844 entry edge into a region and adjust instantiate_scev calls.
11845 (create_data_ref): Likewise.
11846 (graphite_find_data_references_in_stmt): Likewise.
11847 (find_data_references_in_stmt): Pass the loop preheader edge
11848 from the nest argument.
11849 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11850 parameter the edge into the region.
11851 (instantiate_parameters): Use the loop preheader edge as entry.
11852 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11854 (get_instantiated_value_entry): Make instantiate_below parameter
11855 the edge into the region.
11856 (instantiate_scev_name): Likewise. Adjust dominance checks,
11857 when we cannot use loop-based instantiation instantiate by
11858 walking use-def chains.
11859 (instantiate_scev_poly): Adjust.
11860 (instantiate_scev_binary): Likewise.
11861 (instantiate_scev_convert): Likewise.
11862 (instantiate_scev_not): Likewise.
11863 (instantiate_array_ref): Remove.
11864 (instantiate_scev_3): Likewise.
11865 (instantiate_scev_2): Likewise.
11866 (instantiate_scev_1): Likewise.
11867 (instantiate_scev_r): Do not blindly handle N-operand trees.
11868 Do not instantiate array-refs. Handle all constants and invariants.
11869 (instantiate_scev): Make instantiate_below parameter
11870 the edge into the region.
11871 (resolve_mixers): Use the loop preheader edge for the region
11872 parameter to instantiate_scev_r.
11873 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11875 2017-10-13 Richard Biener <rguenther@suse.de>
11877 PR tree-optimization/82525
11878 * graphite-isl-ast-to-gimple.c
11879 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11881 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11882 Fail code generation when we cannot represent the isl integer.
11883 (binary_op_to_tree): Elide modulo operations that are no-ops
11884 in the type we code generate. Remove now superfluous code
11887 2017-10-13 Richard Biener <rguenther@suse.de>
11889 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11890 (scop_detection::harmful_loop_in_region): Remove premature
11891 IV type restriction.
11892 (scop_detection::graphite_can_represent_scev): We can handle
11893 pointer IVs just fine.
11895 2017-10-13 Alan Modra <amodra@gmail.com>
11897 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11898 "Clobbers and Scratch Registers". Add paragraph on
11899 alternative to clobbers for scratch registers and OpenBLAS
11902 2017-10-13 Alan Modra <amodra@gmail.com>
11904 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11905 example of a memory input for a string of known length. Move
11906 commentary out of table. Add a number of new examples
11907 covering array memory inputs.
11909 2017-10-12 Martin Liska <mliska@suse.cz>
11911 PR tree-optimization/82493
11912 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11913 (test_range_functions): New function.
11914 (sbitmap_c_tests): Likewise.
11915 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11916 * selftest.h (sbitmap_c_tests): New function.
11918 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11920 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11922 * config/rs6000/amo.h: Fix spacing issue.
11924 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11927 * config/i386/i386.md (*ashl<mode>3_mask_1,
11928 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11929 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11932 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11934 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11935 (profile_probability): Set max_probability
11936 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11938 * profile-count.c (profile_probability::differs_from_p): Do not
11939 rely on max_probaiblity == 10000
11941 2017-10-12 Jeff Law <law@redhat.com>
11943 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11946 2017-10-12 Martin Sebor <msebor@redhat.com>
11950 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11951 (handle_alias_pairs): Call it.
11952 * common.opt (-Wattribute-alias): New option.
11953 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11954 * doc/invoke.texi (-Wattribute-alias): Document.
11956 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11959 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11961 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11963 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11964 (make_hard_regno_born, make_hard_regno_dead): Update
11965 bb_killed_pseudos and bb_gen_pseudos.
11967 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11969 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11971 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11973 * config/alpha/alpha.c (alpha_split_conditional_move):
11974 Use std::swap instead of manually swapping.
11975 (alpha_stdarg_optimize_hook): Ditto.
11976 (alpha_canonicalize_comparison): Ditto.
11978 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11980 * tree-loop-distribution.c (struct builtin_info): New struct.
11981 (struct partition): Refactor fields into struct builtin_info.
11982 (partition_free): Free struct builtin_info.
11983 (build_size_arg_loc, build_addr_arg_loc): Delete.
11984 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11985 information from struct builtin_info.
11986 (find_single_drs): New function refactored from classify_partition.
11987 Also moved builtin validity checks to this function.
11988 (compute_access_range, alloc_builtin): New functions.
11989 (classify_builtin_st, classify_builtin_ldst): New functions.
11990 (classify_partition): Refactor code into functions find_single_drs,
11991 classify_builtin_st and classify_builtin_ldst.
11992 (distribute_loop): Don't do runtime alias check when distributing
11994 (find_seed_stmts_for_distribution): New function.
11995 (pass_loop_distribution::execute): Refactor code finding seed
11996 stmts into above function. Support distribution for the innermost
11997 two-level loop nest. Adjust dump information.
11999 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12001 * tree-loop-distribution.c: Adjust the general comment.
12002 (NUM_PARTITION_THRESHOLD): New macro.
12003 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12004 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12005 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12006 in call to build_partition_graph.
12007 (finalize_partitions): New parameter. Make loop distribution more
12008 conservative by fusing more partitions.
12009 (distribute_loop): Don't do runtime alias check in case of loop nest
12011 (find_seed_stmts_for_distribution): New function.
12012 (prepare_perfect_loop_nest): New function.
12013 (pass_loop_distribution::execute): Refactor code finding seed stmts
12014 and loop nest into above functions. Support loop nest distribution.
12015 Adjust dump information accordingly.
12017 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12019 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12020 and set PTYPE_SEQUENTIAL for merged partition.
12022 2017-10-12 Richard Biener <rguenther@suse.de>
12024 PR tree-optimization/69728
12026 2017-09-19 Richard Biener <rguenther@suse.de>
12028 PR tree-optimization/69728
12029 * graphite-sese-to-poly.c (schedule_error): New global.
12030 (add_loop_schedule): Handle empty domain by failing the
12032 (build_original_schedule): Handle schedule_error.
12034 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12035 domain by returning an unchanged schedule.
12037 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12039 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12040 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12042 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12044 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12047 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12050 * expr.c (store_field): Don't optimize away bitsize == 0 store
12051 from CALL_EXPR with addressable return type.
12053 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12055 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12056 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12057 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12058 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12059 TARGET_ISEL instead of TARGET_ISEL<sel>.
12061 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12063 * config/rs6000/rs6000.c
12064 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12066 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12068 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12070 (reg_or_cint_operand): Fix comment.
12071 (reg_or_zero_operand): New predicate.
12072 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12073 * config/rs6000/rs6000.c (output_isel): Delete.
12074 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12075 instead of reg_or_cint_operand. Output instruction directly (not via
12077 (isel_unsigned_<mode>): Ditto.
12078 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12079 gpc_reg_operand. Add an instruction alternative for this. Output
12080 instruction directly.
12081 (*isel_reversed_unsigned_<mode>): Ditto.
12083 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12085 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12086 (TARGET_CANONICALIZE_COMPARISON): Define.
12088 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12091 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12092 Check whether the dest is REG before adding REG_EQUIV note.
12094 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12097 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12099 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12100 (make_hard_regno_born, make_hard_regno_dead): Update
12101 bb_killed_pseudos and bb_gen_pseudos.
12103 2017-10-11 Nathan Sidwell <nathan@acm.org>
12105 * incpath.h (enum incpath_kind): Name enum, prefix values.
12106 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12107 * incpath.c (heads, tails): Use INC_MAX.
12108 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12109 (merge_include_chains, split_quote_chain,
12110 register_include_chains): Update incpath_kind names.
12111 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12112 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12114 (add_framework_path, darwin_register_objc_includes): Likewise.
12115 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12117 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12119 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12120 Do not use float_operator operator predicate.
12121 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12122 * config/i386/predicates.md (float_operator): Remove predicate.
12124 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12126 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12127 (*jccxf_i387): Ditto.
12128 (*jcc<mode>_i387): Ditto.
12129 (*jccu<mode>_i387): Ditto.
12130 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12131 (*jcc_*_i387 splitters): Remove.
12132 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12133 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12134 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12137 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12139 * profile-count.h (slow_safe_scale_64bit): New function.
12140 (safe_scale_64bit): New inline.
12141 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12142 * profile-count.c: Include wide-int.h
12143 (slow_safe_scale_64bit): New.
12145 2017-10-11 Nathan Sidwell <nathan@acm.org>
12147 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12148 HAS_DECL_ASSEMBLER_NAME_P.
12149 * gimple-expr.c (gimple_decl_printable_name: Check
12150 HAS_DECL_ASSEMBLER_NAME_P too.
12151 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12152 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12153 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12154 * passes.c (rest_of_decl_compilation): Check
12155 HAS_DECL_ASSEMBLER_NAME_P too.
12156 * recog.c (verify_changes): Likewise.
12157 * tree-pretty-print.c (dump_decl_name): Likewise.
12158 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12160 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12161 (SET_DECL_ASSEMBLER_NAME): Use it.
12162 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12163 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12164 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12166 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12168 * config.gcc (i386, x86_64): Add extra objects.
12169 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12170 (ix86_min_insn_size): Declare.
12171 (ix86_issue_rate): Declare.
12172 (ix86_adjust_cost): Declare.
12173 (ia32_multipass_dfa_lookahead): Declare.
12174 (ix86_macro_fusion_p): Declare.
12175 (ix86_macro_fusion_pair_p): Declare.
12176 (ix86_bd_has_dispatch): Declare.
12177 (ix86_bd_do_dispatch): Declare.
12178 (ix86_core2i7_init_hooks): Declare.
12179 (ix86_atom_sched_reorder): Declare.
12180 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12181 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12182 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12183 (rip_relative_addr_p): Rename to ...
12184 (ix86_rip_relative_addr_p): ... this one; export.
12185 (memory_address_length): Update.
12186 (ix86_issue_rate): Move to x86-tune-sched.c.
12187 (ix86_flags_dependent): Move to x86-tune-sched.c.
12188 (ix86_agi_dependent): Move to x86-tune-sched.c.
12189 (exact_dependency_1): Move to x86-tune-sched.c.
12190 (exact_store_load_dependency): Move to x86-tune-sched.c.
12191 (ix86_adjust_cost): Move to x86-tune-sched.c.
12192 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12193 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12194 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12195 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12196 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12197 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12198 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12199 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12200 (min_insn_size): Rename to ...
12201 (ix86_min_insn_size): ... this one; export.
12202 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12203 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12204 (core2i7_first_cycle_multipass_backtrack): Move to
12205 x86-tune-sched-core.c.
12206 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12207 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12208 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12209 (ix86_avoid_jump_mispredicts): Update.
12210 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12211 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12212 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12213 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12214 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12215 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12216 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12217 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12218 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12219 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12220 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12221 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12222 (BIG): Move to ix86-tune-sched-bd.c.
12223 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12224 (enum insn_path): Move to ix86-tune-sched-bd.c.
12225 (get_mem_group): Move to ix86-tune-sched-bd.c.
12226 (is_cmp): Move to ix86-tune-sched-bd.c.
12227 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12228 (is_branch): Move to ix86-tune-sched-bd.c.
12229 (is_prefetch): Move to ix86-tune-sched-bd.c.
12230 (init_window): Move to ix86-tune-sched-bd.c.
12231 (allocate_window): Move to ix86-tune-sched-bd.c.
12232 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12233 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12234 (process_end_window): Move to ix86-tune-sched-bd.c.
12235 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12236 (find_constant): Move to ix86-tune-sched-bd.c.
12237 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12238 (has_immediate): Move to ix86-tune-sched-bd.c.
12239 (get_insn_path): Move to ix86-tune-sched-bd.c.
12240 (get_insn_group): Move to ix86-tune-sched-bd.c.
12241 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12242 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12243 (add_insn_window): Move to ix86-tune-sched-bd.c.
12244 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12245 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12246 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12247 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12248 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12249 (do_dispatch): Move to ix86-tune-sched-bd.c.
12250 (has_dispatch): Move to ix86-tune-sched-bd.c.
12251 * config/i386/t-i386: Add new object files.
12252 * config/i386/x86-tune-costs.h: New file.
12253 * config/i386/x86-tune-sched-atom.c: New file.
12254 * config/i386/x86-tune-sched-bd.c: New file.
12255 * config/i386/x86-tune-sched-core.c: New file.
12256 * config/i386/x86-tune-sched.c: New file.
12258 2017-10-11 Liu Hao <lh_mouse@126.com>
12260 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12261 the generic version below instead.
12262 (should_colorize): Recognize Windows consoles as terminals
12264 * pretty-print.c [__MINGW32__] (write_all): New function.
12265 [__MINGW32__] (find_esc_head): Likewise.
12266 [__MINGW32__] (find_esc_terminator): Likewise.
12267 [__MINGW32__] (eat_esc_sequence): Likewise.
12268 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12270 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12273 2017-10-11 Richard Biener <rguenther@suse.de>
12275 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12276 Properly call analyze_scalar_evolution with the loop of the stmt.
12278 2017-10-11 Richard Biener <rguenther@suse.de>
12280 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12281 * tree-core.h (tree_base): Add chrec_var union member.
12282 * tree.h (CHREC_VAR): Remove.
12283 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12284 * tree-chrec.h (build_polynomial_chrec): Adjust.
12285 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12286 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12288 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12290 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12291 * match.pd: ... here.
12292 ((T) X == (T) Y): Relax condition.
12294 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12296 PR tree-optimization/82472
12297 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12299 (break_alias_scc_partitions): Update postorder number.
12301 2017-10-11 Martin Liska <mliska@suse.cz>
12304 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12306 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12308 2017-10-11 Martin Liska <mliska@suse.cz>
12310 * print-rtl.c (print_insn): Move declaration of idbuf
12311 to same scope as name.
12313 2017-10-11 Martin Liska <mliska@suse.cz>
12318 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12319 volatile arguments.
12321 2017-10-11 Martin Liska <mliska@suse.cz>
12324 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12325 volatile arguments.
12327 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12329 * config.gcc (default_gnu_indirect_function): Default to yes for
12330 arm*-*-linux* with glibc.
12332 2017-10-11 Richard Biener <rguenther@suse.de>
12334 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12335 default-defs and types we do not want to analyze.
12336 (interpret_loop_phi): Replace unreachable code with an assert.
12337 (compute_scalar_evolution_in_loop): Remove and inline ...
12338 (analyze_scalar_evolution_1): ... here, replacing condition with
12339 what makes the intent clearer. Remove handling of cases
12340 get_scalar_evolution now handles.
12342 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12344 PR rtl-optimization/81434
12345 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12346 comment for main loop. In sched_group_found if, also add checks for
12347 pass and min_cost_group.
12349 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12351 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12352 (rs6000_insn_cost): New function.
12353 * config/rs6000/rs6000.md (cost): New attribute.
12355 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12356 H.J. Lu <hongjiu.lu@intel.com>
12360 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12361 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12362 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12363 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12364 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12365 OPTION_MASK_ISA_MMX and some other bit set require both
12366 mmx and the ISAs without the mmx bit.
12367 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12368 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12369 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12370 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12371 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12372 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12373 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12374 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12375 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12376 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12377 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12378 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12379 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12380 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12381 Add OPTION_MASK_ISA_MMX.
12383 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12385 * config.gcc (armv7*-*-freebsd*): New target.
12386 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12388 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12390 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12391 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12394 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12396 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12397 a template parameter.
12398 (WIDE_INT_REF_FOR): Update accordingly.
12399 * tree.h (wi::int_traits <const_tree>): Delete.
12400 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12401 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12402 (wi::tree_to_wide_ref): New typedef.
12403 (wi::to_wide): New function.
12404 * calls.c (get_size_range): Use wi::to_wide when operating on
12405 trees as wide_ints.
12406 * cgraph.c (cgraph_node::create_thunk): Likewise.
12407 * config/i386/i386.c (ix86_data_alignment): Likewise.
12408 (ix86_local_alignment): Likewise.
12409 * dbxout.c (stabstr_O): Likewise.
12410 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12411 * expr.c (const_vector_from_tree): Likewise.
12412 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12413 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12414 (fold_negate_expr_1, int_const_binop_1, const_binop)
12415 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12416 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12417 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12418 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12419 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12420 (fold_not_const, round_up_loc): Likewise.
12421 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12422 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12423 (alloca_call_type): Likewise.
12424 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12425 * godump.c (go_output_typedef): Likewise.
12426 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12427 * internal-fn.c (get_min_precision): Likewise.
12428 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12429 * ipa-polymorphic-call.c
12430 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12431 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12432 (ipa_modify_call_arguments): Likewise.
12433 * match.pd: Likewise.
12434 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12435 * print-tree.c (print_node_brief, print_node): Likewise.
12436 * stmt.c (expand_case): Likewise.
12437 * stor-layout.c (layout_type): Likewise.
12438 * tree-affine.c (tree_to_aff_combination): Likewise.
12439 * tree-cfg.c (group_case_labels_stmt): Likewise.
12440 * tree-data-ref.c (dr_analyze_indices): Likewise.
12441 (prune_runtime_alias_test_list): Likewise.
12442 * tree-dump.c (dequeue_and_dump): Likewise.
12443 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12444 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12445 * tree-pretty-print.c (dump_generic_node): Likewise.
12446 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12447 (simple_iv_with_niters): Likewise.
12448 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12449 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12450 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12451 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12452 (refine_value_range_using_guard, number_of_iterations_ne_max)
12453 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12454 (get_cst_init_from_scev, record_nonwrapping_iv)
12455 (scev_var_range_cant_overflow): Likewise.
12456 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12457 * tree-ssa-pre.c (compute_avail): Likewise.
12458 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12459 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12460 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12461 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12462 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12463 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12464 (dump_case_nodes, try_switch_expansion): Likewise.
12465 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12466 (vect_do_peeling): Likewise.
12467 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12468 * tree-vect-stmts.c (vectorizable_load): Likewise.
12469 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12470 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12471 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12472 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12473 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12474 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12475 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12476 (evrp_dom_walker::before_dom_children): Likewise.
12477 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12478 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12479 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12480 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12481 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12482 (get_range_pos_neg): Likewise.
12483 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12484 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12485 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12486 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12487 * config/avr/avr.c (avr_fold_builtin): Likewise.
12488 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12489 * config/msp430/msp430.c (msp430_attr): Likewise.
12490 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12491 * config/powerpcspe/powerpcspe-c.c
12492 (altivec_resolve_overloaded_builtin): Likewise.
12493 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12494 (rs6000_expand_ternop_builtin): Likewise.
12495 * config/rs6000/rs6000-c.c
12496 (altivec_resolve_overloaded_builtin): Likewise.
12497 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12498 (rs6000_expand_ternop_builtin): Likewise.
12499 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12501 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12503 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12504 when copying loop nest with only one inner loop.
12506 2017-10-10 Richard Biener <rguenther@suse.de>
12508 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12509 blocks if SCEV is active.
12510 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12512 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12513 (scev_initialize): Assert we are not yet initialized.
12515 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12517 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12518 inner loop's exit stmt by making it always exit the loop, otherwise
12519 we would generate an infinite empty loop.
12521 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12523 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12524 renaming variables in new preheader if it's deleted.
12526 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12528 * tree-loop-distribution.c (struct partition): Remove unused field
12529 loops of the structure.
12530 (partition_alloc, partition_free): Ditto.
12531 (build_rdg_partition_for_vertex): Ditto.
12533 2017-10-09 Jeff Law <law@redhat.com>
12535 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12536 return type to match prototype and documentation.
12538 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12540 * config/rs6000/rs6000.c (processor_costs): Move to ...
12541 * config/rs6000/rs6000.h: ... here.
12542 (rs6000_cost): Declare.
12544 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12546 * except.c (setjmp_fn): New global variable.
12547 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12548 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12549 if DONT_USE_BUILTIN_SETJMP is defined.
12551 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12553 * target.def (insn_cost): New hook.
12554 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12555 * doc/tm.texi: Regenerate.
12556 * rtlanal.c (insn_cost): Use the new hook.
12558 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12560 * combine.c (combine_validate_cost): Compute the new insn_cost,
12561 not just pattern_cost.
12562 (try_combine): Adjust comment.
12564 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12566 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12568 * combine.c (uid_insn_cost): Adjust comment.
12569 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12571 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12572 * dse.c (find_shift_sequence): Ditto.
12573 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12574 (bb_valid_for_noce_process_p): Use pattern_cost.
12575 * rtl.h (insn_rtx_cost): Delete.
12576 (pattern_cost): New prototype.
12577 (insn_cost): New prototype.
12578 * rtlanal.c (insn_rtx_cost): Rename to...
12579 (pattern_cost): ... this.
12582 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12584 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12585 (*jcc<mode>_0_r_i387): Ditto.
12586 (*jccxf_r_i387): Ditto.
12587 (*jcc<mode>_r_i387): Ditto.
12588 (*jccu<mode>_r_i387): Ditto.
12589 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12590 (*jcc): Rename from *jcc_1.
12592 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12594 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12595 deferred rescans after the lvx/stvx recombination pre-pass.
12597 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12599 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12600 memory operation instruction support.
12601 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12602 (rs6000-ibm-aix[789]*): Likewise.
12603 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12604 Document new functions.
12606 2017-10-09 Richard Biener <rguenther@suse.de>
12608 PR tree-optimization/82397
12609 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12610 equality only for semantically equal trees.
12612 2017-10-09 Richard Biener <rguenther@suse.de>
12614 PR tree-optimization/82449
12615 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12616 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12617 allow constant addresses.
12618 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12621 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12623 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12626 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12629 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12632 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12635 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12637 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12640 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12641 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12643 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12645 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12646 (WI_BINARY_PREDICATE_RESULT): Likewise.
12647 (wi::binary_traits::operator_result): New type.
12648 (wi::binary_traits::predicate_result): Likewise.
12649 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12650 (generic_wide_int::operator==, generic_wide_int::operator!=)
12651 (generic_wide_int::operator&, generic_wide_int::and_not)
12652 (generic_wide_int::operator|, generic_wide_int::or_not)
12653 (generic_wide_int::operator^, generic_wide_int::operator+
12654 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12656 (operator~, unary operator-, operator==, operator!=, operator&)
12657 (operator|, operator^, operator+, binary operator-, operator*): New
12659 * expr.c (get_inner_reference): Use wi::bit_and_not.
12660 * fold-const.c (fold_binary_loc): Likewise.
12661 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12662 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12663 (bit_value_binop): Likewise.
12664 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12665 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12666 (extract_range_from_binary_expr_1): Likewise.
12667 (masked_increment): Likewise.
12668 (simplify_bit_ops_using_ranges): Likewise.
12670 2017-10-09 Martin Jambor <mjambor@suse.cz>
12673 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12674 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12675 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12677 (hsa_fixup_mov_insn_type): New function.
12678 (hsa_op_with_type::get_in_type): Use it.
12679 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12680 immediates in an assert.
12681 (hsa_op_with_type::extend_int_to_32bit): New method.
12682 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12683 types. Convert to dest type if necessary.
12684 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12685 (reg_for_gimple_ssa): Pass false as min32int to
12686 hsa_type_for_scalar_tree_type.
12687 (gen_hsa_addr): Fixup type when creating addresable temporary.
12688 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12689 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12690 necessary. Call hsa_fixup_mov_insn_type.
12691 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12692 extend operands and convert to dest type if necessary.
12693 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12694 to dest type if necessary.
12695 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12696 if conversion nt necessary and size matches.
12697 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12698 to dest type if necessary.
12699 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12700 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12702 (gen_hsa_clrsb): Likewise.
12703 (gen_hsa_ffs): Likewise.
12704 (gen_hsa_divmod): Extend operands and convert to dest type if
12706 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12708 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12710 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12711 Remove empty default arguments. Use a brace block as output
12713 (conditional return): Ditto.
12715 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12716 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12718 (group_ending_nop): Ditto.
12719 (doloop_end): Ditto.
12720 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12721 (splitters for those): Ditto.
12723 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12725 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12726 a conditional jump (and the compare for it) so that pc_rtx is the
12728 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12729 for the deleted and renamed ctr<mode>_internal[234] patterns.
12730 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12731 Delete second conditional return pattern.
12732 (ctr<mode>_internal2): Delete this second bdnz pattern.
12733 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12734 (ctr<mode>_internal4): Delete this second bdz pattern.
12736 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12738 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12739 (always_initialized_rtx_for_ssa_name_p): New predicate.
12740 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12741 (finish_out_of_ssa): Free new field of SA.
12742 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12743 * tree-ssa-coalesce.c: Include tree-ssa.h.
12744 (get_parm_default_def_partitions): Remove extern keyword.
12745 (get_undefined_value_partitions): New function.
12746 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12747 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12748 uninitialized bits.
12749 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12751 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12753 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12755 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12757 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12758 for targets that preffer 128bit.
12760 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12762 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12764 2017-10-08 Olivier Hainque <hainque@adacore.com>
12766 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12767 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12769 (thumb_set_return_address): Likewise.
12771 2017-10-08 Olivier Hainque <hainque@adacore.com>
12773 * common/config/arm/arm-common.c (arm_except_unwind_info):
12774 Handle DWARF2_UNWIND_INFO.
12776 2017-10-07 Michael Collison <michael.collison@arm.com>
12778 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12781 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12783 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12784 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12785 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12786 defined, force the creation of a new block for a dispatch label.
12788 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12790 * invoke.texi (Wsuggest-attribute=cold): Document.
12791 * common.opt (Wsuggest-attribute=cold): New
12792 * ipa-pure-const.c (warn_function_cold): New function.
12793 * predict.c (compute_function_frequency): Use it.
12794 * predict.h (warn_function_cold): Declare.
12796 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12798 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12801 2017-10-06 Martin Liska <mliska@suse.cz>
12803 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12804 keyword for member functions.
12805 (struct sanopt_tree_couple): New struct.
12806 (struct sanopt_tree_couple_hash): New function.
12807 (struct sanopt_ctx): Add new hash_map.
12808 (has_dominating_ubsan_ptr_check): New function.
12809 (record_ubsan_ptr_check_stmt): Likewise.
12810 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12811 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12812 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12814 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12817 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12818 aarch64_simd_valid_immediate on CONST_VECTORs.
12819 (aarch64_reg_or_bic_imm): Likewise.
12821 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12823 PR rtl-optimization/82396
12824 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12826 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12828 * graphite-dependences.c (scop_get_reads): Move code to...
12829 (scop_get_must_writes): Move code to...
12830 (scop_get_may_writes): Move code to...
12831 (scop_get_reads_and_writes): ... here.
12832 (scop_get_dependences): Call scop_get_reads_and_writes.
12834 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12836 PR tree-optimization/82434
12837 * fold-const.h (can_native_encode_type_p,
12838 can_native_encode_string_p): Remove.
12839 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12840 don't encode anything, just return what would be otherwise returned.
12841 (native_encode_fixed, native_encode_complex, native_encode_vector):
12843 (native_encode_string): Likewise. Inline by hand
12844 can_native_encode_string_p.
12845 (can_native_encode_type_p): Remove.
12846 (can_native_encode_string_p): Remove.
12847 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12848 STRING_CSTs using can_native_encode_string_p, test all
12849 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12850 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12851 argument from native_encode_expr.
12852 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12853 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12856 2017-10-06 Richard Biener <rguenther@suse.de>
12858 PR tree-optimization/82397
12859 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12860 operand_equal_p but rely on data_ref_compare_tree for detecting
12862 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12863 to match up with dr_group_sort_cmp.
12865 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12868 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12870 * config/s390/s390-builtin-types.def: Regenerate.
12872 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12875 * config/s390/s390-builtin-types.def: Regenerate.
12876 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12877 Change flag from B_VXE to B_VX.
12878 (s390_vec_min_dbl): Remove B_VXE flag.
12880 2017-10-06 Richard Biener <rguenther@suse.de>
12882 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12883 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12884 translate_isl_ast_to_gimple::is_valid_rename,
12885 translate_isl_ast_to_gimple::get_rename,
12886 translate_isl_ast_to_gimple::get_def_bb_for_const,
12887 translate_isl_ast_to_gimple::get_new_name,
12888 translate_isl_ast_to_gimple::collect_all_ssa_names,
12889 translate_isl_ast_to_gimple::copy_loop_phi_args,
12890 translate_isl_ast_to_gimple::collect_all_ssa_names,
12891 translate_isl_ast_to_gimple::copy_loop_phi_args,
12892 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12893 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12894 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12895 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12896 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12897 translate_isl_ast_to_gimple::copy_cond_phi_args,
12898 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12899 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12900 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12901 translate_isl_ast_to_gimple::rename_uses,
12902 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12903 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12904 (set_rename_for_each_def): Likewise.
12905 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12906 here. Handle rewriting SCEV analyzable uses here.
12907 (copy_bb_and_scalar_dependences): Generate code for PHI
12909 (graphite_regenerate_ast_isl): Adjust.
12910 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12911 (add_write, add_read): New functions.
12912 (build_cross_bb_scalars_def): Use it and simplify.
12913 (build_cross_bb_scalars_use): Likewise.
12914 (graphite_find_cross_bb_scalar_vars): Inline into...
12915 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12916 simulating out-of-SSA. Compute liveout and add dependencies.
12917 (build_scops): Force an empty entry block.
12918 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12920 (sese_build_liveouts): Declare.
12921 (sese_trivially_empty_bb_p): Likewise.
12922 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12923 compute liveout and debug_liveout.
12924 (sese_bad_liveouts_use): Remove.
12925 (sese_reset_debug_liveouts_bb): Likewise.
12926 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12927 (sese_build_liveouts): Build liveout and debug_liveout and store
12929 (new_sese_info): Adjust.
12930 (free_sese_info): Likewise.
12931 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12932 do not build liveout here.
12933 (move_sese_in_condition): Adjust region entry.
12934 (scev_analyzable_p): Match up with chrec_apply requirements.
12935 (sese_trivially_empty_bb_p): New.
12936 * tree-into-ssa.c (get_reaching_def): Properly support generating
12937 default-defs for incremental rewrite of anonymous names.
12939 2017-10-06 Richard Biener <rguenther@suse.de>
12941 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12942 precision do not perform modulo reduction.
12944 2017-10-06 Richard Biener <rguenther@suse.de>
12946 PR tree-optimization/82436
12947 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12948 conservatively choose the vectorization factor when checking
12949 whether we can perform the required load permutation.
12950 (vect_transform_slp_perm_load): Assert when we may not fail.
12952 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12954 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12955 message for incompatible -msdata=* and -mcall-* options.
12957 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12959 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12960 rate for post-reload scheduling.
12962 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12964 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12966 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12968 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12969 to improve monte carlo in scimark.
12971 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12973 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12974 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12975 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12976 pentium4_cost, nocona_cost): Set reassociation width to 1.
12977 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12978 width to 2 for fp operations and 1 otherwise.
12979 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12981 (atom_cost): Set reassociation width to 2.
12982 (slm_cost, generic_cost): Set fp reassociation width
12983 to 2 and 1 otherwise.
12984 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12985 (core_cost): Set fp reassociation width to 4 and vector to 2.
12986 (ix86_reassociation_width): Rewrite using cost table; special case
12987 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12988 and TARGET_AVX128_OPTIMAL.
12989 * config/i386/i386.h (processor_costs): Add
12990 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12991 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12992 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12993 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12994 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12995 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12997 2017-10-05 Nathan Sidwell <nathan@acm.org>
12999 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13001 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13003 * config/arm/arm.c (arm_test_fpu_data): New.
13004 (arm_run_selftests): Call arm_test_fpu_data.
13006 2017-10-04 Nathan Sidwell <nathan@acm.org>
13008 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13010 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13012 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13014 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13015 check type for aarch64_simd_valid_immediate.
13016 (aarch64_output_simd_mov_immediate): Update prototype.
13017 (aarch64_simd_valid_immediate): Update prototype.
13018 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13019 support for ORR-immediate.
13020 (and<mode>3): modified pattern to add support for BIC-immediate.
13021 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13022 now checks for valid immediate for BIC and ORR based on new enum
13024 (aarch64_output_simd_mov_immediate): Function now used to output
13025 BIC/ORR imm as well based on new enum argument.
13026 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13028 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13029 (aarch64_reg_or_bic_imm): Likewise.
13031 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13033 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13034 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13036 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13040 PR rtl-optimization/82396
13041 * haifa-sched.c (autopref_multipass_init): Simplify
13043 (autopref_rank_data): Simplify sort order.
13044 * sched-int.h (autopref_multipass_data_): Remove
13045 multi_mem_insn_p, min_offset and max_offset.
13047 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13049 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13051 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13053 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13054 vect_doubleint_cvt.
13056 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13058 * doc/sourcebuild.texi: Document vect_long_mult.
13060 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13062 PR tree-optimization/82413
13063 * fold-const.c (build_range_check): Use widest_int when comparing
13064 the maximum ETYPE value with HIGH.
13066 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13068 PR rtl-optimization/82396
13069 * haifa-sched.c (autopref_multipass_init): Simplify
13071 (autopref_rank_data): Simplify sort order.
13072 * sched-int.h (autopref_multipass_data_): Remove
13073 multi_mem_insn_p, min_offset and max_offset.
13075 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13077 PR tree-optimization/82381
13078 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13079 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13082 PR tree-optimization/82374
13083 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13084 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13085 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13086 current_function_decl to the new decl.
13088 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13090 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13091 helper macro for IEEE float128 hardware built-in functions.
13092 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13094 (TRUNCF128_ODD): Likewise.
13095 (ADDF128_ODD): Likewise.
13096 (SUBF128_ODD): Likewise.
13097 (MULF128_ODD): Likewise.
13098 (DIVF128_ODD): Likewise.
13099 (FMAF128_ODD): Likewise.
13100 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13101 UNSPEC_TRUNC_ROUND_TO_ODD.
13102 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13103 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13104 floating point round to odd instructions.
13105 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13106 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13107 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13108 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13109 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13110 (trunc<mode>sf2_hw): Change the truncate with round to odd
13111 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13112 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13113 to odd hardware instructions.
13114 (sub<mode>3_odd): Likewise.
13115 (mul<mode>3_odd): Likewise.
13116 (div<mode>3_odd): Likewise.
13117 (sqrt<mode>2_odd): Likewise.
13118 (fma<mode>4_odd): Likewise.
13119 (fms<mode>4_odd): Likewise.
13120 (nfma<mode>4_odd): Likewise.
13121 (nfms<mode>4_odd): Likewise.
13122 (trunc<mode>df2_odd): Change the truncate with round to odd
13123 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13125 * doc/extend.texi (PowerPC built-in functions): Update documentation
13126 for existing IEEE float128-bit built-in functions. Add built-in
13127 functions that generate the IEEE 128-bit floating point round to
13130 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13132 PR rtl-optimization/77729
13133 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13134 to (X&(C1&~C2))|C2 transformations.
13136 2017-10-03 Martin Jambor <mjambor@suse.cz>
13138 PR tree-optimization/82363
13139 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13140 mismatch, mark lacc written regardless of racc.
13142 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13144 PR tree-optimization/82381
13145 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13146 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13147 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13148 fallthrough into reassoc_stmt_dominates_stmt_p.
13151 * combine.c (combine_instructions): Don't combine in unreachable
13154 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13157 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13158 function to not use the have_cpu variable. Do not set cpu_index,
13159 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13160 or the default cpu.
13161 (rs6000_valid_attribute_p): Remove duplicate initializations of
13162 old_optimize and func_optimize.
13163 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13164 (rs6000_activate_target_options): Make global.
13165 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13168 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13170 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13171 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13172 Set *poffset to 0 and *psize and *pmax_size to -1 if
13173 *poffset + *psize overflows in HOST_WIDE_INT.
13175 PR tree-optimization/82387
13176 PR tree-optimization/82388
13177 PR tree-optimization/82389
13178 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13179 instead of live_bytes non-NULL.
13181 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13184 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13187 2017-10-02 Richard Biener <rguenther@suse.de>
13189 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13190 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13191 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13193 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13195 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13196 requested precision matches the type's.
13197 * calls.c (alloc_max_size): Calculate the new candidate size as
13198 a widest_int and use wi::to_widest when comparing it with the
13199 current candidate size.
13200 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13201 zero rather than integer_zero_node.
13202 * match.pd: Check for a no-op conversion before using wi::add
13203 rather than after. Use tree_to_uhwi when summing small shift
13204 counts into an unsigned int.
13206 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13207 Alan Hayward <alan.hayward@arm.com>
13208 David Sherwood <david.sherwood@arm.com>
13211 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13212 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13213 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13214 POINTER_AND_FP_REGS.
13216 2017-10-02 Richard Biener <rguenther@suse.de>
13218 PR tree-optimization/82355
13219 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13220 a mapping for the enclosing loop but avoid generating one for
13221 the loop tree root.
13222 (copy_bb_and_scalar_dependences): Remove premature codegen
13223 error on PHIs in blocks duplicated into multiple places.
13224 * graphite-scop-detection.c
13225 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13226 in the region use it as loop and nest to analyze the DR in.
13227 (try_generate_gimple_bb): Likewise.
13228 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13229 (add_loop_constraints): For blocks in a loop not in the region
13230 create a dimension with a single iteration.
13231 * sese.h (gbb_loop_at_index): Remove assert.
13233 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13235 * omp-expand.c (adjust_context_scope): New function.
13236 (expand_parallel_call): Call adjust_context_scope.
13238 2017-10-01 Jeff Law <law@redhat.com>
13240 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13241 dom_opt_dom_walker class with direct access to private members.
13242 Add comments. Call test_for_singularity.
13243 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13244 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13245 m_dummy_cond anymore.
13246 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13248 (pass_dominator:execute): Build the dummy_cond here and pass it
13249 to the dom_opt_dom_walker ctor.
13250 (test_for_singularity): New function.
13252 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13253 Maya Rashish <coypu@sdf.org>
13255 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13256 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13257 (alpha*-*-netbsd*) Use nbsd_tm_file.
13258 (arm*-*-netbsdelf*) Likewise.
13259 (i[34567]86-*-netbsdelf*) Likewise.
13260 (x86_64-*-netbsd*) Likewise.
13261 (mips*-*-netbsd*) Likewise.
13262 (powerpc-*-netbsd*) Likewise.
13263 (sh*-*-netbsd*) Likewise.
13264 (sparc-*-netbsdelf*) Likewise.
13265 (sparc64-*-netbsd*) Likewise.
13266 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13268 (vax-*-netbsdelf*) Likewise.
13269 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13270 (UINT_FAST8_TYPE) Likewise.
13271 (INT_FAST16_TYPE) Check CHAR_FAST16.
13272 (UINT_FAST16_TYPE) Likewise.
13274 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13277 * config/i386/i386.md
13278 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13279 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13280 *divmodsi4_zext_2): New define_insn_and_split.
13281 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13282 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13283 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13284 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13285 New define_insn_and_split.
13286 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13287 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13288 operands[1] having DImode when mode is SImode.
13290 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13291 always SImode for DIV and MOD in REG_EQUAL notes.
13293 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13295 PR middle-end/82319
13296 * match.pd: Fix handling of NaNs in pattern.
13298 2017-09-29 Jeff Law <law@redhat.com>
13300 * sbitmap.c (bitmap_bit_in_range_p): New function.
13301 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13302 * tree-ssa-dse.c (live_bytes_read): New function.
13303 (dse_classify_store): Ignore reads of dead bytes.
13305 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13306 typos and whitespace errors.
13307 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13308 * config/s390/s390.c (s390_emit_prologue): Likewise.
13310 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13313 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13314 with a symbol for LRA.
13316 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13318 PR rtl-optimization/82338
13319 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13321 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13323 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13324 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13325 (qsort_chk): Declare.
13326 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13327 (qsort_chk): New function.
13329 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13331 PR tree-optimization/82337
13332 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13333 phi definition if the PHI result appears in an abnormal PHI.
13334 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13335 basis appears in an abnormal PHI.
13337 2017-09-29 Richard Biener <rguenther@suse.de>
13339 * graphite-isl-ast-to-gimple.c
13340 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13341 (binary_op_to_tree): Use it.
13342 (get_rename_from_scev): Likewise.
13343 (copy_loop_phi_nodes): Likewise.
13344 (copy_bb_and_scalar_dependences): Likewise.
13345 (translate_pending_phi_nodes): Likewise.
13347 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13350 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13351 for movabsq $(i32 << shift), r64.
13353 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13355 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13356 index when encoding %esp as %rsp to avoid 0x67 prefix.
13358 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13360 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13361 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13363 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13365 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13366 Extensions with more than 16 double VFP registers.
13367 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13368 to_clear_mask and all code related to it. Replace the remaining
13369 entry by a sbitmap and adapt code accordingly.
13371 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13373 * brig-builtins.def: Change pure attributes to const.
13375 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13377 * config.gcc (default_gnu_indirect_function): Default to yes for
13378 sparc*-*-linux* with glibc.
13380 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13382 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13383 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13384 when creating .init_array and .fini_array sections with priority
13387 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13390 * config/aarch64/aarch64.c
13391 (aarch64_builtin_support_vector_misalignment): Always return false
13392 when misalignment is unknown.
13394 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13396 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13397 this function to return false if the definition used by the swap
13398 instruction is artificial, or if the memory address from which the
13399 constant value is loaded is not represented by a base address held
13400 in a register or if the base address register is a frame or stack
13401 pointer. Additionally, return false if the base address of the
13402 loaded constant is a SYMBOL_REF but is not considered to be a
13404 (replace_swapped_load_constant): New function.
13405 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13406 loaded constant vector with a load of a swapped constant vector.
13408 2017-09-27 Carl Love <cel@us.ibm.com>
13410 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13411 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13412 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13415 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13417 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13418 first, always call autopref_rank_data otherwise.
13420 2017-09-27 Richard Biener <rguenther@suse.de>
13422 * graphite-scop-detection.c (find_scop_parameters): Move
13423 loop bound handling ...
13424 (gather_bbs::before_dom_children): ... here, avoiding the need
13425 to build scop_info->loop_nest.
13426 (record_loop_in_sese): Remove.
13427 * sese.h (sese_info_t::loop_nest): Remove.
13428 * sese.c (new_sese_info): Do not allocate loop_nest.
13429 (free_sese_info): Do not free loop_nest.
13431 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13434 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13435 lhs from calls if the lhs has addressable type.
13437 2017-09-27 Richard Biener <rguenther@suse.de>
13439 * graphite.h (scop::max_alias_set): New member.
13440 * graphite-scop-detection.c: Remove references to non-existing
13441 --param in comments.
13442 (build_alias_sets): Record the maximum alias set used for drs.
13443 (build_scops): Support zero as unlimited for
13444 --param graphite-max-arrays-per-scop.
13445 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13446 and inline into ...
13447 (build_poly_sr_1): ... here. Compute alias set based on the
13448 maximum alias set used for drs rather than
13449 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13451 2017-09-27 Richard Biener <rguenther@suse.de>
13453 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13454 --param loop-block-tile-size=0 to disable tiling.
13456 2017-09-27 Richard Biener <rguenther@suse.de>
13458 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13459 (graphite-max-nb-scop-params): Document special value zero.
13460 * domwalk.h (dom_walker::STOP): New symbolical constant.
13461 (dom_walker::dom_walker): Add optional parameter for bb to
13463 (dom_walker::~dom_walker): Declare.
13464 (dom_walker::before_dom_children): Document STOP return value.
13465 (dom_walker::m_user_bb_to_rpo): New member.
13466 (dom_walker::m_bb_to_rpo): Likewise.
13467 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13468 mapping here if not provided by the user.
13469 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13470 provided by the user.
13471 (dom_walker::STOP): Define.
13472 (dom_walker::walk): Do not compute bb to RPO mapping here.
13473 Support STOP return value from before_dom_children to stop
13475 * graphite-optimize-isl.c (optimize_isl): If the schedule
13476 is the same still generate code if -fgraphite-identity
13477 or -floop-parallelize-all are given.
13478 * graphite-scop-detection.c: Include cfganal.h.
13479 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13481 (gather_bbs::before_dom_children): Return STOP for BBs
13483 (build_scops): Compute bb to RPO mapping and pass it to
13484 the domwalk. Treat --param graphite-max-nb-scop-params=0
13485 as not limiting the number of params.
13486 * graphite.c (graphite_initialize): Remove limit on the
13487 number of basic-blocks in a function.
13488 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13489 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13490 default value of 10.
13492 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13494 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13495 Adjust code to eliminate needing to do the shift right 32-bits
13496 operation after XSCVDPSPN.
13498 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13500 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13501 ((X / Y) != 0 -> X >= Y): Likewise.
13503 2017-09-26 Carl Love <cel@us.ibm.com>
13505 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13506 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13507 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13508 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13509 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13510 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13511 definitions and overloading.
13512 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13513 statement for P9V_BUILTIN_XST_LEN_R.
13514 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13515 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13516 define_expand and define_insn for the instructions and builtins.
13517 * doc/extend.texi: Update the built-in documentation file for the new
13518 built-in functions.
13519 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13520 define_insn for the instructions
13522 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13525 * gcc/config/netbsd-protos.h: New file.
13526 * gcc/config/netbsd.c: New file.
13527 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13528 * gcc/config/t-netbsd: New file.
13529 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13530 (tmake_file) Add t-netbsd.
13531 (extra_objs) Add netbsd.o.
13533 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13537 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13539 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13541 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13542 sign extension from a vector register to a GPR by doing a 32-bit
13543 direct move and then an EXTSW.
13544 (extendsi<mode>2 splitter): Likewise.
13545 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13546 right or vector extract after doing XSCVDPSPN. Use
13547 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13549 (movdi_from_sf_zero_ext): Likewise.
13550 (reload_gpr_from_vsxsf): Likewise.
13551 (p8_mfvsrd_4_disf): Delete, no longer used.
13552 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13553 then needing to move the SFmode to a GPR to use the XSCVDPSP
13554 instruction instead of FRSP and XSCVDPSPN.
13555 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13556 it is adjacent to the other XSCVSPDP insns.
13557 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13558 SFmode to be in traditional Altivec registers.
13559 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13560 (vsx_xscvspdpn): Likewise.
13561 (vsx_xscvspdpn_scalar): Likewise.
13563 2017-09-26 Martin Jambor <mjambor@suse.cz>
13565 * tree-sra.c (compare_access_positions): Put integral types first,
13566 stabilize sorting of integral types, remove conditions putting
13567 non-full-precision integers last.
13568 (sort_and_splice_var_accesses): Disable scalarization if a
13569 non-integert would be represented by a non-full-precision integer.
13571 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13573 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13574 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13575 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13576 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13577 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13578 conditionals inside the function instead of around it. Call
13579 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13580 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13582 2017-09-26 Richard Biener <rguenther@suse.de>
13584 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13586 (scop_detection::build_scop_breadth): ... this. Removed.
13587 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13588 (scop_detection::harmful_stmt_in_bb): Likewise.
13589 (scop_detection::graphite_can_represent_stmt): Likewise.
13590 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13591 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13592 (scop_detection::can_represent_loop_1): ... this. Removed.
13593 (scop_detection::harmful_loop_in_region): Simplify after inlining
13594 the above and remove more quadraticness.
13595 (build_scops): Adjust.
13596 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13599 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13602 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13603 REGNO (base) == SP_REG if base is a REG.
13605 PR middle-end/35691
13606 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13607 if it is different SSA_NAME.
13608 (optimize_range_tests_cmp_bitwise): New function.
13609 (optimize_range_tests): Call it.
13611 2017-09-26 Richard Biener <rguenther@suse.de>
13613 PR tree-optimization/82321
13614 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13615 for the def being inside the loop.
13617 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13619 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13621 * config/s390/s390-builtins.def: Fix constraint on op4.
13623 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13625 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13626 independent expanders.
13627 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13628 ("vec_ordered", "vec_unordered"): New expanders.
13630 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13632 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13635 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13637 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13638 vec_unpacks_lo_v16qi.
13639 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13641 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13643 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13644 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13645 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13647 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13649 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13651 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13653 ("*vec_slb<mode>"): New insn pattern.
13654 ("vec_shr_<mode>"): New expander.
13655 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13656 and force the shift count operand to V16QImode.
13657 ("vec_srb<mode>"): Set shift count mode to V16QI.
13659 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13661 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13662 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13663 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13665 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13668 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13670 2017-09-26 Richard Biener <rguenther@suse.de>
13672 PR tree-optimization/82320
13673 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13676 2017-09-25 Jeff Law <law@redhat.com>
13678 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13679 prototype for new argument.
13680 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13681 mostly extracted from rs6000_emit_allocate_stack.
13682 (rs6000_emit_probe_stack_range_stack_clash): New function.
13683 (rs6000_emit_allocate_stack): Call
13684 rs6000_emit_probe_stack_range_stack_clash as needed.
13685 (rs6000_emit_probe_stack_range): Add additional argument
13686 to call to gen_probe_stack_range{si,di}.
13687 (output_probe_stack_range): New.
13688 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13689 (output_probe_stack_range_stack_clash): New.
13690 (rs6000_emit_prologue): Emit notes into dump file as requested.
13691 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13692 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13693 Add additional operand and pass it to output_probe_stack_range.
13695 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13697 PR tree-optimization/82163
13698 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13699 (checking_verify_loop_closed_ssa): New parameter.
13700 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13701 (check_loop_closed_ssa_stmt): Delete.
13702 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13703 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13704 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13707 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13709 * brig-builtins.def: Treat HSAIL barrier builtins as
13710 setjmp/longjump style functions.
13712 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13714 * target.def (constant_alignment): New hook.
13715 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13716 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13717 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13718 * doc/tm.texi: Regenerate.
13719 * targhooks.h (default_constant_alignment): Declare.
13720 (constant_alignment_word_strings): Likewise.
13721 * targhooks.c (default_constant_alignment): New function.
13722 (constant_alignment_word_strings): Likewise.
13723 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13724 instead of CONSTANT_ALIGNMENT.
13725 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13726 (force_const_mem): Likewise.
13727 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13728 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13729 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13730 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13731 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13733 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13734 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13735 constant_alignment_word_strings.
13736 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13737 (CONSTANT_ALIGNMENT): Likewise.
13738 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13739 (arm_constant_alignment): New function.
13740 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13741 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13742 constant_alignment_word_strings.
13743 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13744 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13745 constant_alignment_word_strings.
13746 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13747 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13748 (cris_constant_alignment): New function.
13749 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13750 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13751 (epiphany_constant_alignment): New function.
13752 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13753 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13754 constant_alignment_word_strings.
13755 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13756 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13757 constant_alignment_word_strings.
13758 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13759 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13760 constant_alignment_word_strings.
13761 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13762 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13763 * config/i386/i386.c (ix86_constant_alignment): Make static.
13764 Use the same interface as the target hook.
13765 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13766 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13767 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13768 constant_alignment_word_strings.
13769 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13770 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13771 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13772 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13773 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13774 constant_alignment_word_strings.
13775 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13776 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13777 constant_alignment_word_strings.
13778 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13779 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13780 constant_alignment_word_strings.
13781 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13782 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13784 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13785 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13786 * config/mips/mips.c (mips_constant_alignment): New function.
13787 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13788 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13789 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13790 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13791 (mmix_constant_alignment): Make static. Use the same interface
13792 as the target hook.
13793 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13794 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13795 constant_alignment_word_strings.
13796 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13797 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13798 constant_alignment_word_strings.
13799 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13800 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13801 constant_alignment_word_strings.
13802 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13803 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13804 (rs6000_constant_alignment): New function.
13805 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13806 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13807 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13808 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13809 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13810 (rs6000_constant_alignment): New function.
13811 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13812 * config/s390/s390.c (s390_constant_alignment): New function.
13813 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13814 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13815 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13816 constant_alignment_word_strings.
13817 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13818 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13819 (sparc_constant_alignment): New function.
13820 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13821 * config/spu/spu.c (spu_constant_alignment): New function.
13822 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13823 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13824 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13825 constant_alignment_word_strings.
13826 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13827 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13828 constant_alignment_word_strings.
13829 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13830 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13831 constant_alignment_word_strings.
13832 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13833 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13834 (visium_constant_alignment): New function.
13835 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13836 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13837 (xtensa_constant_alignment): New function.
13838 * system.h (CONSTANT_ALIGNMENT): Poison.
13840 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13842 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13843 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13844 (rs6000_builtin_valid_without_lhs): New helper function.
13845 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13846 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13848 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13850 * target.h (vec_perm_indices): Use unsigned short rather than
13852 (auto_vec_perm_indices): Likewise.
13853 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13854 Use unsigned int rather than unsigned char.
13855 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13857 2017-09-25 Richard Biener <rguenther@suse.de>
13859 * cfgloop.h (sort_sibling_loops): Declare.
13860 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13861 (sort_sibling_loops): New function sorting the sibling loop list
13863 * graphite.c (graphite_transform_loops): Sort sibling loops.
13865 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13867 * target.def (vec_perm_const_ok): Change sel parameter to
13869 * optabs-query.c (can_vec_perm_p): Update accordingly.
13870 * doc/tm.texi: Regenerate.
13871 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13872 auto_vec_perm_indices and remove separate nelt field.
13873 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13874 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13875 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13876 (aarch64_expand_vec_perm_const): Update accordingly.
13877 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13878 to vec_perm_indices.
13879 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13880 auto_vec_perm_indices and remove separate nelt field.
13881 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13882 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13883 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13885 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13886 to vec_perm_indices.
13887 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13888 sel to vec_perm_indices.
13889 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13890 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13891 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13893 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13896 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13899 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13900 on the FUNCTION_DECL function context if it has a DIE that is a
13903 2017-09-25 Richard Biener <rguenther@suse.de>
13905 PR tree-optimization/82285
13906 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13909 2017-09-25 Tom de Vries <tom@codesourcery.com>
13913 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13916 2017-09-25 Richard Biener <rguenther@suse.de>
13918 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13919 ISL errors other than isl_error_quota happen. Dump if the
13920 schedule is the same.
13921 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13922 errors instead of aborting inside ISL.
13924 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13927 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13928 of libgcc_eh for m64.
13929 * config/i386/darwin64.h: Likewise.
13931 2017-09-25 Richard Biener <rguenther@suse.de>
13933 PR middle-end/82144
13934 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13935 attribute for incomplete types nor twice for complete ones.
13937 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13940 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13941 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13944 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13947 * config/i386/i386.opt (mprefer-avx256): Use
13948 ix86_target_flags variable.
13949 * config/i386/i386.c (ix86_target_string): Move
13950 -mprefer-avx256 to flag2_opts.
13952 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13954 PR middle-end/35691
13955 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13956 and x != -1 | y != -1 into (x & y) != -1.
13958 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13960 * config.gcc: Add new case statement to set
13961 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13962 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13963 s390x-*-linux* case statements. Added aarch64 to the list of
13964 supported architectures.
13966 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13968 PR tree-optimization/82289
13969 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13970 STMT_VINFO_RELEVANT_P.
13972 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13973 Alan Hayward <alan.hayward@arm.com>
13974 David Sherwood <david.sherwood@arm.com>
13976 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13977 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13978 (extract_range_from_binary_expr_1): Don't call
13979 extract_range_from_multiplicative_op_1 if !range_int_cst_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 * target.def (preferred_vector_alignment): New hook.
13986 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13988 * doc/tm.texi: Regenerate.
13989 * targhooks.h (default_preferred_vector_alignment): Declare.
13990 * targhooks.c (default_preferred_vector_alignment): New function.
13991 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13993 (DR_TARGET_ALIGNMENT): New macro.
13994 (aligned_access_p): Update commentary.
13995 (vect_known_alignment_in_bytes): New function.
13996 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13998 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13999 Calculate the misalignment based on the target alignment rather than
14001 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14002 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14003 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14004 the target alignment, rather than masking the element misalignment
14005 with the number of elements in a vector. Also use the target
14006 alignment when calculating the maximum number of peels.
14007 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14008 instead of TYPE_ALIGN_UNIT.
14009 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14010 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14011 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14012 (vect_create_data_ref_ptr): Likewise.
14013 (vect_setup_realignment): Realign by ANDing with
14014 -DR_TARGET_MISALIGNMENT.
14015 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14016 the number of peels based on DR_TARGET_ALIGNMENT.
14017 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14018 with the guaranteed alignment boundary when deciding whether
14020 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14021 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14022 (ensure_base_align): Remove stmt_info parameter. Get the
14023 target base alignment from DR_TARGET_ALIGNMENT.
14024 (vectorizable_store): Update call accordingly. Interpret
14025 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14027 (vectorizable_load): Likewise.
14029 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14030 Alan Hayward <alan.hayward@arm.com>
14031 David Sherwood <david.sherwood@arm.com>
14033 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14034 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14035 (vect_enhance_data_refs_alignment): Likewise.
14037 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14039 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14040 error. Only quit immediately if parsing is complete.
14041 (BEGIN): Initialize fatal_err and parse_done.
14042 (begin fpu, end fpu): Check number of arguments.
14043 (begin arch, end arch): Likewise.
14044 (begin cpu, end cpu): Likewise.
14045 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14046 (optalias): Likewise.
14048 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14050 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14051 * config/arm/arm-isa.h: Delete. Move definitions to ...
14052 * arm-cpus.in: ... here. Use new feature and fgroup values.
14053 * config/arm/arm.c (arm_option_override): Use lower case for feature
14055 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14056 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14057 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14059 (print_isa_bits_for): New function.
14060 (gen_isa): New function.
14061 (gen_comm_data): Use print_isa_bits_for.
14062 (define feature): New keyword.
14063 (define fgroup): New keyword.
14064 * config/arm/t-arm (TM_H): Remove.
14065 (GTM_H): Add arm-isa.h.
14066 (arm-isa.h): Add rule to generate file.
14067 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14068 case for feature bit names.
14070 2017-09-22 Richard Biener <rguenther@suse.de>
14072 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14074 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14075 print of no dependency loops ...
14076 * graphite.c (graphite_transform_loops): ... here.
14077 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14079 (same_close_phi_node, remove_duplicate_close_phi,
14080 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14081 (canonicalize_loop_closed_ssa): ... here and simplify.
14082 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14083 (optimize_isl): Use dump_printf_loc to tell when we stopped
14084 optimizing because of an ISL timeout.
14086 2017-09-22 Richard Biener <rguenther@suse.de>
14088 PR tree-optimization/82291
14089 * tree-if-conv.c (predicate_mem_writes): Make sure to
14090 remove writes in blocks predicated with false.
14092 2017-09-22 Richard Biener <rguenther@suse.de>
14094 * sese.c: Include cfganal.h.
14095 (if_region_set_false_region): Remove.
14096 (create_if_region_on_edge): Likewise.
14097 (move_sese_in_condition): Re-implement without destroying
14100 2017-09-22 Richard Biener <rguenther@suse.de>
14102 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14103 Verify both BBs contain loop PHI nodes before dispatching to
14104 copy_loop_phi_args.
14105 (graphite_regenerate_ast_isl): Do not recompute dominators,
14106 do not verify three times. Restructure for clarity.
14107 * graphite-scop-detection.c (same_close_phi_node,
14108 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14109 defined_in_loop_p, canonicalize_loop_closed_ssa,
14110 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14111 checking and SSA rewrite, move to ...
14112 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14113 (graphite_initialize): Do not pass in ctx, do not reset the
14114 SCEV cache, compute only dominators.
14115 (graphite_transform_loops): Allocate ISL ctx after
14116 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14117 Maintain post-dominators only around build_scops.
14118 * sese.c (if_region_set_false_region): Make static. Free
14119 and recompute dominators.
14120 (move_sese_in_condition): Assert we don't get called with
14121 post-dominators computed.
14122 * sese.h (if_region_set_false_region): Remove.
14124 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14126 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14127 mode attribute for TARGET_AVX512VL.
14129 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14131 * config/i386/i386.opt (mprefer-avx256): New option.
14132 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14134 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14135 for TARGET_PREFER_AVX256.
14136 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14138 2017-09-21 Jeff Law <law@redhat.com>
14140 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14141 Fix dump output if the only stack space is for pushed registers.
14143 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14145 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14148 2017-09-21 Martin Sebor <msebor@redhat.com>
14151 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14152 code (in C++) or code that triggers warnings.
14154 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14156 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14158 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14160 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14161 * sched-rgn.c: Ditto.
14162 * sel-sched-ir.c: Ditto.
14164 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14166 * toplev.h (set_random_seed): Adjust return type.
14167 * toplev.c (init_local_tick): Move eager initialization of random_seed
14168 to get_random_seed. Adjust comment.
14169 (init_random_seed): Inline to get_random_seed, delete.
14170 (get_random_seed): Initialize random_seed lazily.
14171 (set_random_seed): Do not return previous value.
14172 (print_switch_value): Do not call get_random_seed.
14174 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14176 * cgraph.c (delete_function_version): New, broken out from...
14177 (cgraph_node::delete_function_version): ...here. Rename to
14178 cgraph_node::delete_function_version_by_decl. Update all uses.
14179 (cgraph_node::remove): Call delete_function_version.
14181 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14184 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14185 VAR_DECLs to which addressable non-volatile parameters are mapped
14186 and for id->retvar after the return value assignment. Clear
14187 id->retval and id->retbnd after inlining.
14189 2017-09-21 Richard Biener <rguenther@suse.de>
14191 PR tree-optimization/82276
14192 PR tree-optimization/82244
14193 * tree-vrp.c (build_assert_expr_for): Set
14194 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14196 (remove_range_assertions): Revert earlier change.
14198 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14201 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14203 2017-09-21 Richard Biener <rguenther@suse.de>
14205 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14206 Restore valid IL after code generation errors.
14207 * graphite.c (graphite_transform_loops): Diagnose code
14208 generation issues as MSG_MISSED_OPTIMIZATION and continue
14209 with processing SCOPs.
14211 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14212 Alan Hayward <alan.hayward@arm.com>
14213 David Sherwood <david.sherwood@arm.com>
14215 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14216 rather than choosing between plus_constant and gen_rtx_<CODE>.
14217 * expr.c (emit_push_insn): Likewise.
14218 (expand_expr_real_2): Likewise.
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 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14226 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14227 Alan Hayward <alan.hayward@arm.com>
14228 David Sherwood <david.sherwood@arm.com>
14230 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14231 calling tree_to_uhwi.
14233 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14234 Alan Hayward <alan.hayward@arm.com>
14235 David Sherwood <david.sherwood@arm.com>
14237 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14238 INTEGER_CST rather than a negative test for ADDR_EXPR.
14240 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14241 Alan Hayward <alan.hayward@arm.com>
14242 David Sherwood <david.sherwood@arm.com>
14244 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14245 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14247 2017-09-21 Richard Biener <rguenther@suse.de>
14249 PR tree-optimization/71351
14250 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14251 graphite_create_new_loop_guard): Remove, fold remaining parts
14253 (translate_isl_ast_node_for): ... here and simplify.
14255 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14258 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14259 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14260 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14261 alternative always use QI mode, for -Os imov (=R,R) alternative
14262 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14265 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14266 Jeff Law <law@redhat.com>
14268 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14269 (allocate_stack_space): New function, partially extracted from
14270 s390_emit_prologue.
14271 (s390_emit_prologue): Track offset to most recent stack probe.
14272 Code to allocate space moved into allocate_stack_space.
14273 Dump actions when no stack is allocated.
14274 (s390_prologue_plus_offset): New function.
14275 (s390_emit_stack_probe): Likewise.
14277 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14279 * common.opt (Wa, Wl, Wp, g, gz=): Add
14281 (gno-column-info): Remove.
14282 (gcolumn-info): Drop RejectNegative.
14283 (gno-): New prefix.
14284 (gno-record-gcc-switches): Remove.
14285 (grecord-gcc-switches): Drop RejectNegative.
14286 (gno-split-dwarf): Remove.
14287 (gsplit-dwarf): Drop RejectNegative.
14288 (gno-strict-dwarf): Remove.
14289 (gstrict-dwarf): Drop RejectNegative.
14290 * config/darwin.opt (gfull, gused): Add RejectNegative.
14291 * dwarf2out.c (gen_producer_string): Drop
14292 gno-record-gcc-switches handler.
14293 * optc-gen.awk: Add g to prefixes with negative forms.
14294 * opts-common.c (remapping_prefix_p): New.
14295 (find_opt): Check it.
14296 (generate_canonical_option): Test g prefix.
14297 (option_map): Add -gno- mapping.
14298 (add_misspelling_candidates): Check remapping_prefix_p.
14300 2017-09-20 Jeff Law <law@redhat.com>
14302 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14303 thinko in stack clash protection support.
14305 * explow.c (compute_stack_clash_protection_loop_data): Use
14306 CONST_INT_P instead of explicit test. Verify object is a
14307 CONST_INT_P before looking at INTVAL.
14308 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14309 instead of explicit test.
14311 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14314 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14315 address instead of to r1 and r11.
14317 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14319 * config.gcc: Support "knm".
14320 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14321 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14323 * config/i386/i386.c (m_KNM): Define.
14324 (processor_target_table): Add "knm".
14326 (ix86_option_override_internal): Add "knm".
14327 (ix86_issue_rate): Add PROCESSOR_KNM.
14328 (ix86_adjust_cost): Ditto.
14329 (ia32_multipass_dfa_lookahead): Ditto.
14330 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14331 (fold_builtin_cpu): Add M_INTEL_KNM.
14332 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14333 (processor_type): Add PROCESSOR_KNM.
14334 * config/i386/x86-tune.def: Add m_KNM.
14335 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14337 2017-09-20 Richard Biener <rguenther@suse.de>
14339 PR tree-optimization/80213
14340 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14341 are allowed in empty BBs as well.
14342 (canonicalize_loop_closed_ssa): Also look for other complex
14344 (scop_detection::get_sese): Include the loop-closed PHI block
14346 (scop_detection::merge_sese): Remove code adding extra blocks.
14347 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14348 (build_scops): Assert the final returned scop is invalid.
14350 2017-09-20 Richard Biener <rguenther@suse.de>
14352 PR tree-optimization/82264
14353 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14355 (vn_phi_lookup): Likewise.
14356 (vn_phi_insert): Likewise.
14358 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14360 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14361 that fits into uhwi or shwi, add DW_AT_const_value regardless
14362 of early_dwarf without going through RTL, using add_AT_unsigned
14365 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14366 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14367 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14368 (DEBUG_MACRO_SECTION): Likewise.
14369 (DEBUG_LTO_MACRO_SECTION): Likewise.
14370 (DEBUG_STR_DWO_SECTION): Likewise.
14371 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14372 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14373 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14374 (DEBUG_LTO_LINE_STR_SECTION): Define.
14375 (init_sections_and_labels): Initialize debug_line_str_section
14376 variable. Initialize debug_loc_section for -gdwarf-5 to
14377 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14379 2017-09-20 Richard Biener <rguenther@suse.de>
14381 * graphite-sese-to-poly.c (extract_affine): Properly handle
14382 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14384 2017-09-20 Richard Biener <rguenther@suse.de>
14386 PR tree-optimization/81373
14387 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14388 Force SESE live-out defs to be handled even if they are
14391 2017-09-19 Jeff Law <law@redhat.com>
14393 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14394 nothing for stack adjustments with REG_STACK_CHECK.
14395 * sched-deps.c (parse_add_or_inc): Reject insns with
14396 REG_STACK_CHECK from dependency breaking.
14397 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14398 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14399 * reg-notes.def (STACK_CHECK): New note.
14401 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14402 (ix86_expand_prologue): Dump stack clash info as needed.
14403 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14405 * function.c (dump_stack_clash_frame_info): New function.
14406 * function.h (dump_stack_clash_frame_info): Prototype.
14407 (enum stack_clash_probes): New enum.
14409 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14410 flag_stack_clash_protection.
14411 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14412 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14413 (arm_frame_pointer_required): Likewise.
14414 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14415 (ia64_expand_prologue): Likewise.
14416 * config/mips/mips.c (mips_expand_prologue): Likewise.
14417 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14418 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14419 (sparc_flat_expand_prologue): Likewise.
14420 * config/spu/spu.c (spu_expand_prologue): Likewise.
14422 * explow.c: Include "params.h".
14423 (anti_adjust_stack_and_probe_stack_clash): New function.
14424 (get_stack_check_protect): Likewise.
14425 (compute_stack_clash_protection_loop_data): Likewise.
14426 (emit_stack_clash_protection_loop_start): Likewise.
14427 (emit_stack_clash_protection_loop_end): Likewise.
14428 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14429 Use anti_adjust_stack_and_probe_stack_clash.
14430 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14431 (emit_stack_clash_protection_loop_start): Likewise.
14432 (emit_stack_clash_protection_loop_end): Likewise.
14433 * rtl.h (get_stack_check_protect): Prototype.
14434 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14435 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14436 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14438 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14440 * doc/tm.texi: Rebuilt.
14441 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14442 get_stack_check_protect.
14443 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14444 * config/arm/arm.c (arm_expand_prologue): Likewise.
14445 (arm_frame_pointer_required): Likewise.
14446 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14447 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14448 * config/mips/mips.c (mips_expand_prologue): Likewise.
14449 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14450 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14451 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14452 (sparc_flat_expand_prologue): Likewise.
14454 * common.opt (-fstack-clash-protection): New option.
14455 * flag-types.h (enum stack_check_type): Note difference between
14456 -fstack-check= and -fstack-clash-protection.
14457 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14458 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14459 * toplev.c (process_options): Issue warnings/errors for cases
14460 not handled with -fstack-clash-protection.
14461 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14462 (-fstack-check): Note additional problem with -fstack-check=generic.
14463 Note that -fstack-check is primarily for Ada and refer users
14464 to -fstack-clash-protection for stack-clash-protection.
14465 Document new params for stack clash protection.
14467 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14469 * config/i386/i386.c (ix86_split_long_move): Do not handle
14470 address used for LEA in a special way.
14472 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14474 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14476 2017-09-19 Martin Sebor <msebor@redhat.com>
14479 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14480 of incompatible types.
14482 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14484 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14485 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14486 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14487 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14489 2017-09-19 Richard Biener <rguenther@suse.de>
14491 PR tree-optimization/82244
14492 * tree-vrp.c (remove_range_assertions): Do not propagate
14493 a constant to abnormals but replace the assert with a copy.
14495 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14497 PR rtl-optimization/57878
14498 PR rtl-optimization/68988
14499 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14500 avoidance test involving non_reload_pseudos. Move frequency test
14501 below the general fragmentation avoidance test.
14503 2017-09-19 Richard Biener <rguenther@suse.de>
14505 PR tree-optimization/69728
14506 * graphite-sese-to-poly.c (schedule_error): New global.
14507 (add_loop_schedule): Handle empty domain by failing the
14509 (build_original_schedule): Handle schedule_error.
14511 2017-09-19 Richard Biener <rguenther@suse.de>
14513 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14514 Do not iterate to sibling loops but only to siblings of inner
14517 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14520 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14521 that the registers are different.
14523 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14525 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14526 to processor_model and "amdfam17h" to arch_names_table.
14527 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14529 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14532 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14534 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14535 Alan Hayward <alan.hayward@arm.com>
14536 David Sherwood <david.sherwood@arm.com>
14538 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14540 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14542 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14543 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14544 vect_schedule_slp_instance.
14545 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14546 Update call to vect_slp_analyze_node_operations. Simplify return
14548 (vect_slp_analyze_bb_1): Update call accordingly.
14549 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14550 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14551 (vect_schedule_slp): Update call accordingly.
14553 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14554 Alan Hayward <alan.hayward@arm.com>
14555 David Sherwood <david.sherwood@arm.com>
14557 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14558 with types that aren't in fact scalar.
14560 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14562 * tree-vect-slp.c (vect_record_max_nunits): New function,
14564 (vect_build_slp_tree_1): ...here.
14565 (vect_build_slp_tree_2): Call it for phis too.
14567 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14569 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14570 to vect_get_vec_def_for_operand when getting the mask operand.
14572 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14573 Alan Hayward <alan.hayward@arm.com>
14574 David Sherwood <david.sherwood@arm.com>
14576 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14579 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14580 Alan Hayward <alan.hayward@arm.com>
14581 David Sherwood <david.sherwood@arm.com>
14583 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14584 calculation for vector booleans.
14586 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14587 Alan Hayward <alan.hayward@arm.com>
14588 David Sherwood <david.sherwood@arm.com>
14590 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14592 (vect_transform_stmt): ...here.
14593 (vect_analyze_stmt): Use it instead of calling
14594 vectorizable_live_operation directly.
14596 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14598 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14599 non-SIMT targets in acc vector loops.
14601 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14603 * configure.ac: Add arc and check if assembler supports gdwarf2.
14604 * configure: Regenerate.
14606 2017-09-18 Richard Biener <rguenther@suse.de>
14608 PR tree-optimization/82220
14609 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14610 epilogue niters from the min_profitable_iters compute.
14612 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14615 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14616 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14617 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14619 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14622 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14623 switching to a new text section.
14625 2017-09-18 Richard Biener <rguenther@suse.de>
14627 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14629 (build_alias_set): Reject aliases with no access function.
14631 2017-09-18 Richard Biener <rguenther@suse.de>
14633 PR tree-optimization/79622
14634 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14636 (build_cross_bb_scalars_use): Likewise.
14638 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14640 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14642 2017-09-18 Alan Modra <amodra@gmail.com>
14645 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14646 stack_pointer_rtx for count 0. Update comments. Break up
14647 large rtl expression.
14649 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14651 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14652 Increase to 20 bytes.
14653 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14654 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14655 or avx version of the stub.
14657 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14660 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14661 compute the minimum stack alignment. Also update preferred stack
14662 boundary for leaf functions.
14664 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14666 PR tree-optimization/82228
14667 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14670 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14672 * common/config/nds32/nds32-common.c
14673 (nds32_option_optimization_table): Refine formatting.
14674 (nds32_option_optimization_table): Use -fsched-pressure and
14675 -fomit-frame-pointer for specific optimization level.
14677 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14679 * config/nds32/nds32.c: Refine formatting and comments.
14680 * config/nds32/nds32.h: Likewise.
14681 * config/nds32/nds32.md: Likewise.
14682 * config/nds32/nds32-cost.c: Likewise.
14683 * config/nds32/nds32-isr.c: Likewise.
14684 * config/nds32/nds32-md-auxiliary.c: Likewise.
14685 * config/nds32/nds32-multiple.md: Likewise.
14686 * config/nds32/nds32-predicates.c: Likewise.
14688 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14689 Jakub Jelinek <jakub@redhat.com>
14691 Add support for -std=c++2a.
14692 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14694 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14696 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14699 * doc/extend.texi (Common Function Attributes): Add
14700 references to ARM, AArch64, and S/390 specific attributes.
14701 (Function Specific Option Pragmas): Add AArch64 and S/390
14702 to list of back ends that support the target pragma.
14704 2017-09-15 Nathan Sidwell <nathan@acm.org>
14706 * doc/standards.texi: Fix C++17 description. Update URLs for
14709 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14711 * common.opt (Wcast-align=strict): New warning option.
14712 * doc/invoke.texi: Document -Wcast-align=strict.
14714 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14716 * cgraph.h (cgraph_thunk_info): Add comments.
14717 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14718 assert for VIRTUAL_* arguments stricter.
14720 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14722 PR tree-optimization/71026
14723 * match.pd: Move RDIV patterns from fold-const.c
14724 * fold-const.c (distribute_real_division): Removed.
14725 (fold_binary_loc): Remove calls to distribute_real_divison.
14727 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14729 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14730 c++1z and gnu++1z as deprecated. Change other references to
14731 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14732 Change -Wc++1z-compat to -Wc++17-compat.
14733 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14734 * dwarf2out.c (highest_c_language): Handle C++17.
14735 (gen_compile_unit_die): Likewise.
14737 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14739 PR rtl-optimization/82192
14740 * combine.c (make_extraction): Don't look through non-paradoxical
14741 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14744 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14745 Alan Hayward <alan.hayward@arm.com>
14746 David Sherwood <david.sherwood@arm.com>
14748 * target.def (function_arg_offset): New hook.
14749 * targhooks.h (default_function_arg_offset): Declare.
14750 * targhooks.c (default_function_arg_offset): New function.
14751 * function.c (locate_and_pad_parm): Use
14752 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14753 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14754 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14755 * doc/tm.texi: Regenerate.
14756 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14757 * config/spu/spu.c (spu_function_arg_offset): New function.
14758 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14759 * system.h (FUNCTION_ARG_OFFSET): Poison.
14761 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14762 Alan Hayard <alan.hayward@arm.com>
14763 David Sherwood <david.sherwood@arm.com>
14765 * target.def (truly_noop_truncation): New hook.
14766 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14767 than TRULY_NOOP_TRUNCATION.
14768 * hooks.h (hook_bool_uint_uint_true): Declare.
14769 * hooks.c (hook_bool_uint_uint_true): New function.
14770 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14771 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14772 * doc/tm.texi: Regenerate.
14773 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14774 rather than TRULY_NOOP_TRUNCATION in comments.
14775 (simplify_comparison): Likewise.
14776 (record_truncated_value): Likewise.
14777 * expmed.c (extract_bit_field_1): Likewise.
14778 (extract_split_bit_field): Likewise.
14779 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14780 instead of TRULY_NOOP_TRUNCATION.
14781 * function.c (assign_parm_setup_block): Likewise.
14782 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14783 * rtlhooks.c: Include target.h.
14784 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14785 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14786 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14787 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14788 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14789 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14790 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14791 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14792 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14793 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14794 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14795 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14796 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14797 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14798 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14799 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14800 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14801 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14802 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14803 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14804 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14805 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14806 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14807 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14808 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14809 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14810 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14811 rather than TRULY_NOOP_TRUNCATION in comments.
14812 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14813 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14814 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14815 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14816 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14817 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14818 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14819 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14820 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14821 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14822 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14823 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14824 rather than TRULY_NOOP_TRUNCATION in comments.
14825 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14826 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14827 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14828 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14829 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14830 TRULY_NOOP_TRUNCATION condition.
14831 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14832 (TRULY_NOOP_TRUNCATION): Delete.
14833 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14834 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14835 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14836 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14837 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14838 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14839 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14840 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14841 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14842 rather than TRULY_NOOP_TRUNCATION in comments.
14843 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14844 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14845 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14846 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14847 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14848 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14850 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14853 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14854 (*cmp_ior): Likewise.
14855 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14856 (*ior_scc_scc_cmp): Likewise.
14857 (*and_scc_scc): Likewise.
14858 (*and_scc_scc_cmp): Likewise.
14860 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14861 Alan Hayard <alan.hayward@arm.com>
14862 David Sherwood <david.sherwood@arm.com>
14864 * target.def (can_change_mode_class): New hook.
14865 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14866 (hard_regno_nregs): Likewise.
14867 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14868 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14869 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14870 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14871 (LOAD_EXTEND_OP): Update accordingly.
14872 * doc/tm.texi: Regenerate.
14873 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14874 CANNOT_CHANGE_MODE_CLASS.
14875 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14876 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14877 * combine.c (simplify_set): Update accordingly.
14878 * emit-rtl.c (validate_subreg): Likewise.
14879 * recog.c (general_operand): Likewise.
14880 * regcprop.c (mode_change_ok): Likewise.
14881 * reload1.c (choose_reload_regs): Likewise.
14882 (inherit_piecemeal_p): Likewise.
14883 * rtlanal.c (simplify_subreg_regno): Likewise.
14884 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14885 instead of CANNOT_CHANGE_MODE_CLASS.
14886 (reload_cse_simplify_operands): Likewise.
14887 * reload.c (push_reload): Use targetm.can_change_mode_class
14888 instead of CANNOT_CHANGE_MODE_CLASS.
14889 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14890 REG_CANNOT_CHANGE_MODE_P.
14891 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14892 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14893 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14894 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14895 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14896 (arm_can_change_mode_class): New function.
14897 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14898 than CANNOT_CHANGE_MODE_CLASS in comments.
14899 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14900 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14901 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14902 (ix86_can_change_mode_class): ...this new function, inverting the
14903 sense of the return value.
14904 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14905 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14906 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14907 (ia64_can_change_mode_class): New function.
14908 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14909 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14910 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14911 (m32c_can_change_mode_class): ...this new function, inverting the
14912 sense of the return value.
14913 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14914 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14915 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14916 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14917 (mips_can_change_mode_class): ...this new function, inverting the
14918 sense of the return value.
14919 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14920 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14921 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14922 (msp430_can_change_mode_class): New function.
14923 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14924 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14925 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14926 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14927 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14928 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14929 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14930 (pa_cannot_change_mode_class): Replace with...
14931 (pa_can_change_mode_class): ...this new function, inverting the
14932 sense of the return value.
14933 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14934 than CANNOT_CHANGE_MODE_CLASS in comments.
14935 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14936 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14937 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14938 (pdp11_cannot_change_mode_class): Replace with...
14939 (pdp11_can_change_mode_class): ...this new function, inverting the
14940 sense of the return value.
14941 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14942 * config/powerpcspe/powerpcspe-protos.h
14943 (rs6000_cannot_change_mode_class_ptr): Delete.
14944 * config/powerpcspe/powerpcspe.c
14945 (rs6000_cannot_change_mode_class_ptr): Delete.
14946 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14947 (rs6000_option_override_internal): Assign to
14948 targetm.can_change_mode_class instead of
14949 rs6000_cannot_change_mode_class_ptr.
14950 (rs6000_cannot_change_mode_class): Replace with...
14951 (rs6000_can_change_mode_class): ...this new function, inverting the
14952 sense of the return value.
14953 (rs6000_debug_cannot_change_mode_class): Replace with...
14954 (rs6000_debug_can_change_mode_class): ...this new function.
14955 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14956 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14957 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14958 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14959 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14961 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14962 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14963 (rs6000_option_override_internal): Assign to
14964 targetm.can_change_mode_class instead of
14965 rs6000_cannot_change_mode_class_ptr.
14966 (rs6000_cannot_change_mode_class): Replace with...
14967 (rs6000_can_change_mode_class): ...this new function, inverting the
14968 sense of the return value.
14969 (rs6000_debug_cannot_change_mode_class): Replace with...
14970 (rs6000_debug_can_change_mode_class): ...this new function.
14971 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14972 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14973 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14974 (s390_can_change_mode_class): ...this new function, inverting the
14975 sense of the return value.
14976 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14977 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14978 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14979 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14980 (sh_cannot_change_mode_class): Replace with...
14981 (sh_can_change_mode_class): ...this new function, inverting the
14982 sense of the return value.
14983 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14984 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14985 (sparc_can_change_mode_class): New function.
14986 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14987 * config/spu/spu.c (spu_can_change_mode_class): New function.
14988 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14989 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14990 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14991 (visium_can_change_mode_class): New function.
14992 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14994 2017-09-15 Richard Biener <rguenther@suse.de>
14996 PR tree-optimization/82217
14997 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14998 but not undefined case.
15000 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15003 * postreload.c (reload_cse_simplify_operands): Skip
15004 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15006 2017-09-15 Richard Biener <rguenther@suse.de>
15008 PR tree-optimization/68823
15009 * graphite-scop-detection.c (build_alias_set): If we have a
15010 possible dependence check whether we can handle them by just
15011 looking at the DRs DR_ACCESS_FNs.
15012 (build_scops): If build_alias_set fails, fail the SCOP.
15014 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15016 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15017 to support float128 built-in functions that require the ISA 3.0
15019 (BU_FLOAT128_3_HW): Likewise.
15020 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15021 built-in functions.
15022 (FMAF128): Likewise.
15024 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15025 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15026 floating point instructions.
15027 (rs6000_invalid_builtin): Likewise.
15028 (rs6000_builtin_mask_names): Likewise.
15029 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15030 (RS6000_BTM_FLOAT128_HW): Likewise.
15031 (RS6000_BTM_COMMON): Likewise.
15032 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15034 * doc/extend.texi (RS/6000 built-in functions): Document the
15035 IEEE 128-bit floating point square root and fused multiply-add
15036 built-in functions.
15038 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15040 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15041 reg (r2) isn't in the set of registers defined in the prologue.
15043 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15044 Alan Hayward <alan.hayward@arm.com>
15045 David Sherwood <david.sherwood@arm.com>
15047 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15048 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15049 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15050 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15051 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15053 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15054 max_vectorization_factor.
15055 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15057 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15058 Alan Hayward <alan.hayward@arm.com>
15059 David Sherwood <david.sherwood@arm.com>
15061 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15062 (vect_worthwhile_without_simd_p): Declare.
15063 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15064 (vectorizable_reduction): Use it.
15065 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15066 (vectorizable_operation): Likewise.
15068 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15069 Alan Hayward <alan.hayward@arm.com>
15070 David Sherwood <david.sherwood@arm.com>
15072 * tree-vectorizer.h (vect_get_num_copies): New function.
15073 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15074 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15075 (vectorizable_induction): Likewise.
15076 (vectorizable_live_operation): Likewise.
15077 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15078 (vectorizable_bswap): Likewise.
15079 (vectorizable_call): Likewise.
15080 (vectorizable_conversion): Likewise.
15081 (vectorizable_assignment): Likewise.
15082 (vectorizable_shift): Likewise.
15083 (vectorizable_operation): Likewise.
15084 (vectorizable_store): Likewise.
15085 (vectorizable_load): Likewise.
15086 (vectorizable_condition): Likewise.
15087 (vectorizable_comparison): Likewise.
15088 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15090 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15091 Alan Hayward <alan.hayward@arm.com>
15092 David Sherwood <david.sherwood@arm.com>
15094 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15095 of vect_init_vector.
15097 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15098 Alan Hayward <alan.hayward@arm.com>
15099 David Sherwood <david.sherwood@arm.com>
15101 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15102 an inline wrapper that provides a location.
15103 (gimple_build_vector): Likewise.
15104 * gimple-fold.c (gimple_build_vector_from_val): New function.
15105 (gimple_build_vector): Likewise.
15106 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15107 functions to build the initial value. Always return a gimple value.
15108 (get_initial_defs_for_reduction): Likewise. Only compute
15110 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15111 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15112 (vectorizable_induction): Use gimple_build_vector rather than
15115 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15116 Alan Hayward <alan.hayward@arm.com>
15117 David Sherwood <david.sherwood@arm.com>
15119 * target.h (vec_perm_indices): New typedef.
15120 (auto_vec_perm_indices): Likewise.
15121 * optabs-query.h: Include target.h
15122 (can_vec_perm_p): Take a vec_perm_indices *.
15123 * optabs-query.c (can_vec_perm_p): Likewise.
15124 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15125 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15126 * tree-vect-generic.c (lower_vec_perm): Likewise.
15127 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15128 (vect_grouped_load_supported): Likewise.
15129 (vect_shift_permute_load_chain): Likewise.
15130 (vect_permute_store_chain): Use auto_vec_perm_indices.
15131 (vect_permute_load_chain): Likewise.
15132 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15133 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15134 Update uses of can_vec_perm_p.
15135 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15136 mode with a number of elements. Take a vec_perm_indices *.
15137 (vect_create_epilog_for_reduction): Update accordingly.
15138 Use auto_vec_perm_indices.
15139 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15140 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15141 (vect_transform_slp_perm_load): Likewise.
15142 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15143 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15144 (vect_gen_perm_mask_checked): Likewise.
15145 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15146 (vect_gen_perm_mask_checked): Likewise.
15147 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15148 (vectorizable_store): Likewise.
15149 (vectorizable_load): Likewise.
15150 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15151 (vectorizable_bswap): Likewise.
15153 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15154 Alan Hayward <alan.hayward@arm.com>
15155 David Sherwood <david.sherwood@arm.com>
15157 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15158 * tree.c (build_vector): Likewise.
15159 (build_vector_from_ctor): Update accordingly.
15160 (build_vector_from_val): Likewise.
15161 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15162 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15163 * tree-vect-generic.c (add_rshift): Likewise.
15164 (expand_vector_divmod): Likewise.
15165 (optimize_vector_constructor): Likewise.
15166 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15167 (vect_transform_slp_perm_load): Likewise.
15168 (vect_schedule_slp_instance): Likewise.
15169 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15170 (vectorizable_call): Likewise.
15171 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15172 * expmed.c (make_tree): Likewise.
15173 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15174 a vector passed to build_vector.
15175 (fold_convert_const): Likewise.
15176 (exact_inverse): Likewise.
15177 (fold_ternary_loc): Likewise.
15178 (fold_relational_const): Likewise.
15179 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15180 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15181 (const_unop): Likewise. Store the reduction accumulator in a
15182 variable rather than an array.
15183 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15184 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15185 the new vector, rather than constructing it after the input arrays.
15186 (native_interpret_vector): Use auto_vec<tree> when building
15187 a vector passed to build_vector. Add elements in order.
15188 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15189 auto_vec<tree> when building a vector passed to build_vector.
15190 (vect_create_epilog_for_reduction): Likewise.
15191 (vectorizable_induction): Likewise.
15192 (get_initial_def_for_reduction): Likewise. Fix indentation of
15194 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15196 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15197 passed to build_vector.
15199 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15200 Alan Hayward <alan.hayward@arm.com>
15201 David Sherwood <david.sherwood@arm.com>
15203 * tree-core.h (tree_base::u): Add an "nelts" field.
15204 (tree_vector): Use VECTOR_CST_NELTS as the length.
15205 * tree.c (tree_size): Likewise.
15206 (make_vector): Initialize VECTOR_CST_NELTS.
15207 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15208 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15209 TYPE_VECTOR_SUBPARTS.
15210 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15211 as the number of units, setting it from VECTOR_CST_NELTS.
15212 (const_vector_from_tree): Likewise.
15213 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15214 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15215 (fold_negate_expr_1): Likewise.
15216 (fold_convert_const): Likewise.
15217 (const_binop): Likewise. Differentiate the number of output and
15219 (const_unop): Likewise.
15220 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15221 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15222 in cases that did the opposite.
15224 2017-09-14 Richard Biener <rguenther@suse.de>
15226 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15229 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15231 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15233 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15236 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15237 if and where to split a bb, except for splitting before debug insn
15238 sequences followed by non-label real insn. Delete debug insns
15239 in between basic blocks.
15241 * combine.c (make_compound_operation_int): Formatting fixes.
15243 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15244 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15245 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15246 * config/sol2.h (LINK_EH_SPEC): Likewise.
15247 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15248 * config/s390/linux.h (LINK_SPEC): Likewise.
15249 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15250 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15251 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15252 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15253 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15254 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15255 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15256 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15258 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15260 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15262 (ENDFILE_LINUX_SPEC): Likewise.
15263 (LINK_EH_SPEC): Likewise.
15264 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15265 (LINK_OS_LINUX_SPEC32): Likewise.
15266 (LINK_OS_LINUX_SPEC64): Likewise.
15267 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15268 (LINK_OS_LINUX_SPEC): Likewise.
15270 2017-09-13 Martin Liska <mliska@suse.cz>
15272 PR middle-end/82154
15273 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15274 CASE_HIGH is NULL_TREE.
15276 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15277 Alan Hayward <alan.hayward@arm.com>
15278 David Sherwood <david.sherwood@arm.com>
15280 * target.def (secondary_memory_needed): New hook.
15281 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15282 instead of SECONDARY_MEMORY_NEEDED.
15283 (secondary_memory_needed_mode): Likewise.
15284 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15285 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15286 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15287 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15288 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15289 * doc/tm.texi: Regenerate.
15290 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15291 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15292 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15293 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15294 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15295 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15296 mode argument first and change the reg_class arguments to reg_class_t.
15297 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15298 Make static. Update the call to inline_secondary_memory_needed.
15299 (ix86_register_move_cost): Update the call to
15300 inline_secondary_memory_needed.
15301 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15302 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15304 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15305 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15307 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15308 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15309 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15310 and match hook interface. Add comment from mips.h.
15311 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15312 * config/mmix/mmix.md (truncdfsf2): Refer to
15313 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15315 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15316 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15317 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15318 (pa_secondary_memory_needed): New function.
15319 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15320 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15321 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15322 (pdp11_secondary_memory_needed): Make static and match hook interface.
15323 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15324 * config/powerpcspe/powerpcspe-protos.h
15325 (rs6000_secondary_memory_needed_ptr): Delete.
15326 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15328 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15329 (rs6000_option_override_internal): Assign to
15330 targetm.secondary_memory_needed rather than
15331 rs6000_secondary_memory_needed_ptr.
15332 (rs6000_secondary_memory_needed): Match hook interface.
15333 (rs6000_debug_secondary_memory_needed): Likewise.
15334 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15335 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15336 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15337 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15338 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15339 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15341 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15342 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15343 (rs6000_option_override_internal): Assign to
15344 targetm.secondary_memory_needed rather than
15345 rs6000_secondary_memory_needed_ptr.
15346 (rs6000_secondary_memory_needed): Match hook interface.
15347 (rs6000_debug_secondary_memory_needed): Likewise.
15348 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15349 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15350 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15351 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15352 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15353 (sparc_secondary_memory_needed): New function.
15354 * lra-constraints.c (check_and_process_move): Refer to
15355 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15357 (curr_insn_transform): Likewise.
15358 (process_alt_operands): Use targetm.secondary_memory_needed
15359 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15360 (check_secondary_memory_needed_p): Likewise.
15361 (choose_split_class): Likewise.
15362 * reload.c: Unconditionally include code that was previously
15363 conditional on SECONDARY_MEMORY_NEEDED.
15364 (push_secondary_reload): Use targetm.secondary_memory_needed
15365 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15366 (push_reload): Likewise.
15367 * reload1.c: Unconditionally include code that was previously
15368 conditional on SECONDARY_MEMORY_NEEDED.
15369 (choose_reload_regs): Use targetm.secondary_memory_needed
15370 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15371 (gen_reload): Likewise.
15372 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15374 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15375 Alan Hayward <alan.hayward@arm.com>
15376 David Sherwood <david.sherwood@arm.com>
15378 * target.def (secondary_memory_needed_mode): New hook:
15379 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15380 * targhooks.h (default_secondary_memory_needed_mode): New function.
15381 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15382 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15383 * doc/tm.texi: Regenerate.
15384 * lra-constraints.c (check_and_process_move): Use
15385 targetm.secondary_memory_needed_mode instead of
15386 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15387 (curr_insn_transform): Likewise.
15388 * reload.c (get_secondary_mem): Likewise.
15389 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15390 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15392 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15393 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15394 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15395 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15396 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15398 * config/powerpcspe/powerpcspe-protos.h
15399 (rs6000_secondary_memory_needed_mode): Delete.
15400 * config/powerpcspe/powerpcspe.c
15401 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15402 (rs6000_secondary_memory_needed_mode): Make static.
15403 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15404 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15406 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15408 (rs6000_secondary_memory_needed_mode): Make static.
15409 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15410 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15411 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15412 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15413 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15415 (sparc_secondary_memory_needed_mode): New function.
15416 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15418 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15420 * config/aarch64/constraints.md (Umq): New constraint.
15421 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15423 (mov<mode>): Update condition.
15425 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15427 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15428 when bitposition is the same.
15430 2017-09-13 Richard Biener <rguenther@suse.de>
15432 * dwarf2out.c (output_die_symbol): Remove.
15433 (output_die): Do not output a DIEs symbol.
15435 2017-09-13 Richard Biener <rguenther@suse.de>
15437 PR middle-end/82128
15438 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15439 default-def to avoid breaking iterator update with the weird
15440 interaction with cgraph_update_edges_for_call_stmt_node.
15442 2017-09-13 Richard Biener <rguenther@suse.de>
15444 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15445 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15446 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15447 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15449 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15451 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15452 Disable pc relative literal load irrespective of
15453 TARGET_FIX_ERR_A53_84341 for default.
15455 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15457 * config/sparc/sparc.c (output_return): Output the source location of
15458 the insn in the delay slot, if any.
15459 (output_sibcall): Likewise.
15461 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15464 * common.opt (-static-pie): New alias.
15465 (shared): Negate static-pie.
15466 (-no-pie): Update help text.
15468 (static-pie): New option.
15469 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15470 -static-pie support.
15471 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15472 (LINK_EH_SPEC): Likewise.
15473 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15474 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15475 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15476 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15477 (init_gcc_specs): Likewise.
15478 (init_spec): Likewise.
15479 (display_help): Update help message for -pie.
15480 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15483 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15485 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15486 (movdi_aarch64): Likewise.
15487 (movti_aarch64): Likewise.
15489 2017-09-12 Simon Wright <simon@pushface.org>
15492 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15493 calculation of the minor version, always output as 0.
15495 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15498 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15499 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15500 on it early, rather than manual conversion late. For
15501 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15502 instead of performing manual conversion.
15504 2017-09-12 Carl Love <cel@us.ibm.com>
15506 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15507 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15509 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15510 VMULOSW): Add definitions.
15511 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15512 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15513 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15514 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15515 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15517 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15519 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15521 (movti_aarch64): Likewise.
15522 (movdf_aarch64): Likewise.
15523 (movtf_aarch64): Likewise.
15524 (load_pairdi): Likewise.
15525 (store_pairdi): Likewise.
15526 (load_pairdf): Likewise.
15527 (store_pairdf): Likewise.
15528 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15529 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15530 (ldr_got_small_<mode>): Likewise.
15531 (ldr_got_small_28k_<mode>): Likewise.
15532 (ldr_got_tiny): Likewise.
15533 * config/aarch64/iterators.md (ldst_sz): New.
15534 (ldpstp_sz): Likewise.
15535 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15537 (thunderx_load): Split load_8 to load_16.
15538 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15540 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15541 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15542 (xgene1_store_pair): Split store_8 to store_16.
15543 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15544 (falkor_st_0_st_sd): Split store_8 to store_16.
15546 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15548 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15549 and store1/2/3/4 to store_4/8/12/16.
15550 * config/aarch64/aarch64.md: Update for rename.
15551 * config/arm/arm.md: Likewise.: Likewise.
15552 * config/arm/arm.c: Likewise.
15553 * config/arm/thumb1.md: Likewise.
15554 * config/arm/thumb2.md: Likewise.
15555 * config/arm/vfp.md: Likewise.
15556 * config/arm/arm-generic.md: Likewise.
15557 * config/arm/arm1020e.md: Likewise.
15558 * config/arm/arm1026ejs.md: Likewise.
15559 * config/arm/arm1136jfs.md: Likewise.
15560 * config/arm/arm926ejs.md: Likewise.
15561 * config/arm/cortex-a15.md: Likewise.
15562 * config/arm/cortex-a17.md: Likewise.
15563 * config/arm/cortex-a5.md: Likewise.
15564 * config/arm/cortex-a53.md: Likewise.
15565 * config/arm/cortex-a57.md: Likewise.
15566 * config/arm/cortex-a7.md: Likewise.
15567 * config/arm/cortex-a8.md: Likewise.
15568 * config/arm/cortex-a9.md: Likewise.
15569 * config/arm/cortex-m4.md: Likewise.
15570 * config/arm/cortex-m7.md: Likewise.
15571 * config/arm/cortex-r4.md: Likewise.
15572 * config/arm/exynos-m1.md: Likewise.
15573 * config/arm/fa526.md: Likewise.
15574 * config/arm/fa606te.md: Likewise.
15575 * config/arm/fa626te.md: Likewise.
15576 * config/arm/fa726te.md: Likewise.
15577 * config/arm/fmp626.md: Likewise.
15578 * config/arm/iwmmxt.md: Likewise.
15579 * config/arm/ldmstm.md: Likewise.
15580 * config/arm/marvell-pj4.md: Likewise.
15581 * config/arm/xgene1.md: Likewise.
15582 * config/aarch64/thunderx.md: Likewise.
15583 * config/aarch64/thunderx2t99.md: Likewise.
15584 * config/aarch64/falkor.md: Likewise.
15586 2017-09-12 Martin Liska <mliska@suse.cz>
15588 * attribs.c (private_lookup_attribute): New function.
15589 * attribs.h (private_lookup_attribute): Declared here.
15590 (lookup_attribute): Called from this place.
15592 2017-09-12 Richard Biener <rguenther@suse.de>
15594 PR tree-optimization/82157
15595 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15596 stmts with side-effects.
15598 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15599 Alan Hayward <alan.hayward@arm.com>
15600 David Sherwood <david.sherwood@arm.com>
15602 * target.def (hard_regno_nregs): New hook.
15603 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15604 * targhooks.h (default_hard_regno_nregs): Declare.
15605 * targhooks.c (default_hard_regno_nregs): New function.
15606 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15607 (TARGET_HARD_REGNO_NREGS): ...this hook.
15608 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15609 (CLASS_MAX_NREGS): Likewise.
15610 * doc/tm.texi: Regenerate.
15611 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15612 instead of HARD_REGNO_NREGS.
15613 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15614 HARD_REGNO_NREGS in the comment.
15615 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15616 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15617 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15618 Return an unsigned int.
15619 (TARGET_HARD_REGNO_NREGS): Redefine.
15620 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15621 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15622 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15623 (arc_hard_regno_nregs): New function.
15624 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15625 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15626 (arm_hard_regno_nregs): New function.
15627 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15628 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15629 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15630 (TARGET_HARD_REGNO_NREGS): Redefine.
15631 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15632 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15633 (HARD_REGNO_NREGS): Delete.
15634 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15635 (cr16_hard_regno_nregs): New function.
15636 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15637 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15638 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15639 (cris_hard_regno_nregs): New function.
15640 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15641 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15642 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15643 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15644 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15645 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15646 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15647 (frv_hard_regno_nregs): Make static. Take and return an
15649 (frv_class_max_nregs): Remove outdated copy of documentation.
15650 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15651 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15652 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15653 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15654 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15655 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15656 (TARGET_HARD_REGNO_NREGS): Redefine.
15657 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15658 (CLASS_MAX_NREGS): Update comment.
15659 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15660 (ia64_hard_regno_nregs): New function.
15661 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15662 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15663 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15664 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15665 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15667 (m32c_hard_regno_nregs): Likewise. Make static.
15668 (TARGET_HARD_REGNO_NREGS): Redefine.
15669 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15670 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15671 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15672 (m68k_hard_regno_nregs): New function.
15673 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15674 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15675 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15676 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15677 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15678 Take and return an unsigned int.
15679 (TARGET_HARD_REGNO_NREGS): Redefine.
15680 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15681 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15682 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15683 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15684 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15685 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15686 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15687 (msp430_hard_regno_nregs): Make static. Take and return an
15689 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15690 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15691 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15692 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15693 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15694 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15695 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15696 (TARGET_HARD_REGNO_NREGS): Redefine.
15697 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15698 (PA_HARD_REGNO_NREGS): ...this.
15699 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15700 (PA_HARD_REGNO_NREGS): ...this.
15701 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15702 (pa_hard_regno_nregs): New function.
15703 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15704 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15705 (pdp11_hard_regno_nregs): New function.
15706 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15707 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15708 (rs6000_hard_regno_nregs_hook): New function.
15709 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15710 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15711 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15712 Take and return an unsigned int. Move earlier in file.
15713 (TARGET_HARD_REGNO_NREGS): Redefine.
15714 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15715 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15716 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15717 (rl78_hard_regno_nregs): Make static. Take and return an
15719 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15720 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15721 (rs6000_hard_regno_nregs_hook): New function.
15722 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15723 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15724 (TARGET_HARD_REGNO_NREGS): Redefine.
15725 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15726 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15727 instead of HARD_REGNO_NREGS.
15728 (s390_hard_regno_nregs): New function.
15729 (s390_hard_regno_mode_ok): Add comment from s390.h.
15730 (TARGET_HARD_REGNO_NREGS): Redefine.
15731 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15732 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15733 (sh_hard_regno_nregs): New function.
15734 (sh_pass_in_reg_p): Use it.
15735 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15736 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15737 (sparc_hard_regno_nregs): New function.
15738 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15739 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15740 (spu_function_arg_advance): Use it, supplying a valid register number.
15741 (TARGET_HARD_REGNO_NREGS): Redefine.
15742 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15743 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15744 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15745 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15746 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15747 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15748 (CLASS_MAX_NREGS): Remove copy of old documentation.
15749 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15750 (visium_hard_regno_nregs): New function.
15751 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15752 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15753 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15754 xtensa_hard_regno_nregs): New function.
15755 * system.h (HARD_REGNO_NREGS): Poison.
15757 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15759 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15760 hard_regno_nregs instead of HARD_REGNO_NREGS.
15761 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15762 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15763 (c6x_expand_epilogue): Likewise.
15764 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15765 (frv_read_iacc_argument): Likewise.
15766 * config/sh/sh.c: Include regs.h.
15767 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15768 (regs_used): Likewise.
15769 (output_stack_adjust): Likewise.
15770 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15771 * expmed.c: Include regs.h.
15772 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15773 * ree.c: Include regs.h.
15774 (combine_reaching_defs): Use hard_regno_nregs instead of
15776 (add_removable_extension): Likewise.
15778 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15780 * regs.h (hard_regno_nregs): Turn into a function.
15781 (end_hard_regno): Update accordingly.
15782 * caller-save.c (setup_save_areas): Likewise.
15783 (save_call_clobbered_regs): Likewise.
15784 (replace_reg_with_saved_mem): Likewise.
15785 (insert_restore): Likewise.
15786 (insert_save): Likewise.
15787 * combine.c (can_change_dest_mode): Likewise.
15788 (move_deaths): Likewise.
15789 (distribute_notes): Likewise.
15790 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15791 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15792 (rs6000_split_multireg_move): Likewise.
15793 (rs6000_register_move_cost): Likewise.
15794 (rs6000_memory_move_cost): Likewise.
15795 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15796 (rs6000_split_multireg_move): Likewise.
15797 (rs6000_register_move_cost): Likewise.
15798 (rs6000_memory_move_cost): Likewise.
15799 * cselib.c (cselib_reset_table): Likewise.
15800 (cselib_lookup_1): Likewise.
15801 * emit-rtl.c (set_mode_and_regno): Likewise.
15802 * function.c (aggregate_value_p): Likewise.
15803 * ira-color.c (setup_profitable_hard_regs): Likewise.
15804 (check_hard_reg_p): Likewise.
15805 (calculate_saved_nregs): Likewise.
15806 (assign_hard_reg): Likewise.
15807 (improve_allocation): Likewise.
15808 (calculate_spill_cost): Likewise.
15809 * ira-emit.c (modify_move_list): Likewise.
15810 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15811 (ira_hard_reg_in_set_p): Likewise.
15812 * ira.c (setup_reg_mode_hard_regset): Likewise.
15813 (clarify_prohibited_class_mode_regs): Likewise.
15814 (check_allocation): Likewise.
15815 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15816 (lra_setup_reg_renumber): Likewise.
15817 (setup_try_hard_regno_pseudos): Likewise.
15818 (spill_for): Likewise.
15819 (assign_hard_regno): Likewise.
15820 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15821 * lra-constraints.c (in_class_p): Likewise.
15822 (lra_constraint_offset): Likewise.
15823 (simplify_operand_subreg): Likewise.
15824 (lra_constraints): Likewise.
15825 (split_reg): Likewise.
15826 (split_if_necessary): Likewise.
15827 (invariant_p): Likewise.
15828 (inherit_in_ebb): Likewise.
15829 * lra-lives.c (process_bb_lives): Likewise.
15830 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15831 (get_hard_regs): Likewise.
15832 (do_remat): Likewise.
15833 * lra-spills.c (assign_spill_hard_regs): Likewise.
15834 * mode-switching.c (create_pre_exit): Likewise.
15835 * postreload.c (reload_combine_recognize_pattern): Likewise.
15836 * recog.c (peep2_find_free_register): Likewise.
15837 * regcprop.c (kill_value_regno): Likewise.
15838 (set_value_regno): Likewise.
15839 (copy_value): Likewise.
15840 (maybe_mode_change): Likewise.
15841 (find_oldest_value_reg): Likewise.
15842 (copyprop_hardreg_forward_1): Likewise.
15843 * regrename.c (check_new_reg_p): Likewise.
15844 (regrename_do_replace): Likewise.
15845 * reload.c (push_reload): Likewise.
15846 (combine_reloads): Likewise.
15847 (find_dummy_reload): Likewise.
15848 (operands_match_p): Likewise.
15849 (find_reloads): Likewise.
15850 (find_equiv_reg): Likewise.
15851 (reload_adjust_reg_for_mode): Likewise.
15852 * reload1.c (count_pseudo): Likewise.
15853 (count_spilled_pseudo): Likewise.
15854 (find_reg): Likewise.
15855 (clear_reload_reg_in_use): Likewise.
15856 (free_for_value_p): Likewise.
15857 (allocate_reload_reg): Likewise.
15858 (choose_reload_regs): Likewise.
15859 (reload_adjust_reg_for_temp): Likewise.
15860 (emit_reload_insns): Likewise.
15861 (delete_output_reload): Likewise.
15862 * rtlanal.c (subreg_get_info): Likewise.
15863 * sched-deps.c (sched_analyze_reg): Likewise.
15864 * sel-sched.c (init_regs_for_mode): Likewise.
15865 (mark_unavailable_hard_regs): Likewise.
15866 (choose_best_reg_1): Likewise.
15867 (verify_target_availability): Likewise.
15868 * valtrack.c (dead_debug_insert_temp): Likewise.
15869 * var-tracking.c (track_loc_p): Likewise.
15870 (emit_note_insn_var_location): Likewise.
15871 * varasm.c (make_decl_rtl): Likewise.
15872 * reginfo.c (choose_hard_reg_mode): Likewise.
15873 (init_reg_modes_target): Refer directly to
15874 this_target_regs->x_hard_regno_nregs.
15876 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15878 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15879 instead of hard_regno_nregs.
15881 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15883 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15884 end_hard_regno instead of hard_regno_nregs.
15885 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15886 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15887 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15888 * ira-color.c (improve_allocation): Likewise.
15889 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15890 * lra-lives.c (mark_regno_live): Likewise.
15891 (mark_regno_dead): Likewise.
15892 * lra-remat.c (operand_to_remat): Likewise.
15893 * lra.c (collect_non_operand_hard_regs): Likewise.
15894 * postreload.c (reload_combine_note_store): Likewise.
15895 (move2add_valid_value_p): Likewise.
15896 * reload.c (regno_clobbered_p): Likewise.
15898 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15900 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15902 * config/v850/v850.c (v850_reorg): Likewise.
15903 * reload.c (refers_to_regno_for_reload_p): Likewise.
15904 (find_equiv_reg): Likewise.
15905 * reload1.c (reload_reg_reaches_end_p): Likewise.
15907 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15909 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15911 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15912 * config/arm/arm.c (output_move_neon): Likewise.
15913 (arm_attr_length_move_neon): Likewise.
15914 (neon_split_vcombine): Likewise.
15915 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15916 (c6x_mark_reg_written): Likewise.
15917 (c6x_dwarf_register_span): Likewise.
15918 * config/i386/i386.c (ix86_save_reg): Likewise.
15919 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15920 (rws_access_reg): Likewise.
15921 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15922 * mode-switching.c (create_pre_exit): Likewise.
15923 * ree.c (combine_reaching_defs): Likewise.
15924 (add_removable_extension): Likewise.
15925 * regcprop.c (find_oldest_value_reg): Likewise.
15926 (copyprop_hardreg_forward_1): Likewise.
15927 * reload.c (reload_inner_reg_of_subreg): Likewise.
15928 (push_reload): Likewise.
15929 (combine_reloads): Likewise.
15930 (find_dummy_reload): Likewise.
15931 (reload_adjust_reg_for_mode): Likewise.
15932 * reload1.c (find_reload_regs): Likewise.
15933 (forget_old_reloads_1): Likewise.
15934 (reload_reg_free_for_value_p): Likewise.
15935 (reload_adjust_reg_for_temp): Likewise.
15936 (emit_reload_insns): Likewise.
15937 (delete_output_reload): Likewise.
15938 * sel-sched.c (choose_best_reg_1): Likewise.
15939 (choose_best_pseudo_reg): Likewise.
15941 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15942 Alan Hayward <alan.hayward@arm.com>
15943 David Sherwood <david.sherwood@arm.com>
15945 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15946 * target.def (slow_unaligned_access): New hook.
15947 * targhooks.h (default_slow_unaligned_access): Declare.
15948 * targhooks.c (default_slow_unaligned_access): New function.
15949 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15950 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15951 * doc/tm.texi: Regenerate.
15952 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15953 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15954 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15956 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15957 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15959 (rs6000_slow_unaligned_access): New function.
15960 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15961 (expand_block_compare): Likewise.
15962 (expand_strn_compare): Likewise.
15963 (rs6000_rtx_costs): Likewise.
15964 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15965 (riscv_slow_unaligned_access): Likewise.
15966 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15967 (riscv_slow_unaligned_access_p): ...this and make static.
15968 (riscv_option_override): Update accordingly.
15969 (riscv_slow_unaligned_access): New function.
15970 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15971 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15972 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15973 (rs6000_slow_unaligned_access): New function.
15974 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15975 (rs6000_rtx_costs): Likewise.
15976 * config/rs6000/rs6000-string.c (expand_block_compare)
15977 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15978 of SLOW_UNALIGNED_ACCESS.
15979 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15980 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15981 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15982 of SLOW_UNALIGNED_ACCESS.
15983 * expmed.c (simple_mem_bitfield_p): Likewise.
15984 * expr.c (alignment_for_piecewise_move): Likewise.
15985 (emit_group_load_1): Likewise.
15986 (emit_group_store): Likewise.
15987 (copy_blkmode_from_reg): Likewise.
15988 (emit_push_insn): Likewise.
15989 (expand_assignment): Likewise.
15990 (store_field): Likewise.
15991 (expand_expr_real_1): Likewise.
15992 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15993 * lra-constraints.c (simplify_operand_subreg): Likewise.
15994 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15995 * gimple-ssa-store-merging.c: Likewise in block comment at start
15997 * tree-ssa-strlen.c: Include target.h.
15998 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15999 of SLOW_UNALIGNED_ACCESS.
16000 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16002 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16004 PR rtl-optimization/82185
16005 * expmed.c (emit_store_flag_int): Only test tem if it has been
16008 2017-09-12 Richard Biener <rguenther@suse.de>
16010 PR middle-end/82149
16011 * match.pd ((FTYPE) N CMP CST): Fix typo.
16013 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16015 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16017 (mips_near_type_p): Add 'short_call' attribute as a synonym
16019 * doc/extend.texi (short_call): Document new function attribute.
16021 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16024 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16025 assertion check that in the condition.
16026 (get_atomic_generic_size): Likewise. Before testing if parameter
16027 has pointer type, if it has array type, call for C++
16028 default_conversion to perform array-to-pointer conversion.
16030 2017-09-12 Richard Biener <rguenther@suse.de>
16032 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16033 for operations we cannot scalarize.
16035 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16037 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16038 vectors heap vectors. Clean up comments.
16039 Make visited_bbs a reference.
16040 (profitable_jump_thread_path): Make GC
16041 vectors heap vectors. Clean up comments.
16043 (convert_and_register_jump_thread_path): Make GC vectors heap
16045 (check_subpath_and_update_thread_path): Same. Clean up comments.
16046 Make visited_bbs a reference.
16047 (handle_phi): Abstract common code to to
16048 register_jump_thread_path_if_profitable.
16049 Rename VAR_BB to DEF_BB.
16051 Make GC vectors heap vectors.
16052 Make visited_bbs a reference.
16053 (handle_assignment): Same.
16054 (register_jump_thread_path_if_profitable): New.
16055 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16057 Make GC vectors heap vectors. Clean up comments.
16058 Make visited_bbs a reference.
16059 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16060 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16063 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16066 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16067 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16069 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16071 Revert r251800 and r251799.
16073 2017-09-11 Martin Jambor <mjambor@suse.cz>
16076 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16077 arguments in advance.
16078 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16079 use it to find predecessor edges.
16080 (naive_outof_ssa): Collect vector of predecessors.
16082 2017-09-08 Jason Merrill <jason@redhat.com>
16084 PR c++/70029 - ICE with ref-qualifier and -flto
16085 * langhooks.h (struct lang_hooks_for_types): Add
16086 copy_lang_qualifiers.
16087 * attribs.c (build_type_attribute_qual_variant): Use it.
16088 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16090 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16091 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16093 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16096 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16097 (*mulsi3_sp64): New instruction.
16098 (mulsi3): New expander.
16100 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16102 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16104 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16106 * sancov.c: Include memmodel.h.
16108 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16111 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16114 2017-09-07 Carl Love <cel@us.ibm.com>
16116 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16117 the sldi instruction.
16119 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16121 * sancov.c: Include tm_p.h.
16123 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16126 * output.h (switch_to_other_text_partition): New declaration.
16127 * varasm.c (switch_to_other_text_partition): New function.
16128 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16129 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16130 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16131 to the other text partition before emitting LCL label and switch back
16132 after emitting the word after it.
16134 2017-09-07 Richard Biener <rguenther@suse.de>
16136 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16137 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16138 critical edges here, after loop init.
16139 (pass_data_pre): Remove PROP_no_crit_edges flags.
16140 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16141 for valueization of call args to avoid leaking VN_TOP.
16142 (visit_use): Assert we do not visit default defs.
16143 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16144 Use error_mark_node to more easily detect leaking VN_TOP.
16145 All default-defs are varying, not VN_TOP. Mark them visited.
16146 (run_scc_vn): Make code match comment.
16148 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16150 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16151 OPTION_MASK_FLOAT128_KEYWORD.
16152 (POWERPC_MASKS): Likewise.
16153 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16154 support for the -mfloat128-type option, and make -mfloat128
16155 default on PowerPC Linux systems. Define or undefine
16156 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16157 Define __float128 to be __ieee128 if IEEE 128-bit support is
16158 enabled, or undefine it.
16159 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16160 Delete defining __FLOAT128_TYPE__.
16161 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16162 -mfloat128-type option and make -mfloat128 default on PowerPC
16164 (TARGET_FLOAT128_TYPE): Likewise.
16165 (-mfloat128-type): Likewise.
16166 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16167 Delete the -mfloat128-type option and make -mfloat128 default on
16168 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16169 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16170 128-bit floating point is enabled. Change tests from using
16171 -mfloat128-type to -mfloat128.
16172 (rs6000_mangle_type): Use the correct mangling for the __float128
16173 type even if normal long double is restricted to 64-bits.
16174 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16176 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16177 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16178 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16180 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16181 documentation for -mfloat128.
16183 2017-09-06 Olivier Hainque <hainque@adacore.com>
16185 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16187 2017-09-06 Wish Wu <wishwu007@gmail.com>
16188 Jakub Jelinek <jakub@redhat.com>
16190 * asan.c (initialize_sanitizer_builtins): Add
16191 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16192 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16193 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16194 BT_FN_VOID_UINT64_PTR variables.
16195 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16196 (BT_FN_VOID_UINT16_UINT16): Likewise.
16197 (BT_FN_VOID_UINT32_UINT32): Likewise.
16198 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16199 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16200 (BT_FN_VOID_UINT64_PTR): Likewise.
16201 * common.opt (flag_sanitize_coverage): New variable.
16202 (fsanitize-coverage=trace-pc): Remove.
16203 (fsanitize-coverage=): Add.
16204 * flag-types.h (enum sanitize_coverage_code): New enum.
16205 * fold-const.c (fold_range_test): Disable non-short-circuit
16206 optimization if flag_sanitize_coverage.
16207 (fold_truth_andor): Likewise.
16208 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16209 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16210 (coverage_sanitizer_opts): New array.
16211 (get_closest_sanitizer_option): Add OPTS argument, handle also
16212 OPT_fsanitize_coverage_.
16213 (parse_sanitizer_options): Adjusted to also handle
16214 OPT_fsanitize_coverage_.
16215 (common_handle_option): Add OPT_fsanitize_coverage_.
16216 * sancov.c (instrument_comparison, instrument_switch): New function.
16217 (sancov_pass): Add trace-cmp support.
16218 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16219 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16220 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16221 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16222 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16223 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16224 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16225 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16226 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16227 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16229 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16231 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16232 error. Only quit immediately if parsing is complete.
16233 (BEGIN): Initialize fatal_err and parse_done.
16234 (begin fpu, end fpu): Check number of arguments.
16235 (begin arch, end arch): Likewise.
16236 (begin cpu, end cpu): Likewise.
16237 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16238 (optalias): Likewise.
16240 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16242 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16243 * config/arm/arm-isa.h: Delete. Move definitions to ...
16244 * arm-cpus.in: ... here. Use new feature and fgroup values.
16245 * config/arm/arm.c (arm_option_override): Use lower case for feature
16247 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16248 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16249 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16251 (print_isa_bits_for): New function.
16252 (gen_isa): New function.
16253 (gen_comm_data): Use print_isa_bits_for.
16254 (define feature): New keyword.
16255 (define fgroup): New keyword.
16256 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16257 (arm-isa.h): Add rule to generate file.
16258 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16259 case for feature bit names.
16261 2017-09-06 Richard Biener <rguenther@suse.de>
16263 * tree-ssa-pre.c (NECESSARY): Remove.
16264 (create_expression_by_pieces): Do not touch pass-local flags.
16265 (insert_into_preds_of_block): Likewise.
16266 (do_pre_regular_insertion): Likewise.
16267 (eliminate_insert): Likewise.
16268 (eliminate_dom_walker::before_dom_children): Likewise.
16269 (fini_eliminate): Do not look at inserted_exprs.
16270 (mark_operand_necessary): Remove.
16271 (remove_dead_inserted_code): Replace with simple work-list
16272 algorithm based on inserted_exprs and SSA uses.
16273 (pass_pre::execute): Re-order fini_eliminate and
16274 remove_dead_inserted_code.
16276 2017-09-06 Olivier Hainque <hainque@adacore.com>
16278 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16279 for VxWorks 7. Adjust surrounding comments.
16281 2017-09-06 Richard Biener <rguenther@suse.de>
16283 * gimple-ssa-strength-reduction.c
16284 (find_candidates_dom_walker::before_dom_children): Also allow
16287 2017-09-06 Richard Biener <rguenther@suse.de>
16289 PR tree-optimization/82108
16290 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16291 for gap in the non-permutation SLP case.
16293 2017-09-06 Martin Jambor <mjambor@suse.cz>
16295 PR tree-optimization/82078
16296 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16297 add_access_to_work_queue...
16298 (build_accesses_from_assign): ...here.
16299 (propagate_all_subaccesses): Make sure racc is the group
16300 representative, if there is one.
16302 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16304 PR middle-end/82095
16305 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16308 2017-09-06 Richard Biener <rguenther@suse.de>
16310 * gimple-ssa-strength-reduction.c
16311 (find_candidates_dom_walker::before_doom_children): Use a
16312 type and not a mode check.
16314 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16317 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16319 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16321 2017-09-05 Jeff Law <law@redhat.com>
16323 PR tree-optimization/64910
16324 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16325 cases where we have 3 or more operands.
16327 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16329 PR middle-end/81768
16330 * omp-low.c (lower_omp_for): Recompute tree invariant if
16331 gimple_omp_for_initial/final is ADDR_EXPR.
16333 PR middle-end/81768
16334 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16335 into gimple val before gimplification fo the COND_EXPR.
16337 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16339 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16340 REGION_COPY argument.
16341 (thread_through_all_blocks): Remove unused argument to
16342 duplicate_thread_path.
16344 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16345 Alan Hayward <alan.hayward@arm.com>
16346 David Sherwood <david.sherwood@arm.com>
16348 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16349 Take a scalar_mode rather than a machine_mode.
16350 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16351 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16352 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16353 (aarch64_gen_adjusted_ldpstp): Likewise.
16354 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16356 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16357 Alan Hayward <alan.hayward@arm.com>
16358 David Sherwood <david.sherwood@arm.com>
16360 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16361 Take a scalar_int_mode instead of a machine_mode.
16362 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16363 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16364 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16365 (aarch64_simd_attr_length_rglist): Delete.
16366 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16367 a scalar_int_mode instead of a machine_mode.
16368 (aarch64_add_offset): Likewise.
16369 (aarch64_internal_mov_immediate): Likewise
16370 (aarch64_add_constant_internal): Likewise.
16371 (aarch64_add_constant): Likewise.
16372 (aarch64_movw_imm): Likewise.
16373 (aarch64_rtx_arith_op_extract_p): Likewise.
16374 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16375 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16376 Remove assert that the mode isn't a vector.
16377 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16378 (aarch64_expand_mov_immediate): Update calls after above changes.
16379 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16380 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16381 (aarch64_move_imm): Likewise.
16382 (aarch64_can_const_movi_rtx_p): Likewise.
16383 (aarch64_strip_extend): Likewise.
16384 (aarch64_extr_rtx_p): Likewise.
16385 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16386 a CONST_INT when the mode parameter is VOIDmode.
16387 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16389 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16391 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16392 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16393 (bitwise_type_for_mode): Update accordingly.
16395 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16397 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16398 * stor-layout.c (mode_for_size_tree): Likewise.
16399 (mode_for_array): Update accordingly.
16400 (layout_decl): Likewise.
16401 (compute_record_mode): Likewise. Only set the mode once.
16403 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16405 * target.def (get_mask_mode): Change return type to opt_mode.
16407 * doc/tm.texi: Regenerate.
16408 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16409 * targhooks.c (default_get_mask_mode): Likewise.
16410 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16411 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16412 targetm.get_mask_mode.
16413 * tree.c (build_truth_vector_type): Likewise.
16415 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16417 * machmode.h (mode_for_vector): Return an opt_mode.
16418 * stor-layout.c (mode_for_vector): Likewise.
16419 (mode_for_int_vector): Update accordingly.
16420 (layout_type): Likewise.
16421 * config/i386/i386.c (emit_memmov): Likewise.
16422 (ix86_expand_set_or_movmem): Likewise.
16423 (ix86_expand_vector_init): Likewise.
16424 (ix86_get_mask_mode): Likewise.
16425 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16427 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16428 * expmed.c (extract_bit_field_1): Likewise.
16429 * expr.c (expand_expr_real_2): Likewise.
16430 * optabs-query.c (can_vec_perm_p): Likewise.
16431 (can_vec_mask_load_store_p): Likewise.
16432 * optabs.c (expand_vec_perm): Likewise.
16433 * targhooks.c (default_get_mask_mode): Likewise.
16434 * tree-vect-stmts.c (vectorizable_store): Likewise.
16435 (vectorizable_load): Likewise.
16436 (get_vectype_for_scalar_type_and_size): Likewise.
16438 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16440 * machmode.h (mode_for_int_vector): New function.
16441 * stor-layout.c (mode_for_int_vector): Likewise.
16442 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16443 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16444 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16445 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16446 (s390_expand_vcond): Likewise.
16448 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16450 * machmode.h (opt_machine_mode): New type.
16451 (opt_mode<T>): Allow construction from anything that can be
16453 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16454 (mode_for_size): Return an opt_machine_mode.
16455 * stor-layout.c (mode_for_size): Likewise.
16456 (mode_for_size_tree): Update call accordingly.
16457 (bitwise_mode_for_mode): Likewise.
16458 (make_fract_type): Likewise.
16459 (make_accum_type): Likewise.
16460 * caller-save.c (replace_reg_with_saved_mem): Update call
16462 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16463 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16464 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16465 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16466 * expmed.c (extract_bit_field_1): Likewise.
16467 * reload.c (get_secondary_mem): Likewise.
16468 * varasm.c (assemble_integer): Likewise.
16469 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16472 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16474 * machmode.h (decimal_float_mode_for_size): New function.
16475 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16476 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16477 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16478 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16479 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16480 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16482 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16484 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16485 (get_builtin_sync_mode): Likewise.
16486 (expand_ifn_atomic_compare_exchange): Likewise.
16487 (expand_builtin_atomic_clear): Likewise.
16488 (expand_builtin_atomic_test_and_set): Likewise.
16489 (fold_builtin_atomic_always_lock_free): Likewise.
16490 * calls.c (compute_argument_addresses): Likewise.
16491 (emit_library_call_value_1): Likewise.
16492 (store_one_arg): Likewise.
16493 * combine.c (combine_instructions): Likewise.
16494 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16495 * config/arm/arm.c (arm_function_value): Likewise.
16496 (aapcs_allocate_return_reg): Likewise.
16497 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16498 * config/i386/i386.c (construct_container): Likewise.
16499 (ix86_gimplify_va_arg): Likewise.
16500 (ix86_expand_sse_cmp): Likewise.
16501 (emit_memmov): Likewise.
16502 (emit_memset): Likewise.
16503 (expand_small_movmem_or_setmem): Likewise.
16504 (ix86_expand_pextr): Likewise.
16505 (ix86_expand_pinsr): Likewise.
16506 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16507 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16509 * config/mips/mips.c (mips_function_value_1) Likewise.
16510 (mips_block_move_straight): Likewise.
16511 (mips_expand_ins_as_unaligned_store): Likewise.
16512 * config/powerpcspe/powerpcspe.c
16513 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16514 (rs6000_darwin64_record_arg_flush): Likewise.
16515 * config/rs6000/rs6000.c
16516 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16517 (rs6000_darwin64_record_arg_flush): Likewise.
16518 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16519 (sparc_function_value_1): Likewise.
16520 * config/spu/spu.c (adjust_operand): Likewise.
16521 (spu_emit_branch_or_set): Likewise.
16522 (arith_immediate_p): Likewise.
16523 * emit-rtl.c (gen_lowpart_common): Likewise.
16524 * expr.c (expand_expr_real_1): Likewise.
16525 * function.c (assign_parm_setup_block): Likewise.
16526 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16527 * reload1.c (alter_reg): Likewise.
16528 * stor-layout.c (mode_for_vector): Likewise.
16529 (layout_type): Likewise.
16531 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16533 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16534 (spu_convert_move): Likewise.
16535 * lower-subreg.c (resolve_simple_move): Likewise.
16537 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16540 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16541 define_insn to a define_expand.
16542 (altivec_vsum2sws_direct): New define_insn.
16543 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16545 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16547 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16550 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16555 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16556 realign stack if stack alignment needed is less than incoming
16559 2017-09-05 Marek Polacek <polacek@redhat.com>
16562 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16565 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16567 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16569 2017-09-05 Richard Biener <rguenther@suse.de>
16571 PR tree-optimization/82084
16572 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16574 2017-09-05 Richard Biener <rguenther@suse.de>
16576 PR tree-optimization/82102
16577 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16579 2017-09-05 Martin Liska <mliska@suse.cz>
16581 PR tree-optimization/82032
16582 * tree-cfg.c (generate_range_test): New function.
16583 * tree-cfg.h (generate_range_test): Declared here.
16584 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16585 (cleanup_control_expr_graph): Use it.
16586 * tree-switch-conversion.c (try_switch_expansion): Remove
16588 (emit_case_nodes): Use generate_range_test.
16590 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16593 * config/i386/i386.md (*<btsc><mode>_mask): Add
16594 TARGET_USE_BT to insn constraint.
16595 (*btr<mode>_mask): Ditto.
16597 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16599 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16600 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16602 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16605 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16606 TARGET_NEON and TARGET_IWMMXT.
16607 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16608 TARGET_NEON and TARGET_IWMMXT.
16609 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16611 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16613 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16614 (ix86_rewrite_tls_address): Ditto.
16615 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16616 (ix86_rewrite_tls_address_1): Ditto.
16617 (ix86_rewrite_tls_address): Ditto.
16618 * config/i386/predicates.md (tls_address_pattern): New predicate.
16619 * config/i386/i386.md (TLS address splitter): New splitter.
16621 2017-09-04 Richard Biener <rguenther@suse.de>
16623 PR tree-optimization/82084
16624 * fold-const.h (can_native_encode_string_p): Declare.
16625 * fold-const.c (can_native_encode_string_p): Factor out from ...
16626 (native_encode_string): ... here.
16627 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16628 vectorizing stores from constants we later cannot handle.
16630 2017-09-04 Marek Polacek <polacek@redhat.com>
16633 * doc/invoke.texi: Update -Wtautological-compare documentation.
16635 2017-09-04 Jeff Law <law@redhat.com>
16637 PR tree-optimization/64910
16638 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16639 swap the first and last operand if the last is a constant.
16641 2017-09-04 Marek Polacek <polacek@redhat.com>
16644 * convert.c (do_narrow): When sanitizing signed integer overflows,
16645 bail out for signed types.
16646 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16648 2017-09-04 Richard Biener <rguenther@suse.de>
16650 PR tree-optimization/82060
16651 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16652 Move devirtualization after stmt folding and before EH/AB/noreturn
16653 cleanup to get the stmt refs canonicalized. Use a bool instead
16654 of gimple_modified_p since that doesn't work for NOPs. Schedule
16655 NOPs generated by folding for removal.
16657 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16658 Alan Hayward <alan.hayward@arm.com>
16659 David Sherwood <david.sherwood@arm.com>
16661 * coretypes.h (pad_direction): New enum.
16662 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16663 (FUNCTION_ARG_PADDING): Likewise.
16664 * target.def (function_arg_padding): New hook.
16665 * targhooks.h (default_function_arg_padding): Declare.
16666 * targhooks.c (default_function_arg_padding): New function.
16667 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16668 (TARGET_FUNCTION_ARG_PADDING): ...this.
16669 * doc/tm.texi: Regenerate.
16670 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16671 instead of direction.
16672 (compute_argument_addresses): Likewise.
16673 (load_register_parameters): Likewise.
16674 (emit_library_call_value_1): Likewise.
16675 (store_one_arg): Use targetm.calls.function_arg_padding instead
16676 of FUNCTION_ARG_PADDING.
16677 (must_pass_in_stack_var_size_or_pad): Likewise.
16678 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16679 (emit_group_store): Likewise.
16680 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16681 instead of FUNCTION_ARG_PADDING.
16682 (emit_push_insn): Likewise, and propagate enum change throughout
16684 * function.h (direction): Delete.
16685 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16687 * function.c (assign_parm_find_stack_rtl): Likewise.
16688 (assign_parm_setup_block_p): Likewise.
16689 (assign_parm_setup_block): Likewise.
16690 (gimplify_parameters): Likewise.
16691 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16692 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16694 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16695 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16696 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16697 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16698 (aarch64_function_arg_padding): ...this new function.
16699 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16700 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16701 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16702 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16703 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16704 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16705 (arm_pad_arg_upward): Replace with...
16706 (arm_function_arg_padding): ...this new function.
16707 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16709 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16710 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16711 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16712 (ia64_hpux_function_arg_padding): Replace with...
16713 (ia64_function_arg_padding): ...this new function. Use pad_direction
16714 instead of direction. Check for TARGET_HPUX.
16715 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16716 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16717 (iq2000_function_arg_padding): New function.
16718 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16719 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16720 (mips_function_arg_padding): ...this new function.
16721 (mips_pad_reg_upward): Update accordingly.
16722 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16723 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16724 targetm.calls.function_arg_padding.
16725 (FUNCTION_ARG_PADDING): Delete.
16726 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16727 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16728 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16729 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16730 (nios2_block_reg_padding): Return pad_direction instead of direction.
16731 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16732 instead of direction.
16733 (nios2_function_arg_padding): Likewise. Make static.
16734 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16735 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16736 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16737 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16738 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16739 (pa_function_arg_padding): Make static. Return pad_direction instead
16741 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16742 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16743 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16744 instead of direction. Use targetm.calls.function_arg_padding.
16745 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16746 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16747 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16748 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16749 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16751 (function_arg_padding): Rename to...
16752 (rs6000_function_arg_padding): ...this. Make static. Return
16753 pad_direction instead of direction.
16754 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16755 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16756 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16757 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16758 instead of direction. Use targetm.calls.function_arg_padding.
16759 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16760 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16761 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16762 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16763 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16764 (function_arg_padding): Rename to...
16765 (rs6000_function_arg_padding): ...this. Make static. Return
16766 pad_direction instead of direction.
16767 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16768 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16769 * config/s390/s390.c (s390_function_arg_padding): New function.
16770 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16771 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16772 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16773 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16774 (function_arg_padding): Rename to...
16775 (sparc_function_arg_padding): ...this. Make static. Return
16776 pad_direction instead of direction.
16777 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16778 * config/spu/spu.c (spu_function_arg_padding): New function.
16779 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16780 * system.h (FUNCTION_ARG_PADDING): Poison.
16782 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16783 Alan Hayward <alan.hayward@arm.com>
16784 David Sherwood <david.sherwood@arm.com>
16786 * target.def (modes_tieable_p): New hook.
16787 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16788 (TARGET_MODES_TIEABLE_P): ...this.
16789 * doc/tm.texi.in: Regenerate.
16790 * hooks.h (hook_bool_mode_mode_true): Declare.
16791 * hooks.c (hook_bool_mode_mode_true): New function.
16792 * combine.c (subst): Use targetm.modes_tieable_p instead of
16794 * dse.c (find_shift_sequence): Likewise.
16795 * expmed.c (extract_low_bits): Likewise.
16796 * lower-subreg.c: Include target.h.
16797 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16799 * rtlanal.c (rtx_cost): Likewise.
16800 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16801 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16802 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16803 (TARGET_MODES_TIEABLE_P): Redefine.
16804 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16805 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16806 (TARGET_MODES_TIEABLE_P): Redefine.
16807 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16808 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16809 (arc_modes_tieable_p): New function.
16810 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16811 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16812 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16813 (arm_modes_tieable_p): Make static.
16814 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16815 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16816 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16817 (TARGET_MODES_TIEABLE_P): Redefine.
16818 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16819 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16820 (TARGET_MODES_TIEABLE_P): Redefine.
16821 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16822 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16823 (cr16_modes_tieable_p): New function.
16824 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16825 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16826 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16827 (TRULY_NOOP_TRUNCATION): Update comment.
16828 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16829 (TRULY_NOOP_TRUNCATION): Update comment.
16830 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16831 (frv_modes_tieable_p): New function.
16832 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16833 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16834 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16835 (TARGET_MODES_TIEABLE_P): Redefine.
16836 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16837 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16838 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16839 (TARGET_MODES_TIEABLE_P): Redefine.
16840 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16841 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16842 (ia64_modes_tieable_p): New function.
16843 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16844 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16845 (iq2000_modes_tieable_p): New function.
16846 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16847 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16848 (lm32_modes_tieable_p): New function.
16849 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16850 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16851 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16852 (TARGET_MODES_TIEABLE_P): Redefine.
16853 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16854 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16855 (m32r_modes_tieable_p): New function.
16856 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16857 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16858 (m68k_modes_tieable_p): New function.
16859 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16860 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16861 (mcore_modes_tieable_p): New function.
16862 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16863 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16865 (TARGET_MODES_TIEABLE_P): Redefine.
16866 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16867 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16868 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16869 (TARGET_MODES_TIEABLE_P): Redefine.
16870 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16871 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16872 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16873 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16874 (mn10300_modes_tieable_p): ...this and make static.
16875 (TARGET_MODES_TIEABLE_P): Redefine.
16876 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16877 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16878 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16879 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16880 (msp430_modes_tieable_p): Make static.
16881 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16882 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16883 (TARGET_MODES_TIEABLE_P): Redefine.
16884 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16885 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16886 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16887 (TARGET_MODES_TIEABLE_P): Redefine.
16888 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16889 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16890 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16891 (TARGET_MODES_TIEABLE_P): Redefine.
16892 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16893 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16894 (pdp11_modes_tieable_p): New function.
16895 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16896 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16897 (rs6000_modes_tieable_p): New function.
16898 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16899 * config/powerpcspe/powerpcspe.md: Update comment.
16900 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16901 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16902 (TARGET_MODES_TIEABLE_P): Redefine.
16903 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16904 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16905 (rl78_modes_tieable_p): New function.
16906 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16907 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16908 (rs6000_modes_tieable_p): New function.
16909 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16910 * config/rs6000/rs6000.md: Update comment.
16911 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16912 * config/rx/rx.c (rx_modes_tieable_p): New function.
16913 (TARGET_MODES_TIEABLE_P): Redefine.
16914 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16915 * config/s390/s390.c (s390_modes_tieable_p): New function.
16916 (TARGET_MODES_TIEABLE_P): Redefine.
16917 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16918 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16919 (sh_modes_tieable_p): New function.
16920 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16921 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16922 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16923 (sparc_modes_tieable_p): Make static.
16924 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16925 * config/spu/spu.c (spu_modes_tieable_p): New function.
16926 (TARGET_MODES_TIEABLE_P): Redefine.
16927 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16928 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16929 (TARGET_MODES_TIEABLE_P): Redefine.
16930 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16931 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16932 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16933 * config/v850/v850.c (v850_modes_tieable_p): New function.
16934 (TARGET_MODES_TIEABLE_P): Redefine.
16935 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16936 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16937 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16938 (visium_modes_tieable_p): New function.
16939 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16940 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16941 (xtensa_modes_tieable_p): New function.
16942 * system.h (MODES_TIEABLE_P): Poison.
16944 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16945 Alan Hayward <alan.hayward@arm.com>
16946 David Sherwood <david.sherwood@arm.com>
16948 * target.def (hard_regno_mode_ok): New hook.
16949 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16950 (TARGET_HARD_REGNO_MODE_OK): ...this.
16951 * doc/tm.texi.in: Regenerate.
16952 * hooks.h (hook_bool_uint_mode_true): Declare.
16953 * hooks.c (hook_bool_uint_mode_true): New function.
16954 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16955 HARD_REGNO_MODE_OK.
16956 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16957 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16958 instead of HARD_REGNO_MODE_OK.
16959 * caller-save.c: Include target.h.
16960 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16961 HARD_REGNO_MODE_OK.
16962 * combine.c (can_combine_p): Likewise.
16963 (combinable_i3pat): Likewise.
16964 (can_change_dest_mode): Likewise.
16965 * expr.c (init_expr_target): Likewise.
16966 (convert_move): Likewise.
16967 (convert_modes): Likewise.
16968 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16969 (setup_prohibited_mode_move_regs): Likewise.
16970 * ira.h (target_ira): Likewise.
16971 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16972 * lra-constraints.c (process_alt_operands): Likewise.
16973 (split_reg): Likewise.
16974 * recog.c (peep2_find_free_register): Likewise.
16975 * ree.c (combine_reaching_defs): Likewise.
16976 * regcprop.c (maybe_mode_change): Likewise.
16977 * reginfo.c (init_reg_sets_1): Likewise.
16978 (choose_hard_reg_mode): Likewise.
16979 (simplifiable_subregs): Likewise.
16980 * regrename.c (check_new_reg_p): Likewise.
16981 * reload.c (find_valid_class): Likewise.
16982 (find_valid_class_1): Likewise.
16983 (reload_inner_reg_of_subreg): Likewise.
16984 (push_reload): Likewise.
16985 (combine_reloads): Likewise.
16986 (find_dummy_reload): Likewise.
16987 (find_reloads): Likewise.
16988 * reload1.c (find_reg): Likewise.
16989 (set_reload_reg): Likewise.
16990 (allocate_reload_reg): Likewise.
16991 (choose_reload_regs): Likewise.
16992 (reload_adjust_reg_for_temp): Likewise.
16993 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16994 (simplify_subreg_regno): Likewise.
16995 * sel-sched.c (init_regs_for_mode): Likewise.
16996 * varasm.c (make_decl_rtl): Likewise.
16997 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16998 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16999 HARD_REGNO_MODE_OK.
17000 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17001 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17002 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17003 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17004 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17005 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17006 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17007 (arc_mode_class): Delete.
17008 (HARD_REGNO_MODE_OK): Delete.
17009 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17010 (arc_hard_regno_mode_ok): Rename old array to...
17011 (arc_hard_regno_mode_ok_modes): ...this.
17012 (arc_conditional_register_usage): Update accordingly.
17013 (arc_mode_class): Make static.
17014 (arc_hard_regno_mode_ok): New function.
17015 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17016 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17017 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17018 (arm_hard_regno_mode_ok): Make static.
17019 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17020 HARD_REGNO_MODE_OK.
17021 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17022 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17023 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17025 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17026 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17027 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17028 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17029 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17030 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17031 * config/bfin/predicates.md (valid_reg_operand): Use
17032 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17033 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17034 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17035 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17036 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17037 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17038 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17039 (cr16_hard_regno_mode_ok): Make static and return a bool.
17040 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17041 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17042 (cris_hard_regno_mode_ok): New function.
17043 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17044 (epiphany_mode_class): Delete.
17045 (HARD_REGNO_MODE_OK): Delete.
17046 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17047 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17048 (hard_regno_mode_ok): Rename to...
17049 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17050 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17051 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17052 HARD_REGNO_MODE_OK.
17053 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17054 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17055 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17056 (frv_hard_regno_mode_ok): Make static and return a bool.
17057 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17058 HARD_REGNO_MODE_OK.
17059 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17060 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17061 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17062 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17064 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17065 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17066 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17067 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17069 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17070 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17071 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17072 (ia64_hard_regno_mode_ok): New function.
17073 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17074 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17075 (iq2000_hard_regno_mode_ok): New function.
17076 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17077 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17078 (lm32_hard_regno_mode_ok): New function.
17079 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17080 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17081 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17082 instead of HARD_REGNO_MODE_OK.
17083 (m32c_hard_regno_ok): Rename to...
17084 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17085 (m32c_cannot_change_mode_class): Update accordingly.
17086 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17087 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17088 (m32r_mode_class): Delete.
17089 (HARD_REGNO_MODE_OK): Delete.
17090 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17091 (m32r_hard_regno_mode_ok): Rename to...
17092 (m32r_hard_regno_modes): ...this.
17093 (m32r_mode_class): Make static.
17094 (m32r_hard_regno_mode_ok): New function.
17095 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17096 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17097 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17098 (m68k_hard_regno_mode_ok): Make static.
17099 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17100 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17101 (mcore_hard_regno_mode_ok): New function.
17102 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17103 (HARD_REGNO_MODE_OK): Delete.
17104 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17106 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17107 (microblaze_hard_regno_mode_ok): New function.
17108 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17109 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17110 (mips_hard_regno_mode_ok): Delete.
17111 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17112 (mips_hard_regno_mode_ok_p): ...this and make static.
17113 (mips_hard_regno_mode_ok_p): Rename to...
17114 (mips_hard_regno_mode_ok_uncached): ...this.
17115 (mips_hard_regno_mode_ok): New function.
17116 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17117 of HARD_REGNO_MODE_OK.
17118 (mips_option_override): Update after above name changes.
17119 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17120 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17121 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17122 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17123 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17124 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17125 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17126 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17127 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17128 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17129 (msp430_hard_regno_mode_ok): Make static and return a bool.
17130 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17131 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17132 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17134 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17135 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17136 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17137 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17138 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17139 (PA_HARD_REGNO_MODE_OK): ...this
17140 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17141 (PA_HARD_REGNO_MODE_OK): ...this.
17142 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17143 (pa_hard_regno_mode_ok): New function.
17144 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17145 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17146 (pdp11_hard_regno_mode_ok): New function.
17147 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17148 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17150 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17152 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17153 (rs6000_hard_regno_mode_ok): Rename to...
17154 (rs6000_hard_regno_mode_ok_uncached): ...this.
17155 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17156 (rs6000_hard_regno_mode_ok): New function.
17157 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17158 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17159 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17160 (riscv_hard_regno_mode_ok): ...this and make static.
17161 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17162 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17163 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17164 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17165 (rl78_hard_regno_mode_ok): Make static and return bool.
17166 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17167 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17169 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17170 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17171 (rs6000_hard_regno_mode_ok): Rename to...
17172 (rs6000_hard_regno_mode_ok_uncached): ...this.
17173 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17174 (rs6000_hard_regno_mode_ok): New function.
17175 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17176 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17177 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17178 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17179 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17180 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17181 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17182 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17183 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17184 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17185 (sh_hard_regno_mode_ok): Make static.
17186 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17187 instead of HARD_REGNO_MODE_OK.
17188 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17189 (sparc_mode_class): Delete.
17190 (HARD_REGNO_MODE_OK): Delete.
17191 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17192 (hard_regno_mode_classes): Make static.
17193 (sparc_mode_class): Likewise.
17194 (sparc_hard_regno_mode_ok): New function.
17195 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17196 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17197 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17199 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17200 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17201 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17202 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17203 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17204 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17205 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17206 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17207 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17208 (visium_hard_regno_mode_ok): New function.
17209 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17210 instead of HARD_REGNO_MODE_OK.
17211 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17212 (HARD_REGNO_MODE_OK): Delete.
17213 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17214 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17215 (xtensa_option_override): Update accordingly.
17216 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17217 (xtensa_hard_regno_mode_ok): New function.
17218 * system.h (HARD_REGNO_MODE_OK): Poison.
17220 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17221 Alan Hayward <alan.hayward@arm.com>
17222 David Sherwood <david.sherwood@arm.com>
17224 * target.def (hard_regno_call_part_clobbered): New hook.
17225 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17226 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17227 * doc/tm.texi: Regenerate.
17228 * hooks.h (hook_bool_uint_mode_false): Declare.
17229 * hooks.c (hook_bool_uint_mode_false): New function.
17230 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17231 * cselib.c (cselib_process_insn): Use
17232 targetm.hard_regno_call_part_clobbered instead of
17233 HARD_REGNO_CALL_PART_CLOBBERED.
17234 * ira-conflicts.c (ira_build_conflicts): Likewise.
17235 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17236 * lra-constraints.c (need_for_call_save_p): Likewise.
17237 * lra-lives.c: Include target.h.
17238 (check_pseudos_live_through_calls): Use
17239 targetm.hard_regno_call_part_clobbered instead of
17240 HARD_REGNO_CALL_PART_CLOBBERED.
17241 * regcprop.c: Include target.h.
17242 (copyprop_hardreg_forward_1): Use
17243 targetm.hard_regno_call_part_clobbered instead of
17244 HARD_REGNO_CALL_PART_CLOBBERED.
17245 * reginfo.c (choose_hard_reg_mode): Likewise.
17246 * regrename.c (check_new_reg_p): Likewise.
17247 * reload.c (find_equiv_reg): Likewise.
17248 * reload1.c (emit_reload_insns): Likewise.
17249 * sched-deps.c (deps_analyze_insn): Likewise.
17250 * sel-sched.c (init_regs_for_mode): Likewise.
17251 (mark_unavailable_hard_regs): Likewise.
17252 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17253 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17254 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17256 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17257 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17258 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17260 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17262 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17263 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17264 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17266 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17267 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17268 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17270 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17271 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17273 * config/powerpcspe/powerpcspe.c
17274 (rs6000_hard_regno_call_part_clobbered): New function.
17275 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17276 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17277 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17279 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17280 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17281 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17283 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17284 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17285 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17287 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17288 Alan Hayward <alan.hayward@arm.com>
17289 David Sherwood <david.sherwood@arm.com>
17291 * rtl.h (subreg_memory_offset): Declare.
17292 * emit-rtl.c (subreg_memory_offset): New function.
17293 * expmed.c (store_bit_field_1): Use it.
17294 * expr.c (undefined_operand_subword_p): Likewise.
17295 * simplify-rtx.c (simplify_subreg): Likewise.
17297 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17299 PR rtl-optimization/57448
17302 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17303 using atomic_load pattern.
17304 (expand_atomic_store): Likewise.
17306 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17309 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17310 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17311 handling. Use replace_call_with_value with NULL instead of
17312 gsi_replace, unlink_stmt_vdef and release_defs.
17314 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17317 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17319 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17322 * rtl.h (emit_library_call_value_1): Declare.
17323 (emit_library_call): Replace declaration with a series of overloads.
17324 Remove the parameter count argument.
17325 (emit_library_call_value): Likewise.
17326 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17327 with an "rtx_mode_t *".
17328 (emit_library_call_value): Delete.
17329 (emit_library_call): Likewise.
17330 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17331 (asan_emit_allocas_unpoison): Likewise.
17332 * builtins.c (expand_builtin_powi): Likewise.
17333 (expand_asan_emit_allocas_unpoison): Likewise.
17334 * cfgexpand.c (expand_main_function): Likewise.
17335 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17336 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17337 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17338 * config/arm/arm.c (arm_trampoline_init): Likewise.
17339 (arm_call_tls_get_addr): Likewise.
17340 (arm_expand_divmod_libfunc): Likewise.
17341 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17342 (smulsi3_highpart): Likewise.
17343 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17344 (c6x_expand_compare): Likewise.
17345 (c6x_expand_movmem): Likewise.
17346 * config/frv/frv.c (frv_trampoline_init): Likewise.
17347 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17348 (ix86_expand_divmod_libfunc): Likewise.
17349 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17350 (ia64_expand_compare): Likewise.
17351 (ia64_profile_hook): Likewise.
17352 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17353 (nonlocal_goto): Likewise.
17354 (restore_stack_nonlocal): Likewise.
17355 * config/m32r/m32r.c (block_move_call): Likewise.
17356 (m32r_trampoline_init): Likewise.
17357 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17358 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17359 (m68k_call_m68k_read_tp): Likewise.
17360 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17361 (microblaze_expand_divide): Likewise.
17362 * config/mips/mips.h (mips_args): Likewise.
17363 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17364 (MIPS_ICACHE_SYNC): Likewise.
17365 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17366 (nios2_trampoline_init): Likewise.
17367 * config/pa/pa.c (hppa_tls_call): Likewise.
17368 (pa_trampoline_init): Likewise.
17369 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17370 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17371 (expand_strn_compare): Likewise.
17372 (rs6000_generate_compare): Likewise.
17373 (rs6000_expand_float128_convert): Likewise.
17374 (output_profile_hook): Likewise.
17375 (rs6000_trampoline_init): Likewise.
17376 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17377 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17378 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17379 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17380 (rs6000_generate_compare): Likewise.
17381 (rs6000_expand_float128_convert): Likewise.
17382 (output_profile_hook): Likewise.
17383 (rs6000_trampoline_init): Likewise.
17384 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17385 * config/sh/sh.c (sh_trampoline_init): Likewise.
17386 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17387 (sparc_emit_float_lib_cmp): Likewise.
17388 (sparc32_initialize_trampoline): Likewise.
17389 (sparc64_initialize_trampoline): Likewise.
17390 (sparc_profile_hook): Likewise.
17391 * config/spu/spu.c (ea_load_store): Likewise.
17392 * config/spu/spu.md (floatunssidf2): Likewise.
17393 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17394 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17395 * config/visium/visium.c (expand_block_move_4): Likewise.
17396 (expand_block_move_2): Likewise.
17397 (expand_block_move_1): Likewise.
17398 (expand_block_set_4): Likewise.
17399 (expand_block_set_2): Likewise.
17400 (expand_block_set_1): Likewise.
17401 (visium_trampoline_init): Likewise.
17402 (visium_profile_hook): Likewise.
17403 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17404 (xtensa_setup_frame_addresses): Likewise.
17405 (xtensa_trampoline_init): Likewise.
17406 * except.c (sjlj_emit_function_enter): Likewise.
17407 (sjlj_emit_function_exit): Likewise.
17408 * explow.c (allocate_dynamic_stack_space): Likewise.
17409 (probe_stack_range): Likewise.
17410 * expr.c (convert_mode_scalar): Likewise.
17411 * optabs.c (expand_binop): Likewise.
17412 (expand_twoval_binop_libfunc): Likewise.
17413 (expand_unop): Likewise.
17414 (prepare_cmp_insn): Likewise.
17415 (prepare_float_lib_cmp): Likewise.
17416 (expand_float): Likewise.
17417 (expand_fix): Likewise.
17418 (expand_fixed_convert): Likewise.
17419 (maybe_emit_sync_lock_test_and_set): Likewise.
17420 (expand_atomic_compare_and_swap): Likewise.
17421 (expand_mem_thread_fence): Likewise.
17422 (expand_atomic_fetch_op): Likewise.
17424 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17426 * doc/generic.texi (OpenACC): Adjust URL.
17427 * doc/invoke.texi (C Dialect Options): Ditto.
17429 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17431 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17432 predicate for operand 1. Add (m,<S>) constraint.
17433 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17434 Prevent memory operand 1 with register operand 2.
17436 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17438 PR rtl-optimization/82024
17439 * combine.c (try_combine): If the combination result is a PARALLEL,
17440 and we only need to retain the SET in there that would be placed
17441 at I2, check that we can place that at I3 instead, before doing so.
17443 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17446 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17448 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17449 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17452 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17453 Jeff Law <law@redhat.com>
17455 * varasm.c (bss_initializer_p): Do not put constants into .bss
17456 (categorize_decl_for_section): Handle bss_initializer_p returning
17457 false when DECL_INITIAL is NULL.
17459 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17462 * config/s390/s390.c (s390_can_inline_p): New function.
17464 2017-09-01 Jeff Law <law@redhat.com>
17466 PR tree-optimization/82052
17467 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17468 Always initialize the returned slot after a hash table miss
17469 when INSERT is true.
17471 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17473 * config/s390/s390.md (mem_signal_fence): Remove.
17474 * doc/md.texi (mem_signal_fence): Remove.
17475 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17477 * target-insns.def (mem_signal_fence): Remove.
17479 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17482 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17485 * asan.c (create_odr_indicator): Strip name encoding from assembler
17486 name before appending it after __odr_asan_.
17488 2017-09-01 Martin Liska <mliska@suse.cz>
17490 PR tree-optimization/82059
17491 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17492 frequency only when an edge is redirected.
17494 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17496 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17497 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17498 (arc_conditional_register_usage): Remove ARC600 lp_count
17500 (arc_file_start): Emit Tag_ARC_CPU_variation.
17501 (arc_can_use_doloop_p): New conditions to use ZOLs.
17502 (hwloop_fail): New function.
17503 (hwloop_optimize): Likewise.
17504 (hwloop_pattern_reg): Likewise.
17505 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17506 (arc_reorg_loops): New function, calls reorg_loops.
17507 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17508 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17510 (arc_loop_hazard): Remove function, functionality moved into
17512 (arc_hazard): Remove arc_loop_hazard call.
17513 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17514 into hwloop_optimize.
17515 (arc_label_align): Remove ZOL handling.
17516 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17517 * config/arc/arc.md (doloop_begin): Remove pattern.
17518 (doloop_begin_i): Likewise.
17519 (doloop_end_i): Likewise.
17520 (doloop_fallback): Likewise.
17521 (doloop_fallback_m): Likewise.
17522 (doloop_end): Reimplement expand.
17523 (arc_lp): New pattern for LP instruction.
17524 (loop_end): New pattern.
17525 (loop_fail): Likewise.
17526 (decrement_and_branch_until_zero): Likewise.
17527 * config/arc/arc.opt (mlpc-width): New option.
17528 * doc/invoke.texi (mlpc-width): Document option.
17530 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17532 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17533 (arc_ccfsm_advance): Fix checking for delay slots.
17534 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17536 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17538 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17540 (movhi_insn): Update store instruction constraint which are saving
17541 6-bit short immediates.
17542 (movsi_insn): Consider also short scaled load operations.
17543 (zero_extendhisi2_i): Use Usd constraint instead of T.
17544 (extendhisi2_i): Add q constraint.
17545 (arc_clzsi2): Add type and length attributes.
17546 (arc_ctzsi2): Likewise.
17547 * config/arc/constraints.md (Usc): Update constraint, the
17548 assembler can parse two relocations for a single instruction.
17550 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17552 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17553 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17555 2017-08-31 Olivier Hainque <hainque@adacore.com>
17557 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17558 match as powerpc-wrs-vxworks*.
17560 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17562 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17563 register constraint for by-element operand.
17564 (aarch64_mls_elt_merge<mode>): Likewise.
17566 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17568 * config/arc/arc.c (arc_can_follow_jump): Check for short
17571 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17573 * config.gcc: Use g.opt for arc.
17574 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17575 functionality moved to ...
17576 (legitimate_scaled_address_p): New function, ...here.
17577 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17578 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17579 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17581 (arc_override_options): Handle G option.
17582 (arc_output_pic_addr_const): Correct function definition.
17583 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17584 (arc_decl_anon_ns_mem_p): Delete.
17585 (arc_in_small_data_p): Overhaul this function to take into
17586 consideration the value given via G option.
17587 (arc_rewrite_small_data_1): Renamed and corrected old
17588 arc_rewrite_small_data function.
17589 (arc_rewrite_small_data): New function.
17590 (small_data_pattern): Don't use pic_offset_table_rtx.
17591 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17592 * config/arc/simdext.md (movmisalignv2hi): Use
17593 prepare_move_operands function.
17595 (movmisalign*): Likewise.
17596 * doc/invoke.texi (ARC options): Document -G option.
17598 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17600 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17602 * config/arc/arc.c (arc_print_operand): Output scalled address for
17603 sdata whenever is possible.
17604 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17605 load/stores are available.
17606 (compact_sda_memory_operand): Check for the alignment required by
17607 code density instructions.
17608 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17610 * config/arc/constraints.md (Usd): Update constraint.
17611 (Us0): New constraint.
17612 (Usc): Update constraint.
17614 2017-08-31 Richard Biener <rguenther@suse.de>
17616 PR middle-end/82054
17617 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17618 function only once.
17620 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17622 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17623 Resize type_signature.
17625 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17626 Alan Hayward <alan.hayward@arm.com>
17627 David Sherwood <david.sherwood@arm.com>
17629 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17630 subregs whose inner modes can be stored in GPRs.
17631 (aarch64_classify_index): Likewise.
17633 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17634 Alan Hayward <alan.hayward@arm.com>
17635 David Sherwood <david.sherwood@arm.com>
17637 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17638 (V_INT_EQUIV): ...this.
17639 (v_cmp_result): Rename to...
17640 (v_int_equiv): ...this.
17641 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17642 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17643 (copysign<mode>3): Likewise.
17644 (aarch64_simd_bsl<mode>_internal): Likewise.
17645 (aarch64_simd_bsl<mode>): Likewise.
17646 (vec_cmp<mode><mode>): Likewise.
17647 (vcond<mode><mode>): Likewise.
17648 (vcond<v_cmp_mixed><mode>): Likewise.
17649 (vcondu<mode><v_cmp_mixed>): Likewise.
17650 (aarch64_cm<optab><mode>): Likewise.
17651 (aarch64_cmtst<mode>): Likewise.
17652 (aarch64_fac<optab><mode>): Likewise.
17653 (vec_perm_const<mode>): Likewise.
17654 (vcond_mask_<mode><v_cmp_result>): Rename to...
17655 (vcond_mask_<mode><v_int_equiv>): ...this.
17656 (vec_cmp<mode><v_cmp_result>): Rename to...
17657 (vec_cmp<mode><v_int_equiv>): ...this.
17659 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17660 Alan Hayward <alan.hayward@arm.com>
17661 David Sherwood <david.sherwood@arm.com>
17663 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17665 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17666 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17667 (UNSPEC_LD4_DREG): New unspecs.
17668 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17669 (aarch64_ld2<mode>_dreg_be): Replace with...
17670 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17672 (aarch64_ld3<mode>_dreg_le)
17673 (aarch64_ld3<mode>_dreg_be): Replace with...
17674 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17676 (aarch64_ld4<mode>_dreg_le)
17677 (aarch64_ld4<mode>_dreg_be): Replace with...
17678 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17681 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17683 PR tree-optimization/81987
17684 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17685 insert an initializer in a location not dominated by the stride
17688 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17690 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17691 on the entire header of the finally block in the fallthru case.
17693 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17695 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17697 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17699 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17700 rs6000_emit_move_from_cr and call renamed function.
17701 (rs6000_emit_prologue): Call renamed functions.
17702 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17703 movesi_from_cr, remove volatile CRs.
17705 2017-08-30 Jon Beniston <jon@beniston.com>
17706 Richard Biener <rguenther@suse.de>
17708 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17709 of VECTOR_MODE_P check.
17710 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17711 element vector types.
17713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17715 * df.h (df_read_modify_subreg_p): Remove in favor of...
17716 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17717 const_rtx instead of an rtx.
17718 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17719 * df-problems.c (df_word_lr_mark_ref): Likewise.
17720 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17721 (mark_pseudo_reg_dead): Likewise.
17722 (mark_ref_dead): Likewise.
17723 * reginfo.c (init_subregs_of_mode): Likewise.
17724 * sched-deps.c (sched_analyze_1): Likewise.
17725 * df-scan.c (df_def_record_1): Likewise.
17726 (df_uses_record): Likewise.
17727 (df_read_modify_subreg_p): Remove in favor of...
17728 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17729 const_rtx instead of an rtx.
17731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17732 Alan Hayward <alan.hayward@arm.com>
17733 David Sherwood <david.sherwood@arm.com>
17735 * rtl.h (partial_subreg_p): New function.
17736 * caller-save.c (save_call_clobbered_regs): Use it.
17737 * calls.c (expand_call): Likewise.
17738 * combine.c (combinable_i3pat): Likewise.
17739 (simplify_set): Likewise.
17740 (make_extraction): Likewise.
17741 (make_compound_operation_int): Likewise.
17742 (gen_lowpart_or_truncate): Likewise.
17743 (force_to_mode): Likewise.
17744 (make_field_assignment): Likewise.
17745 (reg_truncated_to_mode): Likewise.
17746 (record_truncated_value): Likewise.
17747 (move_deaths): Likewise.
17748 * cse.c (record_jump_cond): Likewise.
17749 (cse_insn): Likewise.
17750 * cselib.c (cselib_lookup_1): Likewise.
17751 * expmed.c (extract_bit_field_using_extv): Likewise.
17752 * function.c (assign_parm_setup_reg): Likewise.
17753 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17754 * ira-build.c (create_insn_allocnos): Likewise.
17755 * lra-coalesce.c (merge_pseudos): Likewise.
17756 * lra-constraints.c (match_reload): Likewise.
17757 (simplify_operand_subreg): Likewise.
17758 (curr_insn_transform): Likewise.
17759 * lra-lives.c (process_bb_lives): Likewise.
17760 * lra.c (new_insn_reg): Likewise.
17761 (lra_substitute_pseudo): Likewise.
17762 * regcprop.c (mode_change_ok): Likewise.
17763 (maybe_mode_change): Likewise.
17764 (copyprop_hardreg_forward_1): Likewise.
17765 * reload.c (push_reload): Likewise.
17766 (find_reloads): Likewise.
17767 (find_reloads_subreg_address): Likewise.
17768 * reload1.c (alter_reg): Likewise.
17769 (eliminate_regs_1): Likewise.
17770 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17772 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17774 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17775 back to if statements, including unpack.
17777 2017-08-30 Martin Liska <mliska@suse.cz>
17779 PR inline-asm/82001
17780 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17782 (func_checker::compare_asm_inputs_outputs): ... this function.
17783 (func_checker::compare_gimple_asm): Use the function to compare
17784 also ASM constrains.
17785 * ipa-icf-gimple.h: Rename the function.
17787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17788 Alan Hayward <alan.hayward@arm.com>
17789 David Sherwood <david.sherwood@arm.com>
17791 * coretypes.h (complex_mode): New type.
17792 * gdbhooks.py (build_pretty_printer): Handle it.
17793 * machmode.h (complex_mode): New class.
17794 (complex_mode::includes_p): New function.
17795 (is_complex_int_mode): Likewise.
17796 (is_complex_float_mode): Likewise.
17797 * genmodes.c (get_mode_class): Handle complex mode classes.
17798 * function.c (expand_function_end): Use is_complex_int_mode.
17800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17801 Alan Hayward <alan.hayward@arm.com>
17802 David Sherwood <david.sherwood@arm.com>
17804 * coretypes.h (scalar_mode_pod): New typedef.
17805 * gdbhooks.py (build_pretty_printer): Handle it.
17806 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17807 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17808 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17809 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17810 as_a <scalar_mode>.
17812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17813 Alan Hayward <alan.hayward@arm.com>
17814 David Sherwood <david.sherwood@arm.com>
17816 * machmode.h (mode_for_vector): Take a scalar_mode instead
17818 * stor-layout.c (mode_for_vector): Likewise.
17819 * explow.c (promote_mode): Use as_a <scalar_mode>.
17820 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17823 Alan Hayward <alan.hayward@arm.com>
17824 David Sherwood <david.sherwood@arm.com>
17826 * target.def (preferred_simd_mode): Take a scalar_mode
17827 instead of a machine_mode.
17828 * targhooks.h (default_preferred_simd_mode): Likewise.
17829 * targhooks.c (default_preferred_simd_mode): Likewise.
17830 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17831 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17832 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17833 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17834 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17835 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17836 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17837 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17839 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17840 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17841 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17842 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17843 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17844 * doc/tm.texi: Regenerate.
17845 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17849 Alan Hayward <alan.hayward@arm.com>
17850 David Sherwood <david.sherwood@arm.com>
17852 * target.def (scalar_mode_supported_p): Take a scalar_mode
17853 instead of a machine_mode.
17854 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17855 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17856 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17857 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17858 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17859 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17860 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17861 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17862 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17863 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17864 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17865 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17866 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17867 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17869 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17870 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17871 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17872 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17873 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17875 * doc/tm.texi: Regenerate.
17877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17878 Alan Hayward <alan.hayward@arm.com>
17879 David Sherwood <david.sherwood@arm.com>
17881 * coretypes.h (opt_scalar_mode): New typedef.
17882 * gdbhooks.py (build_pretty_printers): Handle it.
17883 * machmode.h (mode_iterator::get_2xwider): Add overload for
17885 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17887 * expr.c (convert_mode_scalar): Likewise.
17888 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17889 * optabs.c (expand_float): Likewise.
17890 (expand_fix): Likewise.
17891 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17894 Alan Hayward <alan.hayward@arm.com>
17895 David Sherwood <david.sherwood@arm.com>
17897 * optabs.c (expand_float): Explicitly check for scalars before
17898 using a branching expansion.
17899 (expand_fix): Likewise.
17901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17902 Alan Hayward <alan.hayward@arm.com>
17903 David Sherwood <david.sherwood@arm.com>
17905 * expr.c (convert_mode): Split scalar handling out into...
17906 (convert_mode_scalar): ...this new function. Treat the modes
17909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17910 Alan Hayward <alan.hayward@arm.com>
17911 David Sherwood <david.sherwood@arm.com>
17913 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17915 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17918 Alan Hayward <alan.hayward@arm.com>
17919 David Sherwood <david.sherwood@arm.com>
17921 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17922 rather than a machine_mode.
17923 (fixed_from_string): Likewise.
17924 (fixed_convert): Likewise.
17925 (fixed_convert_from_int): Likewise.
17926 (fixed_convert_from_real): Likewise.
17927 (real_convert_from_fixed): Likewise.
17928 * fixed-value.c (fixed_from_double_int): Likewise.
17929 (fixed_from_string): Likewise.
17930 (fixed_convert): Likewise.
17931 (fixed_convert_from_int): Likewise.
17932 (fixed_convert_from_real): Likewise.
17933 (real_convert_from_fixed): Likewise.
17934 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17937 Alan Hayward <alan.hayward@arm.com>
17938 David Sherwood <david.sherwood@arm.com>
17940 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17941 of separate mode class checks. Do not allow vector modes here.
17942 (immed_wide_int_const): Use as_a <scalar_mode>.
17943 * explow.c (trunc_int_for_mode): Likewise.
17944 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17945 (wi::shwi): Likewise.
17946 (wi::min_value): Likewise.
17947 (wi::max_value): Likewise.
17948 * dwarf2out.c (loc_descriptor): Likewise.
17949 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17950 for CONST_WIDE_INT.
17952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17953 Alan Hayward <alan.hayward@arm.com>
17954 David Sherwood <david.sherwood@arm.com>
17956 * tree.h (SCALAR_TYPE_MODE): New macro.
17957 * expr.c (expand_expr_addr_expr_1): Use it.
17958 (expand_expr_real_2): Likewise.
17959 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17960 (fold_convert_const_fixed_from_int): Likewise.
17961 (fold_convert_const_fixed_from_real): Likewise.
17962 (native_encode_fixed): Likewise
17963 (native_encode_complex): Likewise
17964 (native_encode_vector): Likewise.
17965 (native_interpret_fixed): Likewise.
17966 (native_interpret_real): Likewise.
17967 (native_interpret_complex): Likewise.
17968 (native_interpret_vector): Likewise.
17969 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17970 (simd_clone_adjust_argument_types): Likewise.
17971 (simd_clone_init_simd_arrays): Likewise.
17972 (simd_clone_adjust): Likewise.
17973 * stor-layout.c (layout_type): Likewise.
17974 * tree.c (build_minus_one_cst): Likewise.
17975 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17976 * tree-inline.c (estimate_move_cost): Likewise.
17977 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17978 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17979 (vectorizable_reduction): Likewise.
17980 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17981 (vect_recog_mixed_size_cond_pattern): Likewise.
17982 (check_bool_pattern): Likewise.
17983 (adjust_bool_pattern): Likewise.
17984 (search_type_for_mask_1): Likewise.
17985 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17986 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17987 (vectorizable_load): Likewise.
17988 (vectorizable_store): Likewise.
17989 * ubsan.c (ubsan_encode_value): Likewise.
17990 * varasm.c (output_constant): Likewise.
17992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17993 Alan Hayward <alan.hayward@arm.com>
17994 David Sherwood <david.sherwood@arm.com>
17996 * coretypes.h (scalar_mode): New class.
17997 * machmode.h (scalar_mode): Likewise.
17998 (scalar_mode::includes_p): New function.
17999 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18000 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18001 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18002 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18003 * expmed.c (store_bit_field_1): Likewise.
18004 (extract_bit_field_1): Likewise.
18005 * expr.c (write_complex_part): Likewise.
18006 (read_complex_part): Likewise.
18007 (emit_move_complex_push): Likewise.
18008 (expand_expr_real_2): Likewise.
18009 * function.c (assign_parm_setup_reg): Likewise.
18010 (assign_parms_unsplit_complex): Likewise.
18011 * optabs.c (expand_binop): Likewise.
18012 * rtlanal.c (subreg_get_info): Likewise.
18013 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18014 * varasm.c (output_constant_pool_2): Likewise.
18016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18017 Alan Hayward <alan.hayward@arm.com>
18018 David Sherwood <david.sherwood@arm.com>
18020 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18022 (expmed_mult_highpart_optab): Likewise.
18023 (expmed_mult_highpart): 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 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18030 instead of a machine_mode.
18031 (builtin_memset_read_str): Likewise.
18032 * builtins.c (c_readstr): Likewise.
18033 (builtin_memcpy_read_str): Likewise.
18034 (builtin_strncpy_read_str): Likewise.
18035 (builtin_memset_read_str): Likewise.
18036 (builtin_memset_gen_str): Likewise.
18037 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18038 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18039 instead of a machine_mode.
18040 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18042 (make_extraction): Likewise.
18043 (try_widen_shift_mode): Take and return scalar_int_modes instead
18045 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18046 a scalar_int_mode instead of a machine_mode.
18047 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18048 (avr_addr_space_pointer_mode): Likewise.
18049 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18050 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18051 (msp430_unwind_word_mode): Likewise.
18052 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18053 (spu_addr_space_pointer_mode): Likewise.
18054 (spu_addr_space_address_mode): Likewise.
18055 (spu_libgcc_cmp_return_mode): Likewise.
18056 (spu_libgcc_shift_count_mode): Likewise.
18057 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18058 (rl78_addr_space_pointer_mode): Likewise.
18059 (fl78_unwind_word_mode): Likewise.
18060 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18062 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18063 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18064 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18065 (mips_valid_pointer_mode): Likewise.
18066 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18067 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18068 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18070 (ft32_addr_space_address_mode): Likewise.
18071 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18072 scalar_int_mode instead of a machine_mode.
18073 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18075 (m32c_addr_space_address_mode): Likewise.
18076 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18077 (rs6000_eh_return_filter_mode): Likewise.
18078 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18079 (rs6000_eh_return_filter_mode): Likewise.
18080 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18081 (s390_libgcc_shift_count_mode): Likewise.
18082 (s390_unwind_word_mode): Likewise.
18083 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18085 * target.def (mode_rep_extended): Likewise.
18086 (valid_pointer_mode): Likewise.
18087 (addr_space.valid_pointer_mode): Likewise.
18088 (eh_return_filter_mode): Return a scalar_int_mode rather than
18090 (libgcc_cmp_return_mode): Likewise.
18091 (libgcc_shift_count_mode): Likewise.
18092 (unwind_word_mode): Likewise.
18093 (addr_space.pointer_mode): Likewise.
18094 (addr_space.address_mode): Likewise.
18095 * doc/tm.texi: Regenerate.
18096 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18098 (do_jump): Use scalar_int_mode for local variables.
18099 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18100 rather than a machine_mode.
18101 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18102 (scompare_loc_descriptor_wide): Likewise.
18103 (scompare_loc_descriptor_narrow): Likewise.
18104 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18106 * except.c (sjlj_emit_dispatch_table): Likewise.
18107 (expand_builtin_eh_copy_values): Likewise.
18108 * explow.c (convert_memory_address_addr_space_1): Likewise.
18109 Take a scalar_int_mode rather than a machine_mode.
18110 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18111 than a machine_mode.
18112 (memory_address_addr_space): Use scalar_int_mode for local variables.
18113 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18114 rather than a machine_mode.
18115 * expmed.c (mask_rtx): Likewise.
18116 (init_expmed_one_conv): Likewise.
18117 (expand_mult_highpart_adjust): Likewise.
18118 (extract_high_half): Likewise.
18119 (expmed_mult_highpart_optab): Likewise.
18120 (expmed_mult_highpart): Likewise.
18121 (expand_smod_pow2): Likewise.
18122 (expand_sdiv_pow2): Likewise.
18123 (emit_store_flag_int): Likewise.
18124 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18126 (extract_low_bits): Likewise.
18127 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18129 * expr.c (pieces_addr::adjust): Likewise.
18130 (can_store_by_pieces): Likewise.
18131 (store_by_pieces): Likewise.
18132 (clear_by_pieces_1): Likewise.
18133 (expand_expr_addr_expr_1): Likewise.
18134 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18135 (expand_expr_real_1): Likewise.
18136 (try_casesi): Likewise.
18137 * final.c (shorten_branches): Likewise.
18138 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18139 type of "mode" to machine_mode.
18140 * internal-fn.c (expand_arith_overflow_result_store): Take a
18141 scalar_int_mode rather than a machine_mode.
18142 (expand_mul_overflow): Use scalar_int_mode for local variables.
18143 * loop-doloop.c (doloop_modify): Likewise.
18144 (doloop_optimize): Likewise.
18145 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18146 than a machine_mode.
18147 (expand_doubleword_shift_condmove): Likewise.
18148 (expand_doubleword_shift): Likewise.
18149 (expand_doubleword_clz): Likewise.
18150 (expand_doubleword_popcount): Likewise.
18151 (expand_doubleword_parity): Likewise.
18152 (expand_absneg_bit): Use scalar_int_mode for local variables.
18153 (prepare_float_lib_cmp): Likewise.
18154 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18155 rather than a machine_mode.
18156 (convert_memory_address_addr_space): Likewise.
18157 (get_mode_bounds): Likewise.
18158 (get_address_mode): Return a scalar_int_mode rather than a
18160 * rtlanal.c (get_address_mode): Likewise.
18161 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18162 than a machine_mode.
18163 * targhooks.c (default_mode_rep_extended): Likewise.
18164 (default_valid_pointer_mode): Likewise.
18165 (default_addr_space_valid_pointer_mode): Likewise.
18166 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18167 than a machine_mode.
18168 (default_libgcc_cmp_return_mode): Likewise.
18169 (default_libgcc_shift_count_mode): Likewise.
18170 (default_unwind_word_mode): Likewise.
18171 (default_addr_space_pointer_mode): Likewise.
18172 (default_addr_space_address_mode): Likewise.
18173 * targhooks.h (default_eh_return_filter_mode): Likewise.
18174 (default_libgcc_cmp_return_mode): Likewise.
18175 (default_libgcc_shift_count_mode): Likewise.
18176 (default_unwind_word_mode): Likewise.
18177 (default_addr_space_pointer_mode): Likewise.
18178 (default_addr_space_address_mode): Likewise.
18179 (default_mode_rep_extended): Take a scalar_int_mode rather than
18181 (default_valid_pointer_mode): Likewise.
18182 (default_addr_space_valid_pointer_mode): Likewise.
18183 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18185 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18186 rather than a machine_mode.
18187 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18188 for local variables.
18189 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18190 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18191 than a machine_mode.
18193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18194 Alan Hayward <alan.hayward@arm.com>
18195 David Sherwood <david.sherwood@arm.com>
18197 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18198 the mode argument to scalar_int_mode.
18199 (do_jump_by_parts_zero_rtx): Likewise.
18200 (do_jump_by_parts_equality_rtx): Likewise.
18201 (do_jump_by_parts_greater): Take a mode argument.
18202 (do_jump_by_parts_equality): Likewise.
18203 (do_jump_1): Update calls accordingly.
18205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18206 Alan Hayward <alan.hayward@arm.com>
18207 David Sherwood <david.sherwood@arm.com>
18209 * is-a.h (safe_dyn_cast): New function.
18210 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18211 (jump_table_for_label): Likewise.
18212 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18213 instead of an rtx_insn *.
18214 (shorten_branches): Use dyn_cast instead of LABEL_P and
18215 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18216 rtx_jump_table_data::get_data_mode.
18217 (final_scan_insn): Likewise.
18219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18220 Alan Hayward <alan.hayward@arm.com>
18221 David Sherwood <david.sherwood@arm.com>
18223 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18224 trying to combine a full-register integer set with a subreg
18227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18228 Alan Hayward <alan.hayward@arm.com>
18229 David Sherwood <david.sherwood@arm.com>
18231 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18232 that is always either address_mode or pointer_mode.
18234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18235 Alan Hayward <alan.hayward@arm.com>
18236 David Sherwood <david.sherwood@arm.com>
18238 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18239 when the two are known to be equal.
18241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18242 Alan Hayward <alan.hayward@arm.com>
18243 David Sherwood <david.sherwood@arm.com>
18245 * simplify-rtx.c (simplify_const_unary_operation): Use
18246 is_a <scalar_int_mode> instead of checking for a nonzero
18247 precision. Forcibly convert op_mode to a scalar_int_mode
18248 in that case. More clearly differentiate the operand and
18249 result modes and use the former when deciding what the value
18250 of a count-bits operation should be. Use is_int_mode instead
18251 of checking for a MODE_INT. Remove redundant check for whether
18252 this mode has a zero precision.
18254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18255 Alan Hayward <alan.hayward@arm.com>
18256 David Sherwood <david.sherwood@arm.com>
18258 * optabs.c (widen_leading): Change the type of the mode argument
18259 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18260 (widen_bswap): Likewise.
18261 (expand_parity): Likewise.
18262 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18263 (expand_ffs): Likewise.
18264 (epand_unop): Check for scalar integer modes before calling the
18267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18268 Alan Hayward <alan.hayward@arm.com>
18269 David Sherwood <david.sherwood@arm.com>
18271 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18273 (expand_expr_real_1): Update call accordingly.
18275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18276 Alan Hayward <alan.hayward@arm.com>
18277 David Sherwood <david.sherwood@arm.com>
18279 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18280 value_mode arguments. Use scalar_int_mode internally.
18281 (store_bit_field_1): Rename the new integer mode from imode
18282 to op0_mode and use it instead of GET_MODE (op0). Update calls
18283 to store_split_bit_field, store_bit_field_using_insv and
18284 store_fixed_bit_field.
18285 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18286 Use scalar_int_mode internally. Use a bit count rather than a mode
18287 when calculating the largest bit size for get_best_mode.
18288 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18289 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18290 Remove assertion that OP0 has a scalar integer mode.
18291 (store_split_bit_field): Add op0_mode and value_mode arguments.
18292 Update calls to extract_fixed_bit_field.
18293 (extract_bit_field_using_extv): Add an op0_mode argument.
18294 Use scalar_int_mode internally.
18295 (extract_bit_field_1): Rename the new integer mode from imode to
18296 op0_mode and use it instead of GET_MODE (op0). Update calls to
18297 extract_split_bit_field, extract_bit_field_using_extv and
18298 extract_fixed_bit_field.
18299 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18300 to extract_split_bit_field and extract_fixed_bit_field_1.
18301 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18302 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18303 on the target mode.
18304 (extract_split_bit_field): Add an op0_mode argument. Update call
18305 to extract_fixed_bit_field.
18307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18308 Alan Hayward <alan.hayward@arm.com>
18309 David Sherwood <david.sherwood@arm.com>
18311 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18312 * explow.c (hard_function_value): Likewise.
18313 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18314 convert_to_mode call outside the loop.
18315 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18316 for the mode iterator. Require the mode specified by max_pieces
18318 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18320 (copy_blkmode_to_reg): Likewise.
18321 (set_storage_via_setmem): Likewise.
18322 * optabs.c (prepare_cmp_insn): Likewise.
18323 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18324 * stor-layout.c (finish_bitfield_representative): Likewise.
18326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18327 Alan Hayward <alan.hayward@arm.com>
18328 David Sherwood <david.sherwood@arm.com>
18330 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18331 * expr.c (convert_move): Use them.
18332 (convert_modes): Likewise.
18333 (store_expr_with_bounds): 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 (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18340 parameter for the mode of "x". Remove the "known_x", "known_mode"
18341 and "known_ret" arguments. Change the type of the mode argument
18342 to scalar_int_mode.
18343 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18344 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18345 (reg_num_sign_bit_copies_for_combine): Likewise.
18346 * rtlanal.c (nonzero_bits1): Likewise.
18347 (num_sign_bit_copies1): Likewise.
18348 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18349 (reg_num_sign_bit_copies_general): Likewise.
18350 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18351 (reg_nonzero_bits_general): Likewise.
18353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18354 Alan Hayward <alan.hayward@arm.com>
18355 David Sherwood <david.sherwood@arm.com>
18357 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18358 than in subroutines. Return 1 for non-integer modes.
18359 (cached_num_sign_bit_copies): Change the type of the mode parameter
18360 to scalar_int_mode.
18361 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18362 classes. Handle CONST_INT_P first and then check whether X also
18363 has a scalar integer mode. Check the same thing for inner registers
18364 of a SUBREG and for values that are being extended or truncated.
18366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18367 Alan Hayward <alan.hayward@arm.com>
18368 David Sherwood <david.sherwood@arm.com>
18370 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18371 in subroutines. Return the mode mask for non-integer modes.
18372 (cached_nonzero_bits): Change the type of the mode parameter
18373 to scalar_int_mode.
18374 (nonzero_bits1): Likewise. Remove early exit for other mode
18375 classes. Handle CONST_INT_P first and then check whether X
18376 also has a scalar integer mode.
18378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18379 Alan Hayward <alan.hayward@arm.com>
18380 David Sherwood <david.sherwood@arm.com>
18382 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18383 Return a scalar_int_mode and assert that the size is greater than
18385 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18386 (op_by_pieces_d::op_by_pieces_d): Likewise.
18387 (op_by_pieces_d::run): Likewise.
18388 (can_store_by_pieces): Likewise.
18390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18391 Alan Hayward <alan.hayward@arm.com>
18392 David Sherwood <david.sherwood@arm.com>
18394 * combine.c (extract_left_shift): Add a mode argument and update
18396 (make_compound_operation_int): Change the type of the mode parameter
18397 to scalar_int_mode and update the call to extract_left_shift.
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 (simplify_and_const_int): Change the type of the mode
18404 parameter to scalar_int_mode.
18405 (simplify_and_const_int_1): Likewise. Update recursive call.
18407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18408 Alan Hayward <alan.hayward@arm.com>
18409 David Sherwood <david.sherwood@arm.com>
18411 * combine.c (simplify_compare_const): Check that the mode is a
18412 scalar_int_mode (rather than VOIDmode) before testing its
18414 (simplify_comparison): Move COMPARISON_P handling out of the
18415 loop and restrict the latter part of the loop to scalar_int_modes.
18416 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18417 and when considering SUBREG_REGs. Use is_int_mode instead of
18418 checking GET_MODE_CLASS against MODE_INT.
18420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18421 Alan Hayward <alan.hayward@arm.com>
18422 David Sherwood <david.sherwood@arm.com>
18424 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18425 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18426 shift_unit_mode and for modes involved in scalar shifts.
18428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18429 Alan Hayward <alan.hayward@arm.com>
18430 David Sherwood <david.sherwood@arm.com>
18432 * combine.c (force_int_to_mode): New function, split out from...
18433 (force_to_mode): ...here. Keep xmode up-to-date and use it
18434 instead of GET_MODE (x).
18436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18437 Alan Hayward <alan.hayward@arm.com>
18438 David Sherwood <david.sherwood@arm.com>
18440 * optabs-query.h (extraction_insn::struct_mode): Change type to
18441 opt_scalar_int_mode and update comment.
18442 (extraction_insn::field_mode): Change type to scalar_int_mode.
18443 (extraction_insn::pos_mode): Likewise.
18444 * combine.c (make_extraction): Update accordingly.
18445 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18446 (get_optab_extraction_insn): Likewise.
18447 * recog.c (simplify_while_replacing): Likewise.
18448 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18449 parameter to opt_scalar_int_mode.
18451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18452 Alan Hayward <alan.hayward@arm.com>
18453 David Sherwood <david.sherwood@arm.com>
18455 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18456 to a scalar_int_mode instead of a machine_mode.
18457 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18458 (get_best_mode): Return a boolean and use a pointer argument to store
18459 the selected mode. Replace the limit mode parameter with a bit limit.
18460 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18461 for the values returned by bit_field_mode_iterator::next_mode.
18462 (store_bit_field): Update call to get_best_mode.
18463 (store_fixed_bit_field): Likewise.
18464 (extract_fixed_bit_field): Likewise.
18465 * expr.c (optimize_bitfield_assignment_op): Likewise.
18466 * fold-const.c (optimize_bit_field_compare): Likewise.
18467 (fold_truth_andor_1): Likewise.
18468 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18469 Update for new type of m_mode.
18470 (get_best_mode): As above.
18472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18473 Alan Hayward <alan.hayward@arm.com>
18474 David Sherwood <david.sherwood@arm.com>
18476 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18477 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18478 (store_bit_field): Check is_a <scalar_int_mode> before calling
18479 strict_volatile_bitfield_p.
18480 (extract_bit_field): Likewise.
18482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18483 Alan Hayward <alan.hayward@arm.com>
18484 David Sherwood <david.sherwood@arm.com>
18486 * target.def (cstore_mode): Return a scalar_int_mode.
18487 * doc/tm.texi: Regenerate.
18488 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18489 * targhooks.h (default_cstore_mode): Likewise.
18490 * targhooks.c (default_cstore_mode): Likewise, using a forced
18492 * expmed.c (emit_cstore): Expect the target of the cstore to be
18495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18496 Alan Hayward <alan.hayward@arm.com>
18497 David Sherwood <david.sherwood@arm.com>
18499 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18501 (niter_desc): Likewise mode.
18502 (iv_analyze): Add a mode parameter.
18504 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18505 and change its type to scalar_int_mode.
18506 * loop-iv.c: Update commentary at head of file.
18507 (iv_constant): Pass the mode paraeter before the rtx it describes
18508 and change its type to scalar_int_mode. Remove VOIDmode handling.
18509 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18510 (iv_extend): Likewise.
18511 (shorten_into_mode): Likewise.
18512 (iv_add): Use scalar_int_mode.
18513 (iv_mult): Likewise.
18514 (iv_shift): Likewise.
18515 (canonicalize_iv_subregs): Likewise.
18516 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18517 it describes and change its mode to scalar_int_mode. Also change
18518 the type of the returned inner_mode to scalar_int_mode.
18519 (get_biv_step): Likewise, turning outer_mode from a pointer
18520 into a direct parameter. Update call to get_biv_step_1.
18521 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18522 iv_constant and get_biv_step.
18523 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18524 and change its type to scalar_int_mode. Don't initialise iv->mode
18525 to VOIDmode and remove later checks for its still being VOIDmode.
18526 Update calls to iv_analyze_op and iv_analyze_expr. Check
18527 is_a <scalar_int_mode> when changing the mode under consideration.
18528 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18529 Update call to iv_analyze_expr.
18530 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18531 inner register is not also a scalar_int_mode. Update call to
18533 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18534 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18535 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18536 separate mode class checks. Update calls to iv_analyze. Remove
18537 fix-up of VOIDmodes after iv_analyze_biv.
18538 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18539 don't have a scalar_int_mode. Update call to biv_p.
18541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18542 Alan Hayward <alan.hayward@arm.com>
18543 David Sherwood <david.sherwood@arm.com>
18545 * cfgexpand.c (convert_debug_memory_address): Use
18546 as_a <scalar_int_mode>.
18547 * combine.c (expand_compound_operation): Likewise.
18548 (make_extraction): Likewise.
18549 (change_zero_ext): Likewise.
18550 (simplify_comparison): Likewise.
18551 * cse.c (cse_insn): Likewise.
18552 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18553 (mem_loc_descriptor): Likewise.
18554 (loc_descriptor): Likewise.
18555 * expmed.c (init_expmed_one_mode): Likewise.
18556 (synth_mult): Likewise.
18557 (emit_store_flag_1): Likewise.
18558 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18559 of a comparison with size.
18560 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18561 (reduce_to_bit_field_precision): Likewise.
18562 * function.c (expand_function_end): Likewise.
18563 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18564 * loop-doloop.c (doloop_modify): Likewise.
18565 * optabs.c (expand_binop): Likewise.
18566 (expand_unop): Likewise.
18567 (expand_copysign_absneg): Likewise.
18568 (prepare_cmp_insn): Likewise.
18569 (maybe_legitimize_operand): Likewise.
18570 * recog.c (const_scalar_int_operand): Likewise.
18571 * rtlanal.c (get_address_mode): Likewise.
18572 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18573 (simplify_cond_clz_ctz): Likewise.
18574 * tree-nested.c (get_nl_goto_field): Likewise.
18575 * tree.c (build_vector_type_for_mode): Likewise.
18576 * var-tracking.c (use_narrower_mode): Likewise.
18578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18579 Alan Hayward <alan.hayward@arm.com>
18580 David Sherwood <david.sherwood@arm.com>
18582 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18583 * builtins.c (expand_builtin_signbit): Use it.
18584 * cfgexpand.c (expand_debug_expr): Likewise.
18585 * dojump.c (do_jump): Likewise.
18586 (do_compare_and_jump): Likewise.
18587 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18588 * expmed.c (make_tree): Likewise.
18589 * expr.c (expand_expr_real_2): Likewise.
18590 (expand_expr_real_1): Likewise.
18591 (try_casesi): Likewise.
18592 * fold-const-call.c (fold_const_call_ss): Likewise.
18593 * fold-const.c (unextend): Likewise.
18594 (extract_muldiv_1): Likewise.
18595 (fold_single_bit_test): Likewise.
18596 (native_encode_int): Likewise.
18597 (native_encode_string): Likewise.
18598 (native_interpret_int): Likewise.
18599 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18600 * internal-fn.c (expand_addsub_overflow): Likewise.
18601 (expand_neg_overflow): Likewise.
18602 (expand_mul_overflow): Likewise.
18603 (expand_arith_overflow): Likewise.
18604 * match.pd: Likewise.
18605 * stor-layout.c (layout_type): Likewise.
18606 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18607 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18608 * tree-ssanames.c (get_range_info): Likewise.
18609 * tree-switch-conversion.c (array_value_type) Likewise.
18610 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18611 (vect_recog_divmod_pattern): Likewise.
18612 (vect_recog_mixed_size_cond_pattern): Likewise.
18613 * tree-vrp.c (extract_range_basic): Likewise.
18614 (simplify_float_conversion_using_ranges): Likewise.
18615 * tree.c (int_fits_type_p): Likewise.
18616 * ubsan.c (instrument_bool_enum_load): Likewise.
18617 * varasm.c (mergeable_string_section): Likewise.
18618 (narrowing_initializer_constant_valid_p): Likewise.
18619 (output_constant): Likewise.
18621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18622 Alan Hayward <alan.hayward@arm.com>
18623 David Sherwood <david.sherwood@arm.com>
18625 * machmode.h (NARROWEST_INT_MODE): New macro.
18626 * expr.c (alignment_for_piecewise_move): Use it instead of
18627 GET_CLASS_NARROWEST_MODE (MODE_INT).
18628 (push_block): Likewise.
18629 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18631 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18634 Alan Hayward <alan.hayward@arm.com>
18635 David Sherwood <david.sherwood@arm.com>
18637 * postreload.c (move2add_valid_value_p): Change the type of the
18638 mode parameter to scalar_int_mode.
18639 (move2add_use_add2_insn): Add a mode parameter and use it instead
18641 (move2add_use_add3_insn): Likewise.
18642 (reload_cse_move2add): Update accordingly.
18644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18645 Alan Hayward <alan.hayward@arm.com>
18646 David Sherwood <david.sherwood@arm.com>
18648 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18650 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18651 * optabs.c (expand_unop): Likewise.
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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18658 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18659 (popcount_loc_descriptor): Likewise.
18660 (bswap_loc_descriptor): Likewise.
18661 (rotate_loc_descriptor): Likewise.
18662 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18663 calling the functions above.
18665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18666 Alan Hayward <alan.hayward@arm.com>
18667 David Sherwood <david.sherwood@arm.com>
18669 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18671 (try_combine): Likewise.
18672 (simplify_if_then_else): Likewise.
18673 * cse.c (cse_insn): Likewise.
18674 * dwarf2out.c (mem_loc_descriptor): Likewise.
18675 * emit-rtl.c (gen_lowpart_common): Likewise.
18676 * simplify-rtx.c (simplify_truncation): Likewise.
18677 (simplify_binary_operation_1): Likewise.
18678 (simplify_const_relational_operation): Likewise.
18679 (simplify_ternary_operation): Likewise.
18680 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18683 Alan Hayward <alan.hayward@arm.com>
18684 David Sherwood <david.sherwood@arm.com>
18686 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18687 * reload.c (push_reload): Likewise.
18688 (find_reloads): Likewise.
18690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18691 Alan Hayward <alan.hayward@arm.com>
18692 David Sherwood <david.sherwood@arm.com>
18694 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18695 (make_compound_operation_int): Likewise.
18696 (change_zero_ext): Likewise.
18697 * expr.c (convert_move): Likewise.
18698 (convert_modes): Likewise.
18699 * fwprop.c (forward_propagate_subreg): Likewise.
18700 * loop-iv.c (get_biv_step_1): Likewise.
18701 * optabs.c (widen_operand): Likewise.
18702 * postreload.c (move2add_valid_value_p): Likewise.
18703 * recog.c (simplify_while_replacing): Likewise.
18704 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18705 (simplify_binary_operation_1): Likewise. Remove redundant
18706 mode equality check.
18708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18709 Alan Hayward <alan.hayward@arm.com>
18710 David Sherwood <david.sherwood@arm.com>
18712 * combine.c (combine_simplify_rtx): Add checks for
18713 is_a <scalar_int_mode>.
18714 (simplify_if_then_else): Likewise.
18715 (make_field_assignment): Likewise.
18716 (simplify_comparison): Likewise.
18717 * ifcvt.c (noce_try_bitop): Likewise.
18718 * loop-invariant.c (canonicalize_address_mult): Likewise.
18719 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18722 Alan Hayward <alan.hayward@arm.com>
18723 David Sherwood <david.sherwood@arm.com>
18725 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18726 is_a <scalar_int_mode> instead of != BLKmode.
18728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18729 Alan Hayward <alan.hayward@arm.com>
18730 David Sherwood <david.sherwood@arm.com>
18732 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18733 instead of != VOIDmode.
18734 * combine.c (if_then_else_cond): Likewise.
18735 (change_zero_ext): Likewise.
18736 * dwarf2out.c (mem_loc_descriptor): Likewise.
18737 (loc_descriptor): Likewise.
18738 * rtlanal.c (canonicalize_condition): Likewise.
18739 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18742 Alan Hayward <alan.hayward@arm.com>
18743 David Sherwood <david.sherwood@arm.com>
18745 * simplify-rtx.c (simplify_binary_operation_1): Use
18746 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18749 Alan Hayward <alan.hayward@arm.com>
18750 David Sherwood <david.sherwood@arm.com>
18752 * wide-int.h (int_traits<unsigned char>) New class.
18753 (int_traits<unsigned short>) Likewise.
18754 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18755 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18757 * combine.c (set_nonzero_bits_and_sign_copies): Use
18758 is_a <scalar_int_mode>.
18759 (find_split_point): Likewise.
18760 (combine_simplify_rtx): Likewise.
18761 (simplify_logical): Likewise.
18762 (expand_compound_operation): Likewise.
18763 (expand_field_assignment): Likewise.
18764 (make_compound_operation): Likewise.
18765 (extended_count): Likewise.
18766 (change_zero_ext): Likewise.
18767 (simplify_comparison): Likewise.
18768 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18769 (ucompare_loc_descriptor): Likewise.
18770 (minmax_loc_descriptor): Likewise.
18771 (mem_loc_descriptor): Likewise.
18772 (loc_descriptor): Likewise.
18773 * expmed.c (init_expmed_one_mode): Likewise.
18774 * lra-constraints.c (lra_constraint_offset): Likewise.
18775 * optabs.c (prepare_libcall_arg): Likewise.
18776 * postreload.c (move2add_note_store): Likewise.
18777 * reload.c (operands_match_p): Likewise.
18778 * rtl.h (load_extend_op): Likewise.
18779 * rtlhooks.c (gen_lowpart_general): Likewise.
18780 * simplify-rtx.c (simplify_truncation): Likewise.
18781 (simplify_unary_operation_1): Likewise.
18782 (simplify_binary_operation_1): Likewise.
18783 (simplify_const_binary_operation): Likewise.
18784 (simplify_const_relational_operation): Likewise.
18785 (simplify_subreg): Likewise.
18786 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18787 * var-tracking.c (adjust_mems): Likewise.
18788 (prepare_call_arguments): Likewise.
18790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18791 Alan Hayward <alan.hayward@arm.com>
18792 David Sherwood <david.sherwood@arm.com>
18794 * machmode.h (is_int_mode): New fuction.
18795 * combine.c (find_split_point): Use it.
18796 (combine_simplify_rtx): Likewise.
18797 (simplify_if_then_else): Likewise.
18798 (simplify_set): Likewise.
18799 (simplify_shift_const_1): Likewise.
18800 (simplify_comparison): Likewise.
18801 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18802 * cse.c (notreg_cost): Likewise.
18803 (cse_insn): Likewise.
18804 * cselib.c (cselib_lookup_1): Likewise.
18805 * dojump.c (do_jump_1): Likewise.
18806 (do_compare_rtx_and_jump): Likewise.
18807 * dse.c (get_call_args): Likewise.
18808 * dwarf2out.c (rtl_for_decl_init): Likewise.
18809 (native_encode_initializer): Likewise.
18810 * expmed.c (emit_store_flag_1): Likewise.
18811 (emit_store_flag): Likewise.
18812 * expr.c (convert_modes): Likewise.
18813 (store_field): Likewise.
18814 (expand_expr_real_1): Likewise.
18815 * fold-const.c (fold_read_from_constant_string): Likewise.
18816 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18817 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18818 * optabs.c (expand_binop): Likewise.
18819 (expand_unop): Likewise.
18820 (expand_abs_nojump): Likewise.
18821 (expand_one_cmpl_abs_nojump): Likewise.
18822 * simplify-rtx.c (mode_signbit_p): Likewise.
18823 (val_signbit_p): Likewise.
18824 (val_signbit_known_set_p): Likewise.
18825 (val_signbit_known_clear_p): Likewise.
18826 (simplify_relational_operation_1): Likewise.
18827 * tree.c (vector_type_mode): Likewise.
18829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18830 Alan Hayward <alan.hayward@arm.com>
18831 David Sherwood <david.sherwood@arm.com>
18833 * machmode.h (smallest_mode_for_size): Fix formatting.
18834 (smallest_int_mode_for_size): New function.
18835 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18836 instead of smallest_mode_for_size.
18837 * combine.c (make_extraction): Likewise.
18838 * config/arc/arc.c (arc_expand_movmem): Likewise.
18839 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18840 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18841 * config/s390/s390.c (s390_expand_insv): Likewise.
18842 * config/sparc/sparc.c (assign_int_registers): Likewise.
18843 * config/spu/spu.c (spu_function_value): Likewise.
18844 (spu_function_arg): Likewise.
18845 * coverage.c (get_gcov_type): Likewise.
18846 (get_gcov_unsigned_t): Likewise.
18847 * dse.c (find_shift_sequence): Likewise.
18848 * expmed.c (store_bit_field_1): Likewise.
18849 * expr.c (convert_move): Likewise.
18850 (store_field): Likewise.
18851 * internal-fn.c (expand_arith_overflow): Likewise.
18852 * optabs-query.c (get_best_extraction_insn): Likewise.
18853 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18854 * stor-layout.c (layout_type): Likewise.
18855 (initialize_sizetypes): Likewise.
18856 * targhooks.c (default_get_mask_mode): Likewise.
18857 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18860 Alan Hayward <alan.hayward@arm.com>
18861 David Sherwood <david.sherwood@arm.com>
18863 * machmode.h (opt_mode::else_blk): New function.
18864 (int_mode_for_mode): Declare.
18865 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18866 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18868 * cfgexpand.c (expand_debug_expr): Likewise.
18869 * combine.c (gen_lowpart_or_truncate): Likewise.
18870 (gen_lowpart_for_combine): Likewise.
18871 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18872 * config/avr/avr.c (avr_to_int_mode): Likewise.
18873 (avr_out_plus_1): Likewise.
18874 (avr_out_plus): Likewise.
18875 (avr_out_round): Likewise.
18876 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18877 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18878 (s390_expand_vcond): Likewise.
18879 * config/spu/spu.c (spu_split_immediate): Likewise.
18880 (spu_expand_mov): Likewise.
18881 * dse.c (get_stored_val): Likewise.
18882 * expmed.c (store_bit_field_1): Likewise.
18883 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18885 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18886 (extract_low_bits): Likewise.
18887 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18888 handling rather than repeating the check.
18889 (emit_group_store): Likewise.
18890 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18891 * optabs.c (expand_absneg_bit): Likewise.
18892 (expand_copysign_absneg): Likewise.
18893 (expand_copysign_bit): Likewise.
18894 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18895 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18896 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18897 * var-tracking.c (prepare_call_arguments): Likewise.
18898 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18899 int_mode_for_mode instead of mode_for_size.
18900 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18903 Alan Hayward <alan.hayward@arm.com>
18904 David Sherwood <david.sherwood@arm.com>
18906 * machmode.h (int_mode_for_size): New function.
18907 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18908 instead of mode_for_size.
18909 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18911 * combine.c (expand_field_assignment): Use int_mode_for_size
18912 instead of mode_for_size.
18913 (make_extraction): Likewise.
18914 (simplify_shift_const_1): Likewise.
18915 (simplify_comparison): Likewise.
18916 * dojump.c (do_jump): Likewise.
18917 * dwarf2out.c (mem_loc_descriptor): Likewise.
18918 * emit-rtl.c (init_derived_machine_modes): Likewise.
18919 * expmed.c (flip_storage_order): Likewise.
18920 (convert_extracted_bit_field): Likewise.
18921 * expr.c (copy_blkmode_from_reg): Likewise.
18922 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18923 * internal-fn.c (expand_mul_overflow): Likewise.
18924 * lower-subreg.c (simple_move): Likewise.
18925 * optabs-libfuncs.c (init_optabs): Likewise.
18926 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18927 * tree.c (vector_type_mode): Likewise.
18928 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18929 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18930 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18931 * tree-vect-stmts.c (vectorizable_load): Likewise.
18932 (vectorizable_store): Likewise.
18934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18935 Alan Hayward <alan.hayward@arm.com>
18936 David Sherwood <david.sherwood@arm.com>
18938 * coretypes.h (pod_mode): New type.
18939 (scalar_int_mode_pod): New typedef.
18940 * machmode.h (pod_mode): New class.
18941 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18942 * genmodes.c (emit_mode_int_n): Update accordingly.
18943 * lower-subreg.h (target_lower_subreg): Change type to
18944 scalar_int_mode_pod.
18945 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18946 scalar_int_mode_pod.
18948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18949 Alan Hayward <alan.hayward@arm.com>
18950 David Sherwood <david.sherwood@arm.com>
18952 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18953 machine_mode to scalar_int_mode.
18954 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18955 (rs6000_option_override_internal): Remove cast to int.
18956 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18957 machine_mode to scalar_int_mode.
18958 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18959 (rs6000_option_override_internal): Remove cast to int.
18960 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18961 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18963 * config/s390/s390.c (s390_expand_builtin): Likewise.
18964 * coretypes.h (scalar_int_mode): New type.
18965 (opt_scalar_int_mode): New typedef.
18966 * machmode.h (scalar_int_mode): New class.
18967 (scalar_int_mode::includes_p): New function.
18968 (byte_mode): Change type to scalar_int_mode.
18969 (word_mode): Likewise.
18970 (ptr_mode): Likewise.
18971 * emit-rtl.c (byte_mode): Likewise.
18972 (word_mode): Likewise.
18973 (ptr_mode): Likewise.
18974 (init_derived_machine_modes): Update accordingly.
18975 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18976 and MODE_PARTIAL_INT.
18977 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18978 opt_scalar_int_mode.
18980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18981 Alan Hayward <alan.hayward@arm.com>
18982 David Sherwood <david.sherwood@arm.com>
18984 * target.def (libgcc_floating_mode_supported_p): Take a
18986 * doc/tm.texi: Regenerate.
18987 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18989 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18990 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18994 Alan Hayward <alan.hayward@arm.com>
18995 David Sherwood <david.sherwood@arm.com>
18997 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18998 * doc/tm.texi: Regenerate.
18999 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19000 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19001 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19002 * targhooks.h (default_floatn_mode): Likewise.
19003 * targhooks.c (default_floatn_mode): Likewise.
19004 * tree.c (build_common_tree_nodes): Update accordingly.
19006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19007 Alan Hayward <alan.hayward@arm.com>
19008 David Sherwood <david.sherwood@arm.com>
19010 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19011 (mode_iterator::iterate_p): Likewise.
19012 (mode_iterator::get_wider): Likewise.
19013 * expr.c (init_expr_target): Use opt_scalar_float_mode.
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 * coretypes.h (opt_scalar_float_mode): New typedef.
19020 * machmode.h (float_mode_for_size): New function.
19021 * emit-rtl.c (double_mode): Delete.
19022 (init_emit_once): Use float_mode_for_size.
19023 * stor-layout.c (layout_type): Likewise.
19024 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19027 Alan Hayward <alan.hayward@arm.com>
19028 David Sherwood <david.sherwood@arm.com>
19030 * output.h (assemble_real): Take a scalar_float_mode.
19031 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19032 * config/arm/arm.md (consttable_4): Likewise.
19033 (consttable_8): Likewise.
19034 (consttable_16): Likewise.
19035 * config/mips/mips.md (consttable_float): Likewise.
19036 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19037 * varasm.c (assemble_real): Take a scalar_float_mode.
19038 (output_constant_pool_2): Update accordingly.
19039 (output_constant): Likewise.
19041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19042 Alan Hayward <alan.hayward@arm.com>
19043 David Sherwood <david.sherwood@arm.com>
19045 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19046 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19047 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19048 (native_encode_real): Likewise.
19049 (native_interpret_real): Likewise.
19050 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19051 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19054 Alan Hayward <alan.hayward@arm.com>
19055 David Sherwood <david.sherwood@arm.com>
19057 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19058 <scalar_float_mode>. Simplify.
19059 (gen_extend_conv_libfunc): Likewise.
19061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19062 Alan Hayward <alan.hayward@arm.com>
19063 David Sherwood <david.sherwood@arm.com>
19065 * coretypes.h (scalar_float_mode): New type.
19066 * machmode.h (mode_traits::from_int): Use machine_mode if
19067 USE_ENUM_MODES is defined.
19068 (is_a): New function.
19070 (dyn_cast): Likewise.
19071 (scalar_float_mode): New class.
19072 (scalar_float_mode::includes_p): New function.
19073 (is_float_mode): Likewise.
19074 * gdbhooks.py (MachineModePrinter): New class.
19075 (build_pretty_printer): Use it for scalar_float_mode.
19076 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19077 (format_helper::format_helper): Turn into a template.
19078 * genmodes.c (get_mode_class): New function.
19079 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19080 or machine_mode if none.
19081 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19082 as_a <scalar_float_mode>.
19083 * dwarf2out.c (mem_loc_descriptor): Likewise.
19084 (insert_float): Likewise.
19085 (add_const_value_attribute): Likewise.
19086 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19087 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19088 (expand_unop): Update accordingly.
19089 (expand_abs_nojump): Likewise.
19090 (expand_copysign_absneg): Take a scalar_float_mode.
19091 (expand_copysign_bit): Likewise.
19092 (expand_copysign): Update accordingly.
19094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19095 Alan Hayward <alan.hayward@arm.com>
19096 David Sherwood <david.sherwood@arm.com>
19098 * coretypes.h (opt_mode): New class.
19099 * machmode.h (opt_mode): Likewise.
19100 (opt_mode::else_void): New function.
19101 (opt_mode::require): Likewise.
19102 (opt_mode::exists): Likewise.
19103 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19104 (GET_MODE_2XWIDER_MODE): Likewise.
19105 (mode_iterator::get_wider): Update accordingly.
19106 (mode_iterator::get_2xwider): Likewise.
19107 (mode_iterator::get_known_wider): Likewise, turning into a template.
19108 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19109 forcing a wider mode to exist.
19110 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19111 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19112 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19113 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19114 * lower-subreg.c (init_lower_subreg): Likewise.
19115 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19116 on the final iteration.
19117 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19118 a wider mode exists before asking for a move pattern.
19119 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19120 forcing a wider mode to exist.
19121 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19122 returning false if no such mode exists.
19123 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19124 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19125 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19126 Avoid checking for a MODE_INT if we already know the mode is not a
19128 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19129 forcing a wider mode to exist.
19130 (expmed_mult_highpart_optab): Likewise.
19131 (expmed_mult_highpart): Likewise.
19132 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19134 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19135 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19136 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19137 * internal-fn.c (expand_mul_overflow): Update use of
19138 GET_MODE_2XWIDER_MODE.
19139 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19140 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19141 GET_MODE_WIDER_MODE.
19142 (convert_plusminus_to_widen): Likewise.
19143 * tree-switch-conversion.c (array_value_type): Likewise.
19144 * var-tracking.c (emit_note_insn_var_location): Likewise.
19145 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19146 Return false inside rather than outside the loop if no wider mode
19148 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19149 and GET_MODE_2XWIDER_MODE
19150 (can_compare_p): Use else_void.
19151 * gdbhooks.py (OptMachineModePrinter): New class.
19152 (build_pretty_printer): Use it for opt_mode.
19154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19155 Alan Hayward <alan.hayward@arm.com>
19156 David Sherwood <david.sherwood@arm.com>
19158 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19159 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19162 Alan Hayward <alan.hayward@arm.com>
19163 David Sherwood <david.sherwood@arm.com>
19165 * machmode.h (mode_traits): New structure.
19166 (get_narrowest_mode): New function.
19167 (mode_iterator::start): Likewise.
19168 (mode_iterator::iterate_p): Likewise.
19169 (mode_iterator::get_wider): Likewise.
19170 (mode_iterator::get_known_wider): Likewise.
19171 (mode_iterator::get_2xwider): Likewise.
19172 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19173 (FOR_EACH_MODE): Likewise.
19174 (FOR_EACH_MODE_FROM): Likewise.
19175 (FOR_EACH_MODE_UNTIL): Likewise.
19176 (FOR_EACH_WIDER_MODE): Likewise.
19177 (FOR_EACH_2XWIDER_MODE): Likewise.
19178 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19179 * combine.c (simplify_comparison): Likewise
19180 * config/i386/i386.c (type_natural_mode): Likewise.
19181 * cse.c (cse_insn): Likewise.
19182 * dse.c (find_shift_sequence): Likewise.
19183 * emit-rtl.c (init_derived_machine_modes): Likewise.
19184 (init_emit_once): Likewise.
19185 * explow.c (hard_function_value): Likewise.
19186 * expmed.c (extract_fixed_bit_field_1): Likewise.
19187 (extract_bit_field_1): Likewise.
19188 (expand_divmod): Likewise.
19189 (emit_store_flag_1): Likewise.
19190 * expr.c (init_expr_target): Likewise.
19191 (convert_move): Likewise.
19192 (alignment_for_piecewise_move): Likewise.
19193 (widest_int_mode_for_size): Likewise.
19194 (emit_block_move_via_movmem): Likewise.
19195 (copy_blkmode_to_reg): Likewise.
19196 (set_storage_via_setmem): Likewise.
19197 (compress_float_constant): Likewise.
19198 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19199 * optabs-query.c (get_best_extraction_insn): Likewise.
19200 * optabs.c (expand_binop): Likewise.
19201 (expand_twoval_unop): Likewise.
19202 (expand_twoval_binop): Likewise.
19203 (widen_leading): Likewise.
19204 (widen_bswap): Likewise.
19205 (expand_parity): Likewise.
19206 (expand_unop): Likewise.
19207 (prepare_cmp_insn): Likewise.
19208 (prepare_float_lib_cmp): Likewise.
19209 (expand_float): Likewise.
19210 (expand_fix): Likewise.
19211 (expand_sfix_optab): Likewise.
19212 * postreload.c (move2add_use_add2_insn): Likewise.
19213 * reg-stack.c (reg_to_stack): Likewise.
19214 * reginfo.c (choose_hard_reg_mode): Likewise.
19215 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19216 * stor-layout.c (mode_for_size): Likewise.
19217 (smallest_mode_for_size): Likewise.
19218 (mode_for_vector): Likewise.
19219 (finish_bitfield_representative): Likewise.
19220 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19221 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19222 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19223 * var-tracking.c (prepare_call_arguments): Likewise.
19225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19226 Alan Hayward <alan.hayward@arm.com>
19227 David Sherwood <david.sherwood@arm.com>
19229 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19230 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19231 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19232 * machmode.h (mode_size): Move earlier in file.
19233 (mode_precision): Likewise.
19234 (mode_inner): Likewise.
19235 (mode_nunits): Likewise.
19236 (mode_unit_size): Likewise.
19237 (unit_unit_precision): Likewise.
19238 (mode_wider): Likewise.
19239 (mode_2xwider): Likewise.
19240 (machine_mode): New class.
19241 (mode_to_bytes): New function.
19242 (mode_to_bits): Likewise.
19243 (mode_to_precision): Likewise.
19244 (mode_to_inner): Likewise.
19245 (mode_to_unit_size): Likewise.
19246 (mode_to_unit_precision): Likewise.
19247 (mode_to_nunits): Likewise.
19248 (GET_MODE_SIZE): Use mode_to_bytes.
19249 (GET_MODE_BITSIZE): Use mode_to_bits.
19250 (GET_MODE_PRECISION): Use mode_to_precision.
19251 (GET_MODE_INNER): Use mode_to_inner.
19252 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19253 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19254 (GET_MODE_NUNITS): Use mode_to_nunits.
19255 * system.h (ALWAYS_INLINE): New macro.
19256 * config/powerpcspe/powerpcspe-c.c
19257 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19258 int for arg1_mode and arg2_mode.
19260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19261 Alan Hayward <alan.hayward@arm.com>
19262 David Sherwood <david.sherwood@arm.com>
19264 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19265 Prefix mode names with E_ in case statements.
19266 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19267 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19268 (aarch64_split_simd_move): Likewise.
19269 (aarch64_gen_storewb_pair): Likewise.
19270 (aarch64_gen_loadwb_pair): Likewise.
19271 (aarch64_gen_store_pair): Likewise.
19272 (aarch64_gen_load_pair): Likewise.
19273 (aarch64_get_condition_code_1): Likewise.
19274 (aarch64_constant_pool_reload_icode): Likewise.
19275 (get_rsqrte_type): Likewise.
19276 (get_rsqrts_type): Likewise.
19277 (get_recpe_type): Likewise.
19278 (get_recps_type): Likewise.
19279 (aarch64_gimplify_va_arg_expr): Likewise.
19280 (aarch64_simd_container_mode): Likewise.
19281 (aarch64_emit_load_exclusive): Likewise.
19282 (aarch64_emit_store_exclusive): Likewise.
19283 (aarch64_expand_compare_and_swap): Likewise.
19284 (aarch64_gen_atomic_cas): Likewise.
19285 (aarch64_emit_bic): Likewise.
19286 (aarch64_emit_atomic_swap): Likewise.
19287 (aarch64_emit_atomic_load_op): Likewise.
19288 (aarch64_evpc_trn): Likewise.
19289 (aarch64_evpc_uzp): Likewise.
19290 (aarch64_evpc_zip): Likewise.
19291 (aarch64_evpc_ext): Likewise.
19292 (aarch64_evpc_rev): Likewise.
19293 (aarch64_evpc_dup): Likewise.
19294 (aarch64_gen_ccmp_first): Likewise.
19295 (aarch64_gen_ccmp_next): Likewise.
19296 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19297 (alpha_emit_xfloating_libcall): Likewise.
19298 (emit_insxl): Likewise.
19299 (alpha_arg_type): Likewise.
19300 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19301 (arc_preferred_simd_mode): Likewise.
19302 (arc_secondary_reload): Likewise.
19303 (get_arc_condition_code): Likewise.
19304 (arc_print_operand): Likewise.
19305 (arc_legitimate_constant_p): Likewise.
19306 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19307 * config/arc/arc.md (casesi_load): Likewise.
19308 (casesi_compact_jump): Likewise.
19309 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19310 (cc_use_register): Likewise.
19311 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19312 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19313 (arm_init_iwmmxt_builtins): Likewise.
19314 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19315 (neon_expand_vector_init): Likewise.
19316 (arm_attr_length_move_neon): Likewise.
19317 (maybe_get_arm_condition_code): Likewise.
19318 (arm_emit_vector_const): Likewise.
19319 (arm_preferred_simd_mode): Likewise.
19320 (arm_output_iwmmxt_tinsr): Likewise.
19321 (thumb1_output_casesi): Likewise.
19322 (thumb2_output_casesi): Likewise.
19323 (arm_emit_load_exclusive): Likewise.
19324 (arm_emit_store_exclusive): Likewise.
19325 (arm_expand_compare_and_swap): Likewise.
19326 (arm_evpc_neon_vuzp): Likewise.
19327 (arm_evpc_neon_vzip): Likewise.
19328 (arm_evpc_neon_vrev): Likewise.
19329 (arm_evpc_neon_vtrn): Likewise.
19330 (arm_evpc_neon_vext): Likewise.
19331 (arm_validize_comparison): Likewise.
19332 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19333 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19334 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19335 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19336 (c6x_preferred_simd_mode): Likewise.
19337 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19338 (epiphany_rtx_costs): Likewise.
19339 * config/epiphany/predicates.md (proper_comparison_operator):
19341 * config/frv/frv.c (condexec_memory_operand): Likewise.
19342 (frv_emit_move): Likewise.
19343 (output_move_single): Likewise.
19344 (output_condmove_single): Likewise.
19345 (frv_hard_regno_mode_ok): Likewise.
19346 (frv_matching_accg_mode): Likewise.
19347 * config/h8300/h8300.c (split_adds_subs): Likewise.
19348 (h8300_rtx_costs): Likewise.
19349 (h8300_print_operand): Likewise.
19350 (compute_mov_length): Likewise.
19351 (output_logical_op): Likewise.
19352 (compute_logical_op_length): Likewise.
19353 (compute_logical_op_cc): Likewise.
19354 (h8300_shift_needs_scratch_p): Likewise.
19355 (output_a_shift): Likewise.
19356 (compute_a_shift_length): Likewise.
19357 (compute_a_shift_cc): Likewise.
19358 (expand_a_rotate): Likewise.
19359 (output_a_rotate): Likewise.
19360 * config/i386/i386.c (classify_argument): Likewise.
19361 (function_arg_advance_32): Likewise.
19362 (function_arg_32): Likewise.
19363 (function_arg_64): Likewise.
19364 (function_value_64): Likewise.
19365 (ix86_gimplify_va_arg): Likewise.
19366 (ix86_legitimate_constant_p): Likewise.
19367 (put_condition_code): Likewise.
19368 (split_double_mode): Likewise.
19369 (ix86_avx256_split_vector_move_misalign): Likewise.
19370 (ix86_expand_vector_logical_operator): Likewise.
19371 (ix86_split_idivmod): Likewise.
19372 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19373 (ix86_build_const_vector): Likewise.
19374 (ix86_build_signbit_mask): Likewise.
19375 (ix86_match_ccmode): Likewise.
19376 (ix86_cc_modes_compatible): Likewise.
19377 (ix86_expand_branch): Likewise.
19378 (ix86_expand_sse_cmp): Likewise.
19379 (ix86_expand_sse_movcc): Likewise.
19380 (ix86_expand_int_sse_cmp): Likewise.
19381 (ix86_expand_vec_perm_vpermi2): Likewise.
19382 (ix86_expand_vec_perm): Likewise.
19383 (ix86_expand_sse_unpack): Likewise.
19384 (ix86_expand_int_addcc): Likewise.
19385 (ix86_split_to_parts): Likewise.
19386 (ix86_vectorize_builtin_gather): Likewise.
19387 (ix86_vectorize_builtin_scatter): Likewise.
19388 (avx_vpermilp_parallel): Likewise.
19389 (inline_memory_move_cost): Likewise.
19390 (ix86_tieable_integer_mode_p): Likewise.
19391 (x86_maybe_negate_const_int): Likewise.
19392 (ix86_expand_vector_init_duplicate): Likewise.
19393 (ix86_expand_vector_init_one_nonzero): Likewise.
19394 (ix86_expand_vector_init_one_var): Likewise.
19395 (ix86_expand_vector_init_concat): Likewise.
19396 (ix86_expand_vector_init_interleave): Likewise.
19397 (ix86_expand_vector_init_general): Likewise.
19398 (ix86_expand_vector_set): Likewise.
19399 (ix86_expand_vector_extract): Likewise.
19400 (emit_reduc_half): Likewise.
19401 (ix86_emit_i387_round): Likewise.
19402 (ix86_mangle_type): Likewise.
19403 (ix86_expand_round_sse4): Likewise.
19404 (expand_vec_perm_blend): Likewise.
19405 (canonicalize_vector_int_perm): Likewise.
19406 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19407 (expand_vec_perm_1): Likewise.
19408 (expand_vec_perm_interleave3): Likewise.
19409 (expand_vec_perm_even_odd_pack): Likewise.
19410 (expand_vec_perm_even_odd_1): Likewise.
19411 (expand_vec_perm_broadcast_1): Likewise.
19412 (ix86_vectorize_vec_perm_const_ok): Likewise.
19413 (ix86_expand_vecop_qihi): Likewise.
19414 (ix86_expand_mul_widen_hilo): Likewise.
19415 (ix86_expand_sse2_abs): Likewise.
19416 (ix86_expand_pextr): Likewise.
19417 (ix86_expand_pinsr): Likewise.
19418 (ix86_preferred_simd_mode): Likewise.
19419 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19420 * config/i386/sse.md (*andnot<mode>3): Likewise.
19421 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19422 (*<code><mode>3): Likewise.
19423 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19424 (ia64_expand_atomic_op): Likewise.
19425 (ia64_arg_type): Likewise.
19426 (ia64_mode_to_int): Likewise.
19427 (ia64_scalar_mode_supported_p): Likewise.
19428 (ia64_vector_mode_supported_p): Likewise.
19429 (expand_vec_perm_broadcast): Likewise.
19430 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19431 (iq2000_function_arg_advance): Likewise.
19432 (iq2000_function_arg): Likewise.
19433 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19434 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19435 (m68k_libcall_value): Likewise.
19436 (m68k_function_value): Likewise.
19437 (sched_attr_op_type): Likewise.
19438 * config/mcore/mcore.c (mcore_output_move): Likewise.
19439 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19441 (microblaze_function_arg): Likewise.
19442 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19443 (mips_print_operand): Likewise.
19444 (mips_mode_ok_for_mov_fmt_p): Likewise.
19445 (mips_vector_mode_supported_p): Likewise.
19446 (mips_preferred_simd_mode): Likewise.
19447 (mips_expand_vpc_loongson_even_odd): Likewise.
19448 (mips_expand_vec_unpack): Likewise.
19449 (mips_expand_vi_broadcast): Likewise.
19450 (mips_expand_vector_init): Likewise.
19451 (mips_expand_vec_reduc): Likewise.
19452 (mips_expand_msa_cmp): Likewise.
19453 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19454 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19455 (cc_flags_for_mode): Likewise.
19456 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19457 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19458 (nds32_output_casesi_pc_relative): Likewise.
19459 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19460 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19461 (nvptx_gen_unpack): Likewise.
19462 (nvptx_gen_pack): Likewise.
19463 (nvptx_gen_shuffle): Likewise.
19464 (nvptx_gen_wcast): Likewise.
19465 (nvptx_preferred_simd_mode): Likewise.
19466 * config/pa/pa.c (pa_secondary_reload): Likewise.
19467 * config/pa/predicates.md (base14_operand): Likewise.
19468 * config/powerpcspe/powerpcspe-c.c
19469 (altivec_resolve_overloaded_builtin): Likewise.
19470 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19472 (rs6000_preferred_simd_mode): Likewise.
19473 (output_vec_const_move): Likewise.
19474 (rs6000_expand_vector_extract): Likewise.
19475 (rs6000_split_vec_extract_var): Likewise.
19476 (reg_offset_addressing_ok_p): Likewise.
19477 (rs6000_legitimate_offset_address_p): Likewise.
19478 (rs6000_legitimize_address): Likewise.
19479 (rs6000_emit_set_const): Likewise.
19480 (rs6000_const_vec): Likewise.
19481 (rs6000_emit_move): Likewise.
19482 (spe_build_register_parallel): Likewise.
19483 (rs6000_darwin64_record_arg_recurse): Likewise.
19484 (swap_selector_for_mode): Likewise.
19485 (spe_init_builtins): Likewise.
19486 (paired_init_builtins): Likewise.
19487 (altivec_init_builtins): Likewise.
19488 (do_load_for_compare): Likewise.
19489 (rs6000_generate_compare): Likewise.
19490 (rs6000_expand_float128_convert): Likewise.
19491 (emit_load_locked): Likewise.
19492 (emit_store_conditional): Likewise.
19493 (rs6000_output_function_epilogue): Likewise.
19494 (rs6000_handle_altivec_attribute): Likewise.
19495 (rs6000_function_value): Likewise.
19496 (emit_fusion_gpr_load): Likewise.
19497 (emit_fusion_p9_load): Likewise.
19498 (emit_fusion_p9_store): Likewise.
19499 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19500 (fusion_gpr_mem_load): Likewise.
19501 (fusion_addis_mem_combo_load): Likewise.
19502 (fusion_addis_mem_combo_store): Likewise.
19503 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19504 (fusion_gpr_mem_load): Likewise.
19505 (fusion_addis_mem_combo_load): Likewise.
19506 (fusion_addis_mem_combo_store): Likewise.
19507 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19509 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19510 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19511 (rs6000_preferred_simd_mode): Likewise.
19512 (output_vec_const_move): Likewise.
19513 (rs6000_expand_vector_extract): Likewise.
19514 (rs6000_split_vec_extract_var): Likewise.
19515 (reg_offset_addressing_ok_p): Likewise.
19516 (rs6000_legitimate_offset_address_p): Likewise.
19517 (rs6000_legitimize_address): Likewise.
19518 (rs6000_emit_set_const): Likewise.
19519 (rs6000_const_vec): Likewise.
19520 (rs6000_emit_move): Likewise.
19521 (rs6000_darwin64_record_arg_recurse): Likewise.
19522 (swap_selector_for_mode): Likewise.
19523 (paired_init_builtins): Likewise.
19524 (altivec_init_builtins): Likewise.
19525 (rs6000_expand_float128_convert): Likewise.
19526 (emit_load_locked): Likewise.
19527 (emit_store_conditional): Likewise.
19528 (rs6000_output_function_epilogue): Likewise.
19529 (rs6000_handle_altivec_attribute): Likewise.
19530 (rs6000_function_value): Likewise.
19531 (emit_fusion_gpr_load): Likewise.
19532 (emit_fusion_p9_load): Likewise.
19533 (emit_fusion_p9_store): Likewise.
19534 * config/rx/rx.c (rx_gen_move_template): Likewise.
19535 (flags_from_mode): Likewise.
19536 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19537 (s390_slb_comparison): Likewise.
19538 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19539 (s390_vector_mode_supported_p): Likewise.
19540 (s390_cc_modes_compatible): Likewise.
19541 (s390_match_ccmode_set): Likewise.
19542 (s390_canonicalize_comparison): Likewise.
19543 (s390_emit_compare_and_swap): Likewise.
19544 (s390_branch_condition_mask): Likewise.
19545 (s390_rtx_costs): Likewise.
19546 (s390_secondary_reload): Likewise.
19547 (__SECONDARY_RELOAD_CASE): Likewise.
19548 (s390_expand_cs): Likewise.
19549 (s390_preferred_simd_mode): Likewise.
19550 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19551 * config/sh/sh.c (sh_print_operand): Likewise.
19552 (dump_table): Likewise.
19553 (sh_secondary_reload): Likewise.
19554 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19555 * config/sh/sh.md (casesi_worker_1): Likewise.
19556 (casesi_worker_2): Likewise.
19557 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19558 (fcc_comparison_operator): Likewise.
19559 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19560 (emit_soft_tfmode_cvt): Likewise.
19561 (sparc_preferred_simd_mode): Likewise.
19562 (output_cbranch): Likewise.
19563 (sparc_print_operand): Likewise.
19564 (sparc_expand_vec_perm_bmask): Likewise.
19565 (vector_init_bshuffle): Likewise.
19566 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19567 (spu_vector_mode_supported_p): Likewise.
19568 (spu_expand_insv): Likewise.
19569 (spu_emit_branch_or_set): Likewise.
19570 (spu_handle_vector_attribute): Likewise.
19571 (spu_builtin_splats): Likewise.
19572 (spu_builtin_extract): Likewise.
19573 (spu_builtin_promote): Likewise.
19574 (spu_expand_sign_extend): Likewise.
19575 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19576 (tilegx_simd_int): Likewise.
19577 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19578 (tilepro_simd_int): Likewise.
19579 * config/v850/v850.c (const_double_split): Likewise.
19580 (v850_print_operand): Likewise.
19581 (ep_memory_offset): Likewise.
19582 * config/vax/vax.c (vax_rtx_costs): Likewise.
19583 (vax_output_int_move): Likewise.
19584 (vax_output_int_add): Likewise.
19585 (vax_output_int_subtract): Likewise.
19586 * config/visium/predicates.md (visium_branch_operator): Likewise.
19587 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19588 (visium_print_operand_address): Likewise.
19589 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19590 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19591 (xtensa_expand_conditional_branch): Likewise.
19592 (xtensa_copy_incoming_a7): Likewise.
19593 (xtensa_output_literal): Likewise.
19594 * dfp.c (decimal_real_maxval): Likewise.
19595 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19598 Alan Hayward <alan.hayward@arm.com>
19599 David Sherwood <david.sherwood@arm.com>
19601 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19602 (mode_nunits_inline): Likewise.
19603 (mode_inner_inline): Likewise.
19604 (mode_unit_size_inline): Likewise.
19605 (mode_unit_precision_inline): Likewise.
19606 (emit_insn_modes_h): Likewise. Also emit a #define of the
19608 (emit_mode_wider): Add an E_ prefix to mode names.
19609 (emit_mode_complex): Likewise.
19610 (emit_mode_inner): Likewise.
19611 (emit_mode_adjustments): Likewise.
19612 (emit_mode_int_n): Likewise.
19613 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19614 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19615 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19616 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19617 (CRC32_BUILTIN, ENTRY): Likewise.
19618 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19619 (aarch64_pop_regs): Likewise.
19620 (aarch64_process_components): Likewise.
19621 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19622 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19623 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19624 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19625 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19626 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19628 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19629 (pcmpistr): Likewise.
19630 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19631 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19632 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19634 * config/rl78/rl78.c (mduc_regs): Likewise.
19635 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19636 (htm_expand_builtin): Likewise.
19637 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19638 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19639 * config/xtensa/xtensa.c (print_operand): Likewise.
19640 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19641 (NUM_MODE_VECTOR_INT): Likewise.
19642 * genoutput.c (null_operand): Likewise.
19643 (output_operand_data): Likewise.
19644 * genrecog.c (print_parameter_value): Likewise.
19645 * lra.c (debug_operand_data): Likewise.
19647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19649 * dwarf2out.c (scompare_loc_descriptor_wide)
19650 (scompare_loc_descriptor_narrow): New functions, split out from...
19651 (scompare_loc_descriptor): ...here.
19652 * expmed.c (emit_store_flag_int): New function, split out from...
19653 (emit_store_flag): ...here.
19655 2017-08-30 Richard Biener <rguenther@suse.de>
19657 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19658 (dwarf2out_early_finish): Move setting of AT_pubnames from
19659 early debug output to early finish.
19661 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19663 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19664 and -mdata-region to the assembler.
19666 2017-08-30 Richard Biener <rguenther@suse.de>
19668 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19670 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19671 (dwarf2out_early_global_decl): Only generate a DIE for the
19672 abstract origin if it doesn't already exist or is a declaration DIE.
19673 (resolve_addr): Do not add the linkage name twice when
19674 generating a stub DIE for the DW_TAG_GNU_call_site target.
19676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19678 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19679 Use machine_mode rather than int for arg1_mode.
19681 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19684 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19685 that the second argument of the built-in functions to unpack
19686 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19687 switch statement instead a lot of if statements.
19688 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19689 Allow 64-bit values to be in Altivec registers as well as
19690 traditional floating point registers.
19691 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19693 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19695 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19696 MAX_REGS_PER_ADDRESS == 1.
19698 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19700 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19701 * config/i386/i386.c (ix86_option_override_internal): Simplify
19702 setting of opts->x_flag_entry.
19704 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19705 Jakub Jelinek <jakub@redhat.com>
19706 Richard Biener <rguenther@suse.de>
19708 PR tree-optimization/81503
19709 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19710 folded constant fits in the target type; reorder tests for clarity.
19712 2017-08-29 Martin Liska <mliska@suse.cz>
19714 * passes.def: Include pass_lower_switch.
19715 * stmt.c (dump_case_nodes): Remove and move to
19716 tree-switch-conversion.
19717 (case_values_threshold): Likewise.
19718 (expand_switch_as_decision_tree_p): Likewise.
19719 (emit_case_decision_tree): Likewise.
19720 (expand_case): Likewise.
19721 (balance_case_nodes): Likewise.
19722 (node_has_low_bound): Likewise.
19723 (node_has_high_bound): Likewise.
19724 (node_is_bounded): Likewise.
19725 (emit_case_nodes): Likewise.
19726 (struct simple_case_node): New struct.
19727 (add_case_node): Remove.
19728 (emit_case_dispatch_table): Use vector instead of case_list.
19729 (reset_out_edges_aux): Remove.
19730 (compute_cases_per_edge): Likewise.
19731 (expand_case): Build list of simple_case_node.
19732 (expand_sjlj_dispatch_table): Use it.
19733 * tree-switch-conversion.c (struct case_node): Moved from
19734 stmt.c and adjusted.
19735 (emit_case_nodes): Likewise.
19736 (node_has_low_bound): Likewise.
19737 (node_has_high_bound): Likewise.
19738 (node_is_bounded): Likewise.
19739 (case_values_threshold): Likewise.
19740 (reset_out_edges_aux): Likewise.
19741 (compute_cases_per_edge): Likewise.
19742 (add_case_node): Likewise.
19743 (dump_case_nodes): Likewise.
19744 (balance_case_nodes): Likewise.
19745 (expand_switch_as_decision_tree_p): Likewise.
19746 (emit_jump): Likewise.
19747 (emit_case_decision_tree): Likewise.
19748 (try_switch_expansion): Likewise.
19749 (do_jump_if_equal): Likewise.
19750 (emit_cmp_and_jump_insns): Likewise.
19751 (fix_phi_operands_for_edge): New function.
19752 (record_phi_operand_mapping): Likewise.
19753 (class pass_lower_switch): New pass.
19754 (pass_lower_switch::execute): New function.
19755 (make_pass_lower_switch): Likewise.
19756 (conditional_probability):
19757 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19758 * tree-pass.h: Add make_pass_lower_switch.
19760 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19763 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19766 2017-08-29 Richard Biener <rguenther@suse.de>
19768 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19769 we do not add a DW_AT_inline attribute twice.
19770 (gen_subprogram_die): Remove code setting DW_AT_inline on
19771 DECL_ABSTRACT_P nodes.
19773 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19775 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19776 calls to internal functions.
19777 (gimplify_modify_expr): Likewise.
19778 * tree-call-cdce.c (use_internal_fn): Likewise.
19779 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19780 (convert_to_divmod): Set the nothrow flag.
19781 * tree-if-conv.c (predicate_mem_writes): Likewise.
19782 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19783 (vectorizable_call): Likewise.
19784 (vectorizable_store): Likewise.
19785 (vectorizable_load): Likewise.
19786 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19787 (vect_recog_mask_conversion_pattern): Likewise.
19789 2017-08-29 Martin Liska <mliska@suse.cz>
19792 * gcc.c (driver_handle_option): Add new argument.
19793 * opts-common.c (handle_option): Pass
19794 target_option_override_hook.
19795 * opts-global.c (lang_handle_option): Add new option.
19796 (set_default_handlers): Add new argument.
19797 (decode_options): Likewise.
19798 * opts.c (target_handle_option): Likewise.
19799 (common_handle_option): Call target_option_override_hook.
19800 * opts.h (struct cl_option_handler_func): Add hook for
19801 target option override.
19802 (struct cl_option_handlers): Likewise.
19803 (set_default_handlers): Add new argument.
19804 (decode_options): Likewise.
19805 (common_handle_option): Likewise.
19806 (target_handle_option): Likewise.
19807 * toplev.c (toplev::main): Pass targetm.target_option.override
19810 2017-08-29 Richard Biener <rguenther@suse.de>
19811 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19813 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19814 life to the active subtree.
19816 2017-08-28 Jeff Law <law@redhat.com>
19818 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19819 derive_equivalences.
19820 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19821 Code moved into....
19822 (edge_info::derive_equivalences): New private member function
19824 * tree-ssa-dom.c (class edge_info): Changed from a struct
19825 to a class. Add ctor/dtor, methods and data members.
19826 (edge_info::edge_info): Renamed from allocate_edge_info.
19827 Initialize additional members.
19828 (edge_info::~edge_info): New.
19829 (free_dom_edge_info): Delete the edge info.
19830 (record_edge_info): Use new class & associated member functions.
19831 Tighten forms for testing for edge equivalences.
19832 (record_temporary_equivalences): Iterate over the simple
19833 equivalences rather than assuming there's only one per edge.
19834 (cprop_into_successor_phis): Iterate over the simple
19835 equivalences rather than assuming there's only one per edge.
19836 (optimize_stmt): Use operand_equal_p rather than pointer
19837 equality for mini-DSE code.
19839 2017-08-28 Nathan Sidwell <nathan@acm.org>
19841 * gcc.c (execute): Fold SIGPIPE handling into switch
19842 statement. Adjust internal error message.
19844 2017-08-28 Richard Biener <rguenther@suse.de>
19847 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19848 Do nothing for removed DIEs.
19850 2017-08-28 Richard Biener <rguenther@suse.de>
19852 PR tree-optimization/81977
19853 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19856 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19859 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19860 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19861 using targetm.gen_mem_thread_fence.
19863 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19866 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19867 predicate to register_operand. Reorder operands.
19868 (*btr<mode>): Ditto.
19869 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19870 (*btr<mode>_mask): Ditto.
19872 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19874 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19875 * config/rs6000/xmmintrin.h: New file.
19876 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19878 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19881 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19882 parameter and_insn and return it.
19883 (recombine_lvx_pattern): Insert a copy to ensure availability of
19884 the base register of the copied masking operation at the point of
19885 the instruction replacement.
19886 (recombine_stvx_pattern): Likewise.
19888 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19890 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19891 undocumented switches.
19892 (-mpower9-dform-vector): Likewise.
19893 (-mpower9-dform): Likewise.
19894 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19895 comments to delete references to -mpower9-dform* switches.
19896 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19897 Delete reference to -mpower9-dform* switches, test for
19898 -mpower9-vector instead.
19899 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19900 (OTHER_P9_VECTOR_MASKS): Likewise.
19901 (POWERPC_MASKS): Likewise.
19902 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19903 tests against -mpower9-dform* to -mpower9-vector. Delete code
19904 that checked for -mpower9-dform* consistancy with other options.
19905 Add test for -mpower9-misc to enable other power9 switches.
19906 (rs6000_init_hard_regno_mode_ok): Likewise.
19907 (rs6000_option_override_internal): Likewise.
19908 (rs6000_emit_prologue): Likewise.
19909 (rs6000_emit_epilogue): Likewise.
19910 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19911 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19912 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19914 (emit_fusion_p9_store): Likewise.
19915 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19916 resetting these macros if the assembler does not support ISA 3.0
19918 (TARGET_P9_DFORM_VECTOR): Likewise.
19919 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19920 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19922 2017-08-25 Alan Modra <amodra@gmail.com>
19924 PR rtl-optimization/81747
19925 * cse.c (cse_extended_basic_block): Don't attempt to record
19926 equivalences for degenerate conditional branches that branch
19927 to their fall-through.
19929 2017-08-24 Martin Sebor <msebor@redhat.com>
19931 PR middle-end/81908
19932 * gimple-fold.c (size_must_be_zero_p): New function.
19933 (gimple_fold_builtin_memory_op): Call it.
19935 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19937 * config/rs6000/mm_malloc.h: New file.
19939 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19941 PR tree-optimization/81913
19942 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19943 analysis when either IVs in condition can wrap.
19945 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19947 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19948 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19950 2017-08-24 Richard Biener <rguenther@suse.de>
19953 * targhooks.c (default_target_can_inline_p): Properly
19954 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19955 is present and always compare.
19956 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19957 infer -mfpmath=sse from TARGET_SSE_P.
19958 (ix86_can_inline_p): Properly use target_option_default_node when
19959 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19961 2017-08-24 Richard Biener <rguenther@suse.de>
19964 * dwarf2out.c (output_die): Handle flag_generate_offload like
19966 (output_comp_unit): Likewise.
19967 (gen_array_type_die): Likewise.
19968 (dwarf2out_early_finish): Likewise.
19969 (note_variable_value_in_expr): Likewise.
19970 (dwarf2out_finish): Likewise. Adjust assert.
19971 * cgraphunit.c (symbol_table::compile): Move setting of
19972 flag_generate_offload earlier ...
19973 (symbol_table::finalize_compilation_unit): ... here, before
19974 early debug finalization.
19976 2017-08-24 Richard Biener <rguenther@suse.de>
19978 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19979 and ipa-fnsummary.h.
19980 (ix86_can_inline_p): When ix86_fpmath flags do not match
19981 check whether the callee uses FP math at all.
19983 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19985 PR middle-end/81931
19986 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19987 instead of TYPE_PRECISION.
19989 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19990 Alan Hayward <alan.hayward@arm.com>
19991 David Sherwood <david.sherwood@arm.com>
19993 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19995 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19997 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19999 (ix86_frame::stack_realign_allocate): New field.
20000 (struct machine_frame_state): Modify comments.
20001 (machine_frame_state::sp_realigned_fp_end): New field.
20002 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20003 layout calculation.
20004 (sp_valid_at): Add assertion to assure no attempt to access invalid
20005 offset of a realigned stack.
20006 (fp_valid_at): Likewise.
20007 (choose_baseaddr): Modify comments.
20008 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20009 ix86_expand_prologue.
20010 (ix86_expand_prologue): Modify stack realignment and allocation.
20011 (ix86_expand_epilogue): Modify comments.
20012 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20013 avx2_runtime, avx512f, and avx512f_runtime.
20015 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20017 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20018 (mstackrealign): Do not init to -1.
20019 * config/i386/i386.c (ix86_option_override_internal):
20020 Check opts_set, not opts when setting default value of
20021 opts->x_ix86_force_align_arg_pointer.
20023 2017-08-23 Richard Biener <rguenther@suse.de>
20025 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20026 lang_hooks.decl_printable_name.
20027 * print-rtl-function.c (print_rtx_function): Likewise.
20028 * tree-pretty-print.c (dump_function_header): Likewise.
20030 2017-08-23 Richard Biener <rguenther@suse.de>
20033 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20034 -g0 at compile-time.
20036 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20038 PR middle-end/19706
20039 * doc/sourcebuild.texi (Other hardware attributes):
20042 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20044 PR middle-end/19706
20045 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20046 Fix single-use check.
20048 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20050 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20052 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20054 * doc/install.texi: Modify to add more details on running selected
20057 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20059 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20060 is combined with -mabi=ms.
20061 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20064 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20066 PR tree-optimization/81488
20067 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20068 and cached_basis fields.
20069 (MAX_SPREAD): New constant.
20070 (alloc_cand_and_find_basis): Initialize new fields.
20071 (clear_visited): New function.
20072 (create_phi_basis_1): Rename from create_phi_basis, set visited
20073 and cached_basis fields.
20074 (create_phi_basis): New wrapper function.
20075 (phi_add_costs_1): Rename from phi_add_costs, add spread
20076 parameter, set visited field, short-circuit when limits reached.
20077 (phi_add_costs): New wrapper function.
20078 (record_phi_increments_1): Rename from record_phi_increments, set
20080 (record_phi_increments): New wrapper function.
20081 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20082 (phi_incr_cost): New wrapper function.
20083 (all_phi_incrs_profitable_1): Rename from
20084 all_phi_incrs_profitable, set visited field.
20085 (all_phi_incrs_profitable): New wrapper function.
20087 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20088 Alan Hayward <alan.hayward@arm.com>
20089 David Sherwood <david.sherwood@arm.com>
20091 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20092 that takes the outer and inner modes.
20093 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20094 comparison as the canonical test for a paradoxical subreg.
20095 * combine.c (simplify_set): Use paradoxical_subreg_p.
20096 (make_extraction): Likewise.
20097 (force_to_mode): Likewise.
20098 (rtx_equal_for_field_assignment_p): Likewise.
20099 (gen_lowpart_for_combine): Likewise.
20100 (simplify_comparison): Likewise.
20101 * cse.c (equiv_constant): Likewise.
20102 * expmed.c (store_bit_field_1): Likewise.
20103 * final.c (alter_subreg): Likewise.
20104 * fwprop.c (propagate_rtx): Likewise.
20105 (forward_propagate_subreg): Likewise.
20106 * ira-conflicts.c (ira_build_conflicts): Likewise.
20107 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20108 * lra-constraints.c (curr_insn_transform): Likewise.
20109 (split_reg): Likewise.
20110 * lra-eliminations.c (move_plus_up): Likewise.
20111 (lra_eliminate_regs_1): Likewise.
20112 * recog.c (general_operand): Likewise.
20113 * ree.c (combine_reaching_defs): Likewise.
20114 * reload.c (push_reload): Likewise.
20115 (find_reloads): Likewise.
20116 * reload1.c (elimination_effects): Likewise.
20117 (compute_reload_subreg_offset): Likewise.
20118 (choose_reload_regs): Likewise.
20119 * rtlanal.c (subreg_lsb_1): Likewise.
20120 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20121 (simplify_subreg): Likewise.
20122 * var-tracking.c (track_loc_p): Likewise.
20123 * emit-rtl.c (byte_lowpart_offset): Likewise.
20124 (paradoxical_subreg_p): Delete out-of-line definition.
20126 2017-08-22 Jeff Law <law@redhat.com>
20128 PR tree-optimization/81741
20129 PR tree-optimization/71947
20130 * tree-ssa-dom.c: Include tree-inline.h.
20131 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20132 equivalences if one is more expensive to compute than the other.
20133 * tree-ssa-scopedtables.h (class const_or_copies): Make
20134 record_const_or_copy_raw method private.
20135 (class avail_exprs_stack): New method simplify_binary_operation.
20136 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20137 avail_exprs_stack::simplify_binary_operation as needed.
20138 (avail_exprs_stack::simplify_binary_operation): New function.
20140 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20142 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20143 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20144 (DOT_SYMBOLS): Likewise.
20145 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20146 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20147 (RS6000_ABI_NAME): Likewise.
20148 (TARGET_CMODEL): Likewise.
20149 (TOC_SECTION_ASM_OP): Likewise.
20150 (SET_CMODEL): New macro.
20151 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20153 2017-08-22 Richard Biener <rguenther@suse.de>
20155 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20156 to allow for free-lang-data replacements similar to verify_type_variant.
20158 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20159 Alan Hayward <alan.hayward@arm.com>
20160 David Sherwood <david.sherwood@arm.com>
20162 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20163 VOIDmode for the LABEL_REF.
20165 2017-08-22 Richard Biener <rguenther@suse.de>
20167 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20169 2017-08-22 Richard Biener <rguenther@suse.de>
20171 * common.opt (feliminate-dwarf2-dups): Ignore.
20172 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20173 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20174 same_die_p_wrap, compute_section_prefix,
20175 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20176 (comdat_symbol_id, comdat_symbol_number): Likewise.
20177 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20179 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20180 (output_die): Mark unreachable path unreachable.
20181 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20182 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20183 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20184 (dwarf2out_early_finish): Likewise.
20186 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20188 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20190 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20193 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20194 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20195 (avr_attribute_table) <io, io_low, address>: Initialize
20196 .decl_required with true.
20198 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20200 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20201 undocumented debugging options.
20202 (-mvsx-scalar-double): Likewise.
20203 (-mallow-df-permute): Likewise.
20204 (-mvectorize-builtins): Likewise.
20205 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20206 (rs6000_builtin_vectorized_function): Likewise.
20207 (rs6000_builtin_md_vectorized_function): Likewise.
20208 (rs6000_opt_vars): Likewise.
20210 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20213 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20214 (*btrq_imm): Rename from *btrq.
20215 (*btcq_imm): Rename from *btcq.
20216 (btsc): New code attribute.
20217 (*<btsc><mode>): New insn pattern.
20218 (*btr<mode>): Ditto.
20219 (*<btsc><mode>_mask): New insn_and_split pattern.
20220 (*btr<mode>_mask): Ditto.
20222 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20223 Alan Hayward <alan.hayward@arm.com>
20224 David Sherwood <david.sherwood@arm.com>
20226 * function.c (pad_below): Simplify padding calculation.
20228 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20229 Alan Hayward <alan.hayward@arm.com>
20230 David Sherwood <david.sherwood@arm.com>
20232 * target.def (function_prologue): Remove frame size argument.
20233 (function_epilogue): Likewise.
20234 * doc/tm.texi: Regenerate.
20235 * final.c (final_start_function): Update call to function_prologue.
20236 (final_end_function): Update call to function_epilogue.
20237 (default_function_pro_epilogue): Remove frame size argument.
20238 * output.h (default_function_pro_epilogue): Likewise.
20239 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20240 (arm_output_function_prologue): Likewise.
20241 * config/frv/frv.c (frv_function_prologue): Likewise.
20242 (frv_function_epilogue): Likewise.
20243 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20244 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20245 (ia64_output_function_epilogue): Likewise.
20246 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20247 (m32r_output_function_epilogue): Likewise.
20248 * config/microblaze/microblaze.c (microblaze_function_prologue)
20249 (microblaze_function_epilogue): Likewise.
20250 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20251 (mips_output_function_epilogue): Likewise.
20252 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20253 (mmix_target_asm_function_epilogue): Likewise.
20254 * config/msp430/msp430.c (msp430_start_function): Likewise.
20255 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20256 (nds32_asm_function_epilogue): Likewise.
20257 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20258 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20259 (pa_output_function_epilogue): Likewise.
20260 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20261 (rs6000_output_function_epilogue): Likewise.
20262 * config/rl78/rl78.c (rl78_start_function): Likewise.
20263 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20264 (rs6000_output_function_epilogue): Likewise.
20265 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20266 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20267 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20268 (sparc_asm_function_epilogue): Likewise.
20270 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20272 * tree.h (type_has_mode_precision_p): New function.
20273 * convert.c (convert_to_integer_1): Use it.
20274 * expr.c (expand_expr_real_2): Likewise.
20275 (expand_expr_real_1): Likewise.
20276 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20277 * match.pd: Likewise.
20278 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20279 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20280 * tree-tailcall.c (process_assignment): Likewise.
20281 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20282 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20283 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20284 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20285 (vectorizable_assignment): Likewise.
20286 (vectorizable_shift): Likewise.
20287 (vectorizable_operation): Likewise.
20288 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20290 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20292 * match.pd: Add pow (C, x) simplification.
20294 2017-08-21 Richard Biener <rguenther@suse.de>
20296 PR tree-optimization/81900
20297 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20298 for blocks with abnormal predecessors.
20299 (compute_antic): Do not set visited flag prematurely.
20301 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20304 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20306 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20308 * stor-layout.h (vector_type_mode): Move to...
20309 * tree.h (vector_type_mode): ...here.
20310 * stor-layout.c (vector_type_mode): Move to...
20311 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20313 2017-08-21 Richard Biener <rguenther@suse.de>
20315 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20316 register_external_die hooks.
20317 (debug_false_tree_charstarstar_uhwistar): Declare.
20318 (debug_nothing_tree_charstar_uhwi): Likewise.
20319 * debug.c (do_nothing_debug_hooks): Adjust.
20320 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20321 (debug_nothing_tree_charstar_uhwi): Likewise.
20322 * dbxout.c (dbx_debug_hooks): Adjust.
20323 (xcoff_debug_hooks): Likewise.
20324 * sdbout.c (sdb_debug_hooks): Likewise.
20325 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20326 * dwarf2out.c (macinfo_label_base): New global.
20327 (dwarf2out_register_external_die): New function for the
20328 register_external_die hook.
20329 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20330 (dwarf2_debug_hooks): Use them.
20331 (dwarf2_lineno_debug_hooks): Adjust.
20332 (struct die_struct): Add with_offset flag.
20333 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20334 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20335 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20336 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20337 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20338 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20339 defining section names for the early LTO debug variants.
20340 (reset_indirect_string): New helper.
20341 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20342 (print_dw_val): Add support for offsetted symbol references.
20343 (get_ultimate_context): Split out from is_cxx.
20344 (is_cxx): Use get_ultimate_context.
20345 (is_fortran): Add decl overload.
20346 (compute_comp_unit_symbol): Split out worker from
20347 compute_section_prefix.
20348 (compute_section_prefix): Call compute_comp_unit_symbol and
20349 set comdat_type_p here.
20350 (output_die): Skip DIE symbol output for the LTO added one.
20351 Handle DIE symbol references with offset.
20352 (output_comp_unit): Guard section name mangling properly.
20353 For LTO debug sections emit a symbol at the section beginning
20354 which we use to refer to its DIEs.
20355 (add_abstract_origin_attribute): For DIEs registered via
20356 dwarf2out_register_external_die directly refer to the early
20357 DIE rather than indirectly through the shadow one we created.
20358 Remove obsolete call to dwarf2out_abstract_function for
20359 non-function/block origins.
20360 (gen_array_type_die): When generating early LTO debug do
20361 not emit DW_AT_string_length.
20362 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20363 late when in LTO. As suggested place a gcc_unreachable for
20364 the DECL_ABSTRACT_P case.
20365 (gen_subprogram_die): Avoid another specification DIE
20366 for early built declarations/definitions for the late LTO case.
20367 (gen_variable_die): Add type references for late duplicated VLA dies
20369 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20370 we have the abstract instance already.
20371 (process_scope_var): Adjust decl DIE contexts in LTO which
20372 first puts them in limbo.
20373 (gen_decl_die): Do not generate type DIEs late apart from
20374 types for VLAs or for decls we do not yet have a DIE. Do not
20375 call dwarf2out_abstract_function late.
20376 (dwarf2out_early_global_decl): Make sure to create DIEs
20377 for abstract instances of a decl first.
20378 (dwarf2out_late_global_decl): Adjust comment.
20379 (output_macinfo_op): With multiple macro sections use
20380 macinfo_label_base to distinguish labels.
20381 (output_macinfo): Likewise. Update macinfo_label_base.
20382 Pass in the line info label.
20383 (note_variable_value_in_expr): When generating LTO resolve
20384 all variable values here by generating DIEs as needed.
20385 (init_sections_and_labels): Add early LTO debug flag parameter
20386 and generate different sections and names if set. Add generation
20387 counter for the labels so we can have multiple of them.
20388 (reset_dies): Helper to allow DIEs to be output multiple times.
20389 (dwarf2out_finish): When outputting DIEs to the fat part of an
20390 LTO object first reset DIEs.
20391 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20392 (modified_type_die): Check for decl_ultimate_origin being self
20394 (gen_type_die_with_usage): Likewise.
20395 (gen_typedef_die): Allow decl_ultimate_origin being self.
20396 (set_decl_abstract_flags): Remove.
20397 (set_block_abstract_flags): Likewise.
20398 (dwarf2out_abstract_function): Treat the early generated DIEs
20399 as the abstract copy and only add DW_AT_inline and
20400 DW_AT_artificial here and call set_decl_origin_self.
20401 If the DIE has an abstract origin don't do anything.
20402 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20403 if we have none yet (Go fails to build one, PR78628).
20404 (variably_modified_type_p): Prevent endless recursion for Ada
20405 cyclic pointer types.
20406 * lto-streamer-in.c: Include debug.h.
20407 (dref_queue): New global.
20408 (lto_read_tree_1): Stream in DIE references.
20409 (lto_input_tree): Register DIE references.
20410 (input_function): Stream DECL_DEBUG_ARGS.
20411 * lto-streamer-out.c: Include debug.h.
20412 (lto_write_tree_1): Output DIE references.
20413 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20414 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20415 (output_function): Stream DECL_DEBUG_ARGS.
20416 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20417 Stream DECL_ABSTRACT_ORIGIN.
20418 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20419 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20420 DECL_CONTEXT for file-scope decls.
20421 * lto-streamer.h (struct dref_entry): Declare.
20422 (dref_queue): Likewise.
20423 * cfgexpand.c (pass_expand::execute): Do not call the
20424 outlining_inline_function hook here.
20425 * lto-wrapper.c (debug_obj): New global.
20426 (tool_cleanup): Unlink it if required.
20427 (debug_objcopy): New function.
20428 (run_gcc): Handle early debug sections in the IL files by
20429 extracting them to separate files, partially linkin them and
20430 feeding the result back as result to the linker.
20431 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20432 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20433 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20434 sections into a separate segment.
20435 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20437 (darwin_asm_dwarf_section): Likewise.
20438 (darwin_asm_output_dwarf_offset): Likewise.
20439 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20441 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20442 Alan Hayward <alan.hayward@arm.com>
20443 David Sherwood <david.sherwood@arm.com>
20445 * read-md.h (md_reader::record_potential_iterator_use): Replace
20446 pointer argument with an rtx and an index.
20447 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20448 (apply_mode_iterator): Likewise.
20449 (apply_code_iterator): Likewise.
20450 (apply_int_iterator): Likewise.
20451 (apply_subst_iterator): Likewise.
20452 (record_iterator_use): Likewise.
20453 (record_attribute_use): Likewise.
20454 (md_reader::record_potential_iterator_use): Likewise. Update calls
20455 to record_iterator_use and apply_iterator.
20456 (iterator_use): Replace ptr with x and index.
20457 (attribute_use): Likewise.
20458 (apply_attribute_uses): Update calls to apply_iterator.
20459 (apply_iterators): Likewise. Update initialization of iterator_use.
20460 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20461 and record_potential_iterator_use.
20462 (rtx_reader::read_rtx_operand): Likewise.
20464 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20465 Alan Hayward <alan.hayward@arm.com>
20466 David Sherwood <david.sherwood@arm.com>
20468 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20471 2017-08-21 Richard Biener <rguenther@suse.de>
20473 PR middle-end/81884
20474 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20475 at struct end conservatively when comparing common bases.
20477 2017-08-21 Richard Biener <rguenther@suse.de>
20479 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20480 (mem_ref_in_stmt): Remove.
20481 (determine_max_movement): Use ref index to get at the reference.
20482 (invariantness_dom_walker::before_dom_children): Deal with
20483 lim data already initialized.
20484 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20486 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20488 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20489 (TARGET_ISA_ROUND): Ditto.
20490 (TARGET_ROUND): Ditto.
20491 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20492 * config/i386/i386.md: Ditto.
20493 * config/i386/sse.md: Ditto.
20494 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20495 with OPTION_MASK_ISA_SSE4_1.
20497 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20500 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20501 __builtin_ia32_lzcnt_u16.
20503 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20506 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20507 (rs6000_set_current_function): Rewrite function to use it.
20509 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20512 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20513 and TYPE_WARN_IF_NOT_ALIGN.
20514 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20515 (handle_warn_if_not_align): New.
20516 (place_union_field): Call handle_warn_if_not_align.
20517 (place_field): Call handle_warn_if_not_align.
20518 Copy TYPE_WARN_IF_NOT_ALIGN.
20519 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20520 (layout_type): Likewise.
20521 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20523 (tree_decl_common): Add warn_if_not_align.
20524 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20525 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20526 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20527 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20528 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20529 * doc/extend.texi: Document warn_if_not_aligned attribute.
20530 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20532 2017-08-17 Martin Liska <mliska@suse.cz>
20535 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20536 (get_data_dependence): Use it as pointer type.
20537 (distribute_loop): Likewise.
20539 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20541 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20542 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20543 (p8_vmrgow_<mode>_direct): New define_insn.
20544 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20545 handle endianness for vmrgew and vmrgow permute patterns.
20547 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20549 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20550 * config/rs6000/rs6000-cpus.def: Remove comment.
20551 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20552 (POWERPC_MASKS): Likewise.
20553 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20554 use of TARGET_VSX_TIMODE.
20555 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20556 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20557 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20558 (rs6000_option_override_internal): Remove dead code.
20559 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20560 (rs6000_legitimize_reload_address): Likewise.
20561 (rs6000_legitimate_address_p): Likewise.
20562 (rs6000_opt_masks): Delete "vsx-timode".
20563 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20564 from function comment.
20565 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20566 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20567 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20569 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20570 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20571 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20572 (VSX_TI): Likewise.
20574 (define_peephole2): Likewise.
20576 2017-08-17 Martin Sebor <msebor@redhat.com>
20579 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20580 past the end of an array.
20581 (test_pp_format): Add test cases.
20583 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20585 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20586 missing ECF_NOTHROW flags.
20588 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20591 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20592 operands residing in integer registers.
20593 (*vsx_le_perm_load_<mode>): Likewise.
20594 (*vsx_le_perm_store_<mode>): Likewise.
20595 (define_peephole2): Add peepholes to optimize the above.
20597 2017-08-17 Marek Polacek <polacek@redhat.com>
20599 PR middle-end/81814
20600 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20601 to mimic what shorten_compare did. Change the return type to bool.
20602 (fold_cond_expr_with_comparison): Update call to
20603 operand_equal_for_comparison_p.
20604 (fold_ternary_loc): Likewise.
20606 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20608 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20610 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20612 2017-08-17 Richard Biener <rguenther@suse.de>
20614 * tree-ssa-structalias.c (solve_graph): When propagating
20615 to successors update the graphs succ edges and avoid duplicate work.
20617 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20620 * config/i386/i386.c (ix86_option_override_internal): Save target
20621 specific options after ix86_stack_protector_guard_reg was changed.
20623 2017-08-17 Richard Biener <rguenther@suse.de>
20625 PR tree-optimization/81827
20626 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20627 (new_var_info): Initialize it conservatively.
20628 (get_call_vi): Mark register vars.
20629 (new_scalar_tmp_constraint_exp): Likewise.
20630 (handle_rhs_call): Likewise.
20631 (handle_const_call): Likewise.
20632 (create_function_info_for): Likewise.
20633 (solve_constraints): Sort varinfos to separate register from
20634 non-register vars to pack points-to solution bitmaps during
20637 2017-08-17 Marek Polacek <polacek@redhat.com>
20639 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20641 2017-08-17 Richard Biener <rguenther@suse.de>
20643 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20644 to true when overflow is undefined and we saturated the result.
20646 2017-08-17 Alan Modra <amodra@gmail.com>
20649 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20650 Don't use store multiple if only one reg needs saving.
20651 (interesting_frame_related_regno): New function.
20652 (rs6000_frame_related): Don't emit frame info for regs that
20654 (rs6000_emit_epilogue): Likewise.
20656 2017-08-16 Nathan Sidwell <nathan@acm.org>
20658 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20659 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20660 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20661 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20662 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20663 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20664 (verify_type): Adjust for TYPE_BINFO move.
20665 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20666 process TYPE_BINFO directly.
20667 (hash_tree): Likewise.
20668 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20670 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20673 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20675 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20677 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20680 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20681 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20682 Add ix86_binary_operator_ok to insn constraint.
20683 (*iordi_1_bts): Ditto.
20684 (*xordi_1_btc): Ditto.
20685 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20686 Update corresponding peephole2 pattern.
20690 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20692 PR tree-optimization/81832
20693 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20694 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20696 2017-08-16 Marek Polacek <polacek@redhat.com>
20699 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20700 regarding size_zero_node.
20702 2017-08-16 Martin Liska <mliska@suse.cz>
20705 * config.gcc: Respect previously set extra_objs in case
20708 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20710 PR tree-optimization/81835
20711 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20712 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20713 not depend on the phi.
20715 2017-08-16 Alan Modra <amodra@gmail.com>
20717 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20720 2017-08-16 Alan Modra <amodra@gmail.com>
20722 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20723 (save_reg_p): ..into this. Update all callers.
20724 (first_reg_to_save): Simplify.
20726 2017-08-16 Alan Modra <amodra@gmail.com>
20728 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20731 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20735 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20736 constant count if that count is less than 32.
20738 2017-08-15 Nathan Sidwell <nathan@acm.org>
20740 * gcc.c (execute): Emit friendlier message if inferior is killed
20741 by an external cause.
20743 2017-08-15 Richard Biener <rguenther@suse.de>
20745 PR tree-optimization/81790
20746 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20747 CONSTRUCTORs from simplifying and VN.
20749 2017-08-14 Martin Sebor <msebor@redhat.com>
20751 * builtin-attrs.def: Add comments.
20753 2017-08-14 Martin Sebor <msebor@redhat.com>
20756 * doc/extend.texi (attribute nonstring): Document new attribute.
20758 2017-08-14 Martin Sebor <msebor@redhat.com>
20761 * tree-diagnostic.c (default_tree_printer): Handle %G.
20762 * gimple-pretty-print.h (percent_G_format): Declare new function.
20763 * gimple-pretty-print.c (percent_G_format): Define.
20764 * tree-pretty-print.c (percent_K_format): Add argument.
20766 2017-08-14 Martin Sebor <msebor@redhat.com>
20768 PR translation/79998
20769 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20770 Remove a stray space.
20772 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20775 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20776 (*iordi_1_bts): Ditto.
20777 (*xordi_1_btc): Ditto.
20779 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20782 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20783 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20785 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20786 quoted strings, and make more translator-friendly.
20787 (darwin_rs6000_override_options): Likewise.
20788 (rs6000_option_override_internal): Likewise.
20789 (rs6000_return_in_memory): Fix overlong line.
20790 (init_cmulative_args): Use quoted strings, and make more
20791 translator-friendly.
20792 (rs6000_pass_by_reference): Fix overlong line.
20793 (def_builtin): Use quoted strings.
20794 (altivec_expand_predicate_builtin): Use quoted strings, and make
20795 more translator-friendly.
20796 (htm_expand_builtin): Use quoted strings.
20797 (cpu_expand_builtin): Use quoted strings, and make more
20798 translator-friendly.
20799 (altivec_expand_builtin): Likewise.
20800 (paired_expand_predicate_builtin): Likewise.
20801 (rs6000_invalid_builtin): Likewise.
20802 (builtin_function_type): Use quoted strings.
20803 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20804 more translator-friendly.
20805 (rs6000_trampoline_init): Likewise.
20806 (rs6000_handle_altivec_attribute): Likewise.
20807 (rs6000_inner_target_options): Use quoted strings.
20808 (rs6000_disable_incompatible_switches): Likewise.
20809 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20810 strings, and make more translator-friendly.
20811 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20813 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20815 PR tree-optimization/81799
20816 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20817 cond_expr to simple gimple operand.
20819 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20821 PR middle-end/46932
20822 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20824 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20828 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20829 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20830 TARGET_GASISR_PROLOGUES.
20831 * config/avr/avr.c (avr_option_override): Same.
20832 (avr_pass_pre_proep::execute): Same.
20834 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20837 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20838 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20840 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20842 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20843 define_insn_and_split. Split to a memory load from 0 in
20844 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20845 using PTR mode iterator.
20846 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20847 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20848 (*add_tp_<mode>): Redefine as define_insn_and_split.
20849 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20850 address space. Merge with *add_tp_x32 using PTR mode iterator.
20851 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20852 Split to an add with a memory load from 0 in
20853 DEFAULT_TLS_SEG_REG address space.
20855 2017-08-12 Andrew Pinski <apinski@cavium.com>
20857 * config/aarch64/aarch64-option-extensions.def (rdma):
20858 Fix feature string to what Linux prints out in /proc/cpuinfo.
20860 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20863 * dwarf2out.c (modified_type_die): For C typedef types that have
20864 an ultimate origin, process the ultimate origin instead of the
20866 (gen_typedef_die): Assert that input DECLs have no ultimate
20868 (gen_type_die_with_usage): For typedef variants that have an
20869 ultimate origin, just call gen_decl_die on the original DECL.
20870 (process_scope_var): Avoid creating DIEs for local typedefs and
20871 concrete static variables.
20873 2017-08-12 Alan Modra <amodra@gmail.com>
20877 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20878 match gnu-user.h startfile.
20879 (ENDFILE_LINUX_SPEC): Similarly.
20881 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20884 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20886 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20888 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20889 * config/aarch64/aarch64.md (mov<mode>): Change.
20890 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20891 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20892 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20894 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20896 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20897 for storage order barriers.
20899 2017-08-11 Martin Liska <mliska@suse.cz>
20902 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20903 variables of void type.
20905 2017-08-11 Martin Liska <mliska@suse.cz>
20907 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20908 TARGET_SUPPORTS_ALIASES.
20909 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20910 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20911 (optimize_weakref): Likewise.
20912 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20913 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20914 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20916 2017-08-11 Martin Liska <mliska@suse.cz>
20919 * config/i386/i386.c (make_resolver_func): Do complete
20920 refactoring of the function.
20922 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20925 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20926 * config/i386/i386.c (ix86_stack_protect_guard): Use
20927 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20928 * doc/invoke.texi (x86 Options): Document
20929 -mstack-protector-guard-symbol= option.
20931 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20933 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20934 * config/i386/i386.c (ix86_split_stack_guard): New function.
20935 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20936 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20937 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20938 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20939 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20940 (split_stack_space_check): Call ix86_split_stack_guard.
20942 2017-08-10 Martin Sebor <msebor@redhat.com>
20944 * print-tree.c (print_node): Print location using the established
20946 Replace spaces with colons.
20947 (debug_raw, debug): Ditto.
20949 2017-08-10 Martin Sebor <msebor@redhat.com>
20952 * pretty-print.c (pp_format): Correct the handling of %s precision.
20954 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20957 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20959 (ix86_finalize_stack_frame_flags): This. Also clear
20960 frame_pointer_needed if -fno-omit-frame-pointer is used without
20962 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20963 with ix86_finalize_stack_frame_flags.
20964 (ix86_expand_epilogue): Likewise.
20965 (ix86_expand_split_stack_prologue): Likewise.
20966 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20968 2017-08-10 Martin Liska <mliska@suse.cz>
20971 * c-attribs.c (handle_target_attribute):
20972 Report warning for an empty string argument of target attribute.
20974 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20977 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20979 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20980 or DECL_NONLOCAL labels.
20981 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20982 or DECL_NONLOCAL labels here.
20984 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20986 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20987 to indicate when early gimple folding has been disabled.
20988 (rs6000_gimple_fold_builtin): Add debug content.
20989 (rs6000_invalid_builtin): Fix whitespace.
20990 (rs6000_expand_builtin): Fix whitespace.
20991 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20993 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20996 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20997 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20999 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21001 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21002 (qdf24xx): Likewise.
21003 * config/aarch64/aarch64.md: Include falkor.md.
21004 * config/aarch64/falkor.md: New.
21006 2017-08-09 Marek Polacek <polacek@redhat.com>
21009 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21010 * diagnostic.c (emit_diagnostic): Add a comment.
21011 (emit_diagnostic_valist): New function.
21013 2017-08-09 Marek Polacek <polacek@redhat.com>
21016 * input.c (make_location): New overload.
21017 * input.h (make_location): Declare.
21019 2017-08-08 Alan Modra <amodra@gmail.com>
21020 H.J. Lu <hongjiu.lu@intel.com>
21023 * gcc.c (NO_PIE_SPEC): Delete.
21024 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21026 (LINK_PIE_SPEC): ..to here.
21027 (LINK_COMMAND_SPEC): Support -no-pie.
21028 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21029 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21031 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21032 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21033 (ENDFILE_CRTEND_SPEC): Similarly.
21035 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21038 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21039 (mstack-protector-guard-offset=): Ditto.
21040 * config/i386/i386.c (ix86_option_override): Handle
21041 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21043 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21044 ix86_stack_protect_guard_offset variables.
21045 (TARGET_STACK_PROTECT_GUARD): Always define.
21046 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21047 and -mstack-protector-guard-offset= options.
21049 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21051 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21052 boundary case for the last candidate.
21054 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21056 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21057 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21059 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21061 PR middle-end/19706
21062 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21063 * config/aarch64/aarch64-builtins.c
21064 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21065 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21066 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21068 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21069 Andrew Pinski <pinskia@gmail.com>
21071 PR middle-end/19706
21072 * internal-fn.def (XORSIGN): New.
21073 * optabs.def (xorsign_optab): New.
21074 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21075 (convert_expand_mult_copysign): New.
21076 (pass_optimize_widening_mul::execute): Call
21077 convert_expand_mult_copysign.
21079 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21081 PR tree-optimization/81354
21082 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21083 Insert on edges rather than explicitly creating landing pads.
21084 (analyze_candidates_and_replace): Commit edge inserts.
21086 2017-08-08 Richard Biener <rguenther@suse.de>
21088 PR middle-end/81719
21089 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21090 (expand_simple_operations): Also look through ADDR_EXPRs with
21091 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21093 2017-08-08 Richard Biener <rguenther@suse.de>
21095 PR tree-optimization/81723
21096 * tree-vect-slp.c (struct bst_traits): New hash traits.
21097 (bst_fail): New global.
21098 (vect_build_slp_tree_2): New worker, split out from ...
21099 (vect_build_slp_tree): ... this now wrapping it with using
21100 bst_fail set to cache SLP tree build fails. Properly handle
21102 (vect_analyze_slp_instance): Allocate and free bst_fail.
21104 2017-08-08 Martin Liska <mliska@suse.cz>
21107 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21108 LABEL_DECLs that can be from a different function.
21110 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21112 PR tree-optimization/81744
21113 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21114 loop's number of iterations.
21116 2017-08-08 Martin Liska <mliska@suse.cz>
21118 * asan.c: Include header files.
21119 * attribs.c (build_decl_attribute_variant): New function moved
21121 (build_type_attribute_qual_variant): Likewise.
21122 (cmp_attrib_identifiers): Likewise.
21123 (simple_cst_list_equal): Likewise.
21124 (omp_declare_simd_clauses_equal): Likewise.
21125 (attribute_value_equal): Likewise.
21126 (comp_type_attributes): Likewise.
21127 (build_type_attribute_variant): Likewise.
21128 (lookup_ident_attribute): Likewise.
21129 (remove_attribute): Likewise.
21130 (merge_attributes): Likewise.
21131 (merge_type_attributes): Likewise.
21132 (merge_decl_attributes): Likewise.
21133 (merge_dllimport_decl_attributes): Likewise.
21134 (handle_dll_attribute): Likewise.
21135 (attribute_list_equal): Likewise.
21136 (attribute_list_contained): Likewise.
21137 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21138 (lookup_attribute_by_prefix): Likewise.
21139 * bb-reorder.c: Include header files.
21140 * builtins.c: Likewise.
21141 * calls.c: Likewise.
21142 * cfgexpand.c: Likewise.
21143 * cgraph.c: Likewise.
21144 * cgraphunit.c: Likewise.
21145 * convert.c: Likewise.
21146 * dwarf2out.c: Likewise.
21147 * final.c: Likewise.
21148 * fold-const.c: Likewise.
21149 * function.c: Likewise.
21150 * gimple-expr.c: Likewise.
21151 * gimple-fold.c: Likewise.
21152 * gimple-pretty-print.c: Likewise.
21153 * gimple.c: Likewise.
21154 * gimplify.c: Likewise.
21155 * hsa-common.c: Likewise.
21156 * hsa-gen.c: Likewise.
21157 * internal-fn.c: Likewise.
21158 * ipa-chkp.c: Likewise.
21159 * ipa-cp.c: Likewise.
21160 * ipa-devirt.c: Likewise.
21161 * ipa-fnsummary.c: Likewise.
21162 * ipa-inline.c: Likewise.
21163 * ipa-visibility.c: Likewise.
21165 * lto-cgraph.c: Likewise.
21166 * omp-expand.c: Likewise.
21167 * omp-general.c: Likewise.
21168 * omp-low.c: Likewise.
21169 * omp-offload.c: Likewise.
21170 * omp-simd-clone.c: Likewise.
21171 * opts-global.c: Likewise.
21172 * passes.c: Likewise.
21173 * predict.c: Likewise.
21174 * sancov.c: Likewise.
21175 * sanopt.c: Likewise.
21176 * symtab.c: Likewise.
21177 * toplev.c: Likewise.
21178 * trans-mem.c: Likewise.
21179 * tree-chkp.c: Likewise.
21180 * tree-eh.c: Likewise.
21181 * tree-into-ssa.c: Likewise.
21182 * tree-object-size.c: Likewise.
21183 * tree-parloops.c: Likewise.
21184 * tree-profile.c: Likewise.
21185 * tree-ssa-ccp.c: Likewise.
21186 * tree-ssa-live.c: Likewise.
21187 * tree-ssa-loop.c: Likewise.
21188 * tree-ssa-sccvn.c: Likewise.
21189 * tree-ssa-structalias.c: Likewise.
21190 * tree-ssa.c: Likewise.
21191 * tree-streamer-in.c: Likewise.
21192 * tree-vectorizer.c: Likewise.
21193 * tree-vrp.c: Likewise.
21194 * tsan.c: Likewise.
21195 * ubsan.c: Likewise.
21196 * varasm.c: Likewise.
21197 * varpool.c: Likewise.
21198 * tree.c: Remove functions moved to attribs.[ch].
21199 * tree.h: Likewise.
21200 * config/aarch64/aarch64.c: Add attrs.h header file.
21201 * config/alpha/alpha.c: Likewise.
21202 * config/arc/arc.c: Likewise.
21203 * config/arm/arm.c: Likewise.
21204 * config/avr/avr.c: Likewise.
21205 * config/bfin/bfin.c: Likewise.
21206 * config/c6x/c6x.c: Likewise.
21207 * config/cr16/cr16.c: Likewise.
21208 * config/cris/cris.c: Likewise.
21209 * config/darwin.c: Likewise.
21210 * config/epiphany/epiphany.c: Likewise.
21211 * config/fr30/fr30.c: Likewise.
21212 * config/frv/frv.c: Likewise.
21213 * config/ft32/ft32.c: Likewise.
21214 * config/h8300/h8300.c: Likewise.
21215 * config/i386/winnt.c: Likewise.
21216 * config/ia64/ia64.c: Likewise.
21217 * config/iq2000/iq2000.c: Likewise.
21218 * config/lm32/lm32.c: Likewise.
21219 * config/m32c/m32c.c: Likewise.
21220 * config/m32r/m32r.c: Likewise.
21221 * config/m68k/m68k.c: Likewise.
21222 * config/mcore/mcore.c: Likewise.
21223 * config/microblaze/microblaze.c: Likewise.
21224 * config/mips/mips.c: Likewise.
21225 * config/mmix/mmix.c: Likewise.
21226 * config/mn10300/mn10300.c: Likewise.
21227 * config/moxie/moxie.c: Likewise.
21228 * config/msp430/msp430.c: Likewise.
21229 * config/nds32/nds32-isr.c: Likewise.
21230 * config/nds32/nds32.c: Likewise.
21231 * config/nios2/nios2.c: Likewise.
21232 * config/nvptx/nvptx.c: Likewise.
21233 * config/pa/pa.c: Likewise.
21234 * config/pdp11/pdp11.c: Likewise.
21235 * config/powerpcspe/powerpcspe.c: Likewise.
21236 * config/riscv/riscv.c: Likewise.
21237 * config/rl78/rl78.c: Likewise.
21238 * config/rx/rx.c: Likewise.
21239 * config/s390/s390.c: Likewise.
21240 * config/sh/sh.c: Likewise.
21241 * config/sol2.c: Likewise.
21242 * config/sparc/sparc.c: Likewise.
21243 * config/spu/spu.c: Likewise.
21244 * config/stormy16/stormy16.c: Likewise.
21245 * config/tilegx/tilegx.c: Likewise.
21246 * config/tilepro/tilepro.c: Likewise.
21247 * config/v850/v850.c: Likewise.
21248 * config/vax/vax.c: Likewise.
21249 * config/visium/visium.c: Likewise.
21250 * config/xtensa/xtensa.c: Likewise.
21252 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21255 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21256 constraints since the -mupper-regs-* switches have been
21258 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21259 into a vector from a double word element that was extracted from
21260 another vector, and eliminate extra XXPERMDI instructions.
21261 (vsx_concat_<mode>_2): Likewise.
21262 (vsx_concat_<mode>_3): Likewise.
21263 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21264 concat to allow optimizing inserts from previous extracts.
21266 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21268 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21269 memory reference to a SSP offset in TLS address space.
21270 (ix86_print_operand) <case '@'>: Remove.
21271 (ix86_print_operand_punct_valid_p): Remove '@' code.
21272 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21273 UNSPEC_SP_TLS_TEST.
21274 (stack_tls_protect_set_<mode>): Remove.
21275 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21276 (stack_tls_protect_test_<mode>): Remove.
21277 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21279 2017-08-07 Olivier Hainque <hainque@adacore.com>
21282 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21284 2017-08-07 Douglas Rupp <rupp@adacore.com>
21286 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21287 variable was referenced as multidir in command.
21289 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21292 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21295 2017-08-07 Martin Liska <mliska@suse.cz>
21297 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21298 * config/rl78/rl78.c: Add include of attribs.h.
21299 * config/sh/sh.c: Likewise.
21300 * config/v850/v850.c: Likewise.
21302 2017-08-07 Tom de Vries <tom@codesourcery.com>
21304 PR middle-end/78266
21305 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21307 2017-08-07 Martin Liska <mliska@suse.cz>
21309 * config/mips/mips.c: Include attribs.h.
21311 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21314 * doc/invoke.texi: Document change in behvaior for -Ofast for
21317 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21319 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21321 (aarch64_pop_regs): Likewise.
21322 (aarch64_gen_load_pair): Likewise.
21323 (aarch64_save_callee_saves): Likewise.
21324 (aarch64_restore_callee_saves): Likewise.
21326 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21328 * config/i386/i386.c: Revert the last change.
21330 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21333 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21335 (ix86_finalize_stack_frame_flags): This. Also clear
21336 frame_pointer_needed if -fno-omit-frame-pointer is used without
21338 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21339 with ix86_finalize_stack_frame_flags.
21340 (ix86_expand_epilogue): Likewise.
21341 (ix86_expand_split_stack_prologue): Likewise.
21343 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21346 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21347 to P_AES for Westmere.
21349 2017-08-07 Jonathan Yong <10walls@gmail.com>
21351 * config/i386/mingw.opt (fset-stack-executable): Removed.
21352 * config/i386/cygming.opt (fset-stack-executable): Moved
21354 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21356 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21358 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21360 2017-08-07 Marek Polacek <polacek@redhat.com>
21362 PR middle-end/81737
21363 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21365 2017-08-07 Martin Liska <mliska@suse.cz>
21367 * attribs.h (canonicalize_attr_name): New function.
21368 (cmp_attribs): Move from c-format.c and adjusted.
21369 (is_attribute_p): Moved from tree.h.
21370 * tree-inline.c: Add new includes.
21371 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21372 (private_is_attribute_p): Remove.
21373 (private_lookup_attribute): Likewise.
21374 (private_lookup_attribute_by_prefix): Simplify.
21375 (remove_attribute): Use is_attribute_p.
21376 * tree.h: Remove removed declarations.
21378 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21380 PR middle-end/81698
21381 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21382 instead of computing it in the function. Formatting fix.
21383 (expand_case): Don't rely on default_edge being the first edge,
21384 clear it if removing it, pass default_edge to
21385 emit_case_dispatch_table.
21386 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21389 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21391 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21392 insn in the function, emit NOP after the insn.
21394 2017-08-06 Tom de Vries <tom@codesourcery.com>
21396 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21399 2017-08-06 Tom de Vries <tom@codesourcery.com>
21401 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21404 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21406 PR tree-optimization/57371
21407 * match.pd: New pattern.
21409 2017-08-04 Marek Polacek <polacek@redhat.com>
21411 PR middle-end/81695
21412 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21413 perform the computation in offset_int.
21415 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21417 PR tree-optimization/81136
21418 * tree-vectorizer.h: Include tree-hash-traits.h.
21419 (vec_base_alignments): New typedef.
21420 (vec_info): Add a base_alignments field.
21421 (vect_record_base_alignments): Declare.
21422 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21424 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21425 (create_data_ref): Add an is_conditional_in_stmt argument.
21426 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21427 the is_conditional_in_stmt field.
21428 (data_ref_loc): Add an is_conditional_in_stmt field.
21429 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21430 (find_data_references_in_stmt): Update call to create_data_ref.
21431 (graphite_find_data_references_in_stmt): Likewise.
21432 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21433 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21434 (vect_record_base_alignment): New function.
21435 (vect_record_base_alignments): Likewise.
21436 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21437 for nested statements even if we fail to compute a misalignment.
21438 Use pooled base alignments for unconditional references.
21439 (vect_find_same_alignment_drs): Compare base addresses instead
21441 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21442 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21444 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21446 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21447 Add an explicit name for the enum. Use auto_vec for slp_instances
21448 and grouped_stores.
21449 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21451 (_bb_vec_info): Add a constructor and destructor.
21452 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21453 (destroy_loop_vec_info): Delete.
21454 (vect_destroy_datarefs): Likewise.
21455 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21456 (vec_info::vec_info): New function.
21457 (vec_info::~vec_info): Likewise.
21458 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21459 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21460 destroy_loop_vec_info.
21461 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21462 (_loop_vec_info::_loop_vec_info): ...this.
21463 (destroy_loop_vec_info): Replace with...
21464 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21465 the stmt_vec_infos. Leave handling of vec_info information to its
21466 destructor. Remove explicit vector releases.
21467 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21468 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21469 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21470 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21471 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21472 (destroy_bb_vec_info): Replace with...
21473 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21474 information to its destructor.
21475 (vect_slp_analyze_bb_1): Use new and delete instead of
21476 new_bb_vec_info and destroy_bb_vec_info.
21477 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21480 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21482 * tree-data-ref.h (subscript): Add access_fn field.
21483 (data_dependence_relation): Add could_be_independent_p.
21484 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21485 (same_access_functions): Move to tree-data-ref.c.
21486 * tree-data-ref.c (ref_contains_union_access_p): New function.
21487 (access_fn_component_p): Likewise.
21488 (access_fn_components_comparable_p): Likewise.
21489 (dr_analyze_indices): Add a reference to access_fn_component_p.
21490 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21492 (constant_access_functions): Likewise.
21493 (add_other_self_distances): Likewise.
21494 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21495 (initialize_data_dependence_relation): Use XCNEW and remove
21496 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21497 of access functions that have the same type. Allow the
21498 subsequence to end with different bases in some circumstances.
21499 Record the chosen access functions in SUB_ACCESS_FN.
21500 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21501 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21502 (subscript_dependence_tester_1): Likewise dra and drb.
21503 (build_classic_dist_vector): Update calls accordingly.
21504 (subscript_dependence_tester): Likewise.
21505 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21506 DDR_COULD_BE_INDEPENDENT_P.
21507 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21508 comp_alias_ddrs instead of may_alias_ddrs.
21509 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21511 (vect_analyze_data_ref_dependence): Use it if
21512 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21513 distance vectors if that fails.
21514 (dependence_distance_ge_vf): New function.
21515 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21516 LOOP_VINFO_MAY_ALIAS_DDRS.
21518 2017-08-04 Richard Biener <rguenther@suse.de>
21520 PR middle-end/81705
21521 * fold-const.c (fold_binary_loc): Properly restrict
21522 minus_var0 && minus_var1 case when associating undefined overflow
21525 2017-08-04 Tom de Vries <tom@codesourcery.com>
21527 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21529 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21531 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21532 Don't start diagnostic messages with a capital letter.
21533 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21535 (rs6000_invalid_builtin): Likewise.
21536 (rs6000_trampoline_init): Likewise.
21538 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21541 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21542 after setting changeable df flags.
21544 2017-08-03 Richard Biener <rguenther@suse.de>
21546 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21547 up if the use is in USE - X.
21549 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21551 * toplev.c (dumpfile.h): New include.
21552 (internal_error_reentered): New static function. Use it...
21553 (internal_error_function): ...here to handle reentered internal_error.
21555 2017-08-03 Richard Biener <rguenther@suse.de>
21557 PR middle-end/81148
21558 * fold-const.c (split_tree): Add minus_var and minus_con
21559 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21560 here but always use minus_*.
21561 (associate_trees): Assert we never associate with MINUS_EXPR
21562 and NULL first operand. Do not recurse for PLUS_EXPR operands
21563 when associating as MINUS_EXPR either.
21564 (fold_binary_loc): Track minus_var and minus_con.
21566 2017-08-03 Tom de Vries <tom@codesourcery.com>
21569 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21570 ACCEL_COMPILER, apply finish_options on
21571 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21573 2017-08-03 Tom de Vries <tom@codesourcery.com>
21576 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21577 function_entry_patch_area_size > 0.
21579 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21582 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21583 instead of 10??LU, perform unit multiplication in wide_int,
21584 don't change alloc_object_size_limit if the limit is larger
21587 PR tree-optimization/81655
21588 PR tree-optimization/81588
21589 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21590 the case when ranges[i].low and high are 1 for unsigned type with
21593 PR middle-end/81052
21594 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21595 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21597 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21599 * tree-vrp.h: Add include guard.
21601 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21604 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21605 (ud2): New insn pattern.
21606 * config/i386/i386.c (ix86_expand_epilogue):
21607 For naked functions, generate ud2 instead of trap insn.
21609 2017-08-02 Marek Polacek <polacek@redhat.com>
21612 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21614 2017-08-02 Tom de Vries <tom@codesourcery.com>
21615 Cesar Philippidis <cesar@codesourcery.com>
21617 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21618 Add missing edge probabilities.
21620 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21622 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21623 Correct endianness.
21625 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21627 PR middle-end/79499
21628 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21629 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21630 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21631 prologue_seq sequences - if any.
21633 2017-08-02 Richard Biener <rguenther@suse.de>
21635 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21636 via vectors if supported, integer extracts via punning if supported
21637 or otherwise vector extracts.
21639 2017-08-02 Richard Biener <rguenther@suse.de>
21641 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21643 (bitmap_insert_into_set): ... this.
21645 2017-08-02 Richard Biener <rguenther@suse.de>
21647 PR tree-optimization/81633
21649 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21651 PR tree-optimization/71752
21652 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21654 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21656 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21657 (machine_function::call_ms2sysv_pad_out): Remove field.
21658 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21659 (ix86_compute_frame_layout): Likewise.
21661 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21664 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21665 attribute with interrupt attribute.
21667 2017-08-01 Andrew Pinski <apinski@cavium.com>
21669 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21670 BIT_INSERT_EXPR's operand 1
21671 to see if the types precision matches.
21673 2017-08-01 Martin Liska <mliska@suse.cz>
21675 PR middle-end/70140
21676 * builtins.c (expand_builtin_memcpy_args): Remove.
21677 (expand_builtin_memcpy): Call newly added function
21678 expand_builtin_memory_copy_args.
21679 (expand_builtin_memcpy_with_bounds): Likewise.
21680 (expand_builtin_mempcpy): Remove last argument.
21681 (expand_builtin_mempcpy_with_bounds): Likewise.
21682 (expand_builtin_memory_copy_args): New function created from
21683 expand_builtin_mempcpy_args with small modifications.
21684 (expand_builtin_mempcpy_args): Remove.
21685 (expand_builtin_stpcpy): Remove unused argument.
21686 (expand_builtin): Likewise.
21687 (expand_builtin_with_bounds): Likewise.
21689 2017-08-01 Martin Liska <mliska@suse.cz>
21692 Make mempcpy more optimal (PR middle-end/70140).
21694 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21697 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21698 __builtin_vec_cmpne verify both arguments are compatible vectors
21699 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21700 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21701 move computation of aligned to after checking the argument types.
21705 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21706 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21709 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21710 Doug Rupp <rupp@adacore.com>
21711 Olivier Hainque <hainque@adacore.com>
21713 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21714 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21716 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21717 for TARGET_OS_CPP_BUILTIN.
21718 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21719 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21721 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21722 passing required abi options to the assembler for EABI configurations.
21723 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21724 of .text.hot and .text.unlikely sections for kernel modules when
21725 using ARM style exceptions.
21726 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21727 options. Add EXTRA_CC1_SPEC.
21728 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21730 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21732 (ARM_TARGET2_DWARF_FORMAT): Define.
21733 * config/arm/t-vxworks: Adjust multilib control to removal of the
21734 Diab command line options.
21736 2017-08-01 Martin Liska <mliska@suse.cz>
21738 PR gcov-profile/81561
21739 * gcov.c (unblock): Make unblocking safe as we need to preserve
21740 index correspondence of blocks and block_lists.
21742 2017-08-01 Richard Biener <rguenther@suse.de>
21744 PR tree-optimization/81181
21745 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21746 (compute_antic): ... end of iteration here.
21748 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21750 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21751 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21752 (ftree-slp-vectorize): Likewise.
21753 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21754 can no longer be set independent of flag_tree_loop_vectorize.
21755 * omp-general.c (emp_max_vf): Likewise.
21756 * opts.c (enable_fdo_optimizations): Remove references to
21757 flag_tree_vectorize, these are now implicit.
21758 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21759 and leave it for the options machinery.
21761 2017-08-01 Martin Liska <mliska@suse.cz>
21763 PR middle-end/70140
21764 * builtins.c (expand_builtin_memcpy_args): Remove.
21765 (expand_builtin_memcpy): Call newly added function
21766 expand_builtin_memory_copy_args.
21767 (expand_builtin_memcpy_with_bounds): Likewise.
21768 (expand_builtin_mempcpy): Remove last argument.
21769 (expand_builtin_mempcpy_with_bounds): Likewise.
21770 (expand_builtin_memory_copy_args): New function created from
21771 expand_builtin_mempcpy_args with small modifications.
21772 (expand_builtin_mempcpy_args): Remove.
21773 (expand_builtin_stpcpy): Remove unused argument.
21774 (expand_builtin): Likewise.
21775 (expand_builtin_with_bounds): Likewise.
21777 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21780 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21781 print "ds:" only for immediates in generic address space.
21783 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21786 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21787 (ix86_function_ok_for_sibcall): Return false for naked functions.
21789 2017-08-01 Richard Biener <rguenther@suse.de>
21791 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21792 (compute_antic): Seed worklist with exit block predecessors.
21793 * cfganal.c (dfs_find_deadend): For a cycle return the source
21794 of the edge closing it.
21796 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21798 * config/aarch64/aarch64.c
21799 (aarch64_can_const_movi_rtx_p): Move 0 check.
21801 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21803 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21804 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21806 * match.pd: Ditto in address comparison pattern.
21808 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21810 PR tree-optimization/81627
21811 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21812 closed ssa form for store-store chain.
21814 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21816 PR tree-optimization/81620
21817 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21818 for store-store chain.
21820 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21822 PR tree-optimization/81588
21823 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21824 ranges[i].in_p, invert comparison code ccode. For >/>=,
21825 swap rhs1 and rhs2 and comparison code unconditionally,
21826 for </<= don't do that. Don't swap rhs1/rhs2 again if
21827 ranges[i].in_p, instead invert comparison code ccode if
21828 opcode or oe->rank is BIT_IOR_EXPR.
21831 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21832 a direct optab to conversion optab.
21833 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21834 with GET_MODE_INNER as last argument instead of optab_handler.
21835 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21836 vector extraction if possible and optab is available.
21837 * expr.c (store_constructor): Use convert_optab_handler instead
21838 of optab_handler. Use vector initialization from smaller
21839 vectors if possible and optab is available.
21840 * tree-vect-stmts.c (vectorizable_load): Likewise.
21841 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21842 now have two modes.
21843 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21844 of vec_init from half-sized vectors with the same element mode.
21845 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21846 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21847 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21848 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21849 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21850 after mode in gen_vec_extract* calls.
21851 (vec_extract<mode>): Renamed to ...
21852 (vec_extract<mode><ssescalarmodelower>): ... this.
21853 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21854 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21855 element mode after mode in gen_vec_init* calls.
21856 (VEC_INIT_HALF_MODE): New mode iterator.
21857 (vec_init<mode>): Renamed to ...
21858 (vec_init<mode><ssescalarmodelower>): ... this.
21859 (vec_init<mode><ssehalfvecmodelower>): New expander.
21860 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21861 (vec_extractv2sfsf): ... this.
21862 (vec_initv2sf): Renamed to ...
21863 (vec_initv2sfsf): ... this.
21864 (vec_extractv2si): Renamed to ...
21865 (vec_extractv2sisi): ... this.
21866 (vec_initv2si): Renamed to ...
21867 (vec_initv2sisi): ... this.
21868 (vec_extractv4hi): Renamed to ...
21869 (vec_extractv4hihi): ... this.
21870 (vec_initv4hi): Renamed to ...
21871 (vec_initv4hihi): ... this.
21872 (vec_extractv8qi): Renamed to ...
21873 (vec_extractv8qiqi): ... this.
21874 (vec_initv8qi): Renamed to ...
21875 (vec_initv8qiqi): ... this.
21876 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21877 (vec_init<mode>): Renamed to ...
21878 (vec_init<mode><VEC_base_l>): ... this.
21879 (vec_extract<mode>): Renamed to ...
21880 (vec_extract<mode><VEC_base_l>): ... this.
21881 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21882 (vec_initv2sfsf): ... this.
21883 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21884 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21885 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21886 element mode after mode in gen_vec_init* calls.
21887 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21888 (vec_init<mode><Vel>): ... this.
21889 (vec_extract<mode>): Renamed to ...
21890 (vec_extract<mode><Vel>): ... this.
21891 * config/aarch64/iterators.md (Vel): New mode attribute.
21892 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21893 Add element mode after mode in gen_vec_extract* calls.
21894 * config/s390/vector.md (non_vec_l): New mode attribute.
21895 (vec_extract<mode>): Renamed to ...
21896 (vec_extract<mode><non_vec_l>): ... this.
21897 (vec_init<mode>): Renamed to ...
21898 (vec_init<mode><non_vec_l>): ... this.
21899 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21900 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21902 * config/arm/iterators.md (V_elem_l): New mode attribute.
21903 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21904 (vec_extract<mode><V_elem_l>): ... this.
21905 (vec_extractv2di): Renamed to ...
21906 (vec_extractv2didi): ... this.
21907 (vec_init<mode>): Renamed to ...
21908 (vec_init<mode><V_elem_l>): ... this.
21909 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21910 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21911 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21912 Add element mode after gen_vec_extract* calls.
21913 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21914 (vec_init<mode><unitmode>): ... this.
21915 (vec_extract<mode>): Renamed to ...
21916 (vec_extract<mode><unitmode>): ... this.
21917 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21918 (vec_init<mode><unitmode>): ... this.
21919 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21920 (vec_initv2sfsf): ... this.
21921 (vec_extractv2sf): Renamed to ...
21922 (vec_extractv2sfsf): ... this.
21923 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21924 Add element mode after gen_vec_extract* calls.
21925 * config/mips/mips.md (unitmode): New mode iterator.
21926 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21927 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21928 * config/spu/spu.md (inner_l): New mode attribute.
21929 (vec_init<mode>): Renamed to ...
21930 (vec_init<mode><inner_l>): ... this.
21931 (vec_extract<mode>): Renamed to ...
21932 (vec_extract<mode><inner_l>): ... this.
21933 * config/sparc/sparc.md (veltmode): New mode iterator.
21934 (vec_init<VMALL:mode>): Renamed to ...
21935 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21936 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21937 (vec_initv2sisi): ... this.
21938 (vec_initv2sf): Renamed to ...
21939 (vec_initv2sfsf): ... this.
21940 (vec_extractv2sf): Renamed to ...
21941 (vec_extractv2sfsf): ... this.
21942 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21943 (vec_init<mode>): Renamed to ...
21944 (vec_init<mode><VEC_base_l>): ... this.
21945 (vec_extract<mode>): Renamed to ...
21946 (vec_extract<mode><VEC_base_l>): ... this.
21947 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21948 (vec_initv2sfsf): ... this.
21949 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21950 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21951 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21952 gen_vec_init* calls.
21954 2017-08-01 Richard Biener <rguenther@suse.de>
21956 PR tree-optimization/81297
21957 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21958 TREE_OVERFLOW from INTEGER_CSTs.
21960 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21962 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21964 2017-07-31 Carl Love <cel@us.ibm.com>
21966 * config/rs6000/rs6000-c: Add support for built-in functions
21967 vector signed char vec_xl_be (signed long long, signed char *);
21968 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21969 vector signed int vec_xl_be (signed long long, signed int *);
21970 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21971 vector signed long long vec_xl_be (signed long long, signed long long *);
21972 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21973 vector signed short vec_xl_be (signed long long, signed short *);
21974 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21975 vector double vec_xl_be (signed long long, double *);
21976 vector float vec_xl_be (signed long long, float *);
21977 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21978 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21979 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21981 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21982 (altivec_expand_builtin): Add switch statement to call
21983 altivec_expand_xl_be for each builtin.
21984 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21985 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21986 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21987 __builtin_vsx_le_be_v16qi.
21988 * doc/extend.texi: Update the built-in documentation file for the
21989 new built-in functions.
21991 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21994 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21996 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21998 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22000 * config.gcc: Add z14.
22001 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22002 CPU model numbers for z13s and z14.
22003 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22005 * config/s390/s390-opts.h (enum processor_type): Rename
22006 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22007 * config/s390/s390.c (processor_table): Add field for CPU name to
22008 be passed to Binutils.
22009 (s390_asm_output_machine_for_arch): Use the new field in
22010 processor_table for Binutils.
22011 (s390_expand_builtin): Replace arch12 with z14.
22012 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22013 (s390_get_sched_attrmask): Likewise.
22014 (s390_get_unit_mask): Likewise.
22015 * config/s390/s390.opt: Add z14 to processor_type enum.
22017 2017-07-31 Martin Jambor <mjambor@suse.cz>
22020 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22021 regardless of optimization level.
22023 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22024 Martin Liska <mliska@suse.cz>
22026 * predict.def: Remove old comment and adjust probability.
22027 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22028 PREDICT statements.
22030 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22033 * config/i386/i386.c (ix86_function_naked): New function.
22034 (ix86_can_use_return_insn_p): Return false for naked functions.
22035 (ix86_expand_prologue): Skip prologue for naked functions.
22036 (ix86_expand_epilogue): Skip epilogue for naked functions
22037 and emit trap instruction.
22038 (ix86_warn_func_return): New function.
22039 (ix86_attribute_table): Add "naked" attribute specification.
22040 (TARGET_WARN_FUNC_RETURN): Define.
22041 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22043 2017-07-31 Martin Liska <mliska@suse.cz>
22045 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22046 (dump_gimple_bb_header): Always dump BB info.
22047 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22049 2017-07-31 Martin Liska <mliska@suse.cz>
22052 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22053 also with current_function_decl non-null equality.
22055 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22058 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22059 change type to the element type, instead add eltype variable and
22060 use it where we are interested in the element type.
22062 PR tree-optimization/81603
22063 * ipa-polymorphic-call.c
22064 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22065 offset arithmetic in offset_int, bail out if the resulting bit offset
22066 doesn't fit into shwi.
22068 2017-07-31 Martin Liska <mliska@suse.cz>
22070 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22071 (gimplify_save_expr): Fix comment.
22073 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22076 * config/i386/i386.c (ix86_function_arg): Update arguments for
22078 (ix86_compute_frame_layout): Set the initial stack offset to
22079 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22080 INCOMING_FRAME_SP_OFFSET.
22081 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22082 stack before exception handler returns.
22083 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22084 the 'ERROR_CODE' for exception handler.
22086 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22088 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22089 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22090 (ASM_OUTPUT_REG_POP): Ditto.
22091 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22092 instead of asm_fprintf to output pure string.
22094 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22096 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22097 to imported_module_or_decl hook.
22098 (debug_nothing_tree_tree_tree_bool): Remove.
22099 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22100 * debug.c (do_nothing_debug_hooks): Use
22101 debug_nothing_tree_tree_tree_bool_bool instead of
22102 debug_nothing_tree_tree_tree_bool.
22103 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22104 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22105 * sdbout.c (sdb_debug_hooks): Likewise.
22106 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22107 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22109 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22110 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22111 attribute, don't add anything.
22113 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22115 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22116 (fold_build2_stat_loc): Likewise.
22117 (fold_build3_stat_loc): Likewise.
22118 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22119 (fold_build1_loc): Remove macro.
22120 (fold_build2_loc): Likewise.
22121 (fold_build3_loc): Likewise.
22123 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22125 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22126 (gimple_build_debug_bind_source_stat): Likewise.
22127 * gimple.h (gimple_build_debug_bind): Remove macro.
22128 (gimple_build_debug_bind_source): Likewise.
22130 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22132 * bitmap.c (bitmap_alloc): Adjust.
22133 (bitmap_gc_alloc): Likewise.
22134 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22136 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22138 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22139 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22140 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22141 (bitmap_gc_alloc_stat): Likewise.
22142 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22144 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22146 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22147 * rtl.h (shallow_copy_rtx): Remove macro.
22149 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22151 * emit-rtl.c (gen_raw_REG): Adjust.
22152 * gengenrtl.c (gendef): Likewise.
22153 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22154 * rtl.h (rtx_alloc): Remove macro.
22156 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22158 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22159 (build_tree_list_stat): Likewise.
22160 * tree.h (build_tree_list): Remove macro.
22161 (build_tree_list_vec): Likewise.
22163 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22165 * tree.c (make_vector_stat): Remove _stat from name.
22166 (build_vector_stat): Likewise.
22167 * tree.h (make_vector_stat): Remove macro.
22168 (build_vector_stat): Likewise.
22170 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22172 * tree.h (build_var_debug_value): Remove prototype.
22174 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22176 * tree.c (tree_cons_stat): Remove _stat from name.
22177 * tree.h (tree_cons): Remove macro.
22179 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22181 * tree.c (build_vl_exp_stat): Remove _stat from name.
22182 * tree.h (build_vl_exp): Remove macro.
22184 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22186 * tree.c (build_decl_stat): Remove _stat from name.
22187 * tree.h (build_decl): Remove macro.
22189 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22191 * gimple.c (gimple_build_with_ops_stat): Adjust.
22192 (gimple_alloc_stat): Remove _stat from name.
22193 * gimple.h (gimple_alloc): Remove macro.
22195 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22197 * tree.c (make_tree_vec_stat): Remove _stat from name.
22198 (grow_tree_vec_stat): Likewise.
22199 * tree.h (make_tree_vec_stat): Adjust prototype.
22200 (grow_tree_vec_stat): Likewise.
22201 (make_tree_vec): Remove macro.
22202 (grow_tree_vec): Likewise.
22204 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22206 * fold-const.c (fold_build1_stat_loc): Adjust.
22207 (fold_build2_stat_loc): Likewise.
22208 (fold_build3_stat_loc): Likewise.
22209 * tree.c (build0_stat): Remove _stat from name.
22210 (build1_stat): Likewise.
22211 (build2_stat): Likewise.
22212 (build3_stat): Likewise.
22213 (build4_stat): Likewise.
22214 (build5_stat): Likewise.
22215 * tree.h (build1_loc): Remove macro, and rename _stat function
22217 (build2_loc): Likewise.
22218 (build3_loc): Likewise.
22219 (build4_loc): Likewise.
22220 (build5_loc): Likewise.
22222 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22224 * tree.c (make_int_cst_stat): Remove _stat from name.
22225 * tree.h (make_int_cst_stat): Adjust prototype.
22226 (make_int_cst): Remove macro.
22228 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22230 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22231 * tree.h (make_tree_binfo_stat): Adjust prototype.
22232 (make_tree_binfo): Remove.
22234 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22236 * tree.c (copy_node_stat): Rename to copy_node.
22237 (build_distinct_type_copy): Adjust.
22238 * tree.h (copy_node_stat): Adjust prototype.
22239 (copy_node): Remove macro.
22241 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22243 * tree.c (make_node_stat): rename to make_node.
22244 (build_tree_list_stat): Adjust.
22245 (build0_stat): Likewise.
22246 (build2_stat): Likewise.
22247 (build3_stat): Likewise.
22248 (build4_stat): Likewise.
22249 (build5_stat): Likewise.
22250 (build_decl_stat): Likewise.
22251 * tree.h (make_node_stat): Adjust prototype.
22252 (make_node): remove macro.
22254 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22256 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22257 (PPC_FEATURE2_SCV): Likewise.
22258 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22260 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22262 * config/aarch64/aarch64.c
22263 (aarch64_internal_mov_immediate): Add new special pattern.
22264 * config/aarch64/aarch64.md (*movdi_aarch64):
22265 Add reg/32bit const mov case.
22267 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22268 Richard Sandiford <richard.sandiford@linaro.org>
22270 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22271 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22272 Add integer and movi cases.
22273 (movi-split-hf-df-sf split, fp16): New.
22274 (enabled): Added TARGET_FP_F16INST.
22275 * config/aarch64/iterators.md (GPF_HF): New.
22276 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22278 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22280 * config/aarch64/aarch64.c
22281 (aarch64_simd_container_mode): Add prototype.
22282 (aarch64_expand_mov_immediate): Add HI support.
22283 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22284 (aarch64_can_const_movi_rtx_p): New.
22285 (aarch64_preferred_reload_class):
22286 Remove restrictions of using FP registers for certain SIMD operations.
22287 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22288 (aarch64_valid_floating_const): Add integer move validation.
22289 (aarch64_simd_imm_scalar_p): Remove.
22290 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22291 (aarch64_legitimate_constant_p): Expand list of supported cases.
22292 * config/aarch64/aarch64-protos.h
22293 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22294 (aarch64_reinterpret_float_as_int): New.
22295 (aarch64_simd_imm_scalar_p): Remove.
22296 * config/aarch64/constraints.md (Uvi): New.
22297 (Dd): Split into Ds and new Dd.
22298 * config/aarch64/aarch64.md (*movsi_aarch64):
22300 (*movdi_aarch64): Add SIMD mov case.
22302 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22304 * tree-predcom.c: (struct chain): Handle store-store chain in which
22305 stores for elimination only store loop invariant values.
22306 (execute_pred_commoning_chain): Ditto.
22307 (prepare_initializers_chain_store_elim): Ditto.
22308 (prepare_finalizers): Ditto.
22309 (is_inv_store_elimination_chain): New function.
22310 (initialize_root_vars_store_elim_1): New function.
22312 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22314 * tree-predcom.c: Revise general description of the pass.
22315 (enum chain_type): New enum type for store elimination.
22316 (struct chain): New field supporting store elimination.
22317 (struct component): Ditto.
22318 (dump_chain): Dump store-stores chain.
22319 (release_chain): Release resources.
22320 (split_data_refs_to_components): Compute and create component
22321 contains only stores for elimination.
22322 (get_chain_last_ref_at): New function.
22323 (make_invariant_chain): Initialization.
22324 (make_rooted_chain): Specify chain type in parameter and record it.
22325 (add_looparound_copies): Skip for store-stores chain.
22326 (determine_roots_comp): Compute type of chain and pass it to
22328 (initialize_root_vars_store_elim_2): New function.
22329 (finalize_eliminated_stores): New function.
22330 (remove_stmt): Handle store for elimination.
22331 (execute_pred_commoning_chain): Execute predictive commoning on
22332 store-store chains.
22333 (determine_unroll_factor): Skip unroll for store-stores chain.
22334 (prepare_initializers_chain_store_elim): New function.
22335 (prepare_initializers_chain): Hanlde store-store chain.
22336 (prepare_finalizers_chain, prepare_finalizers): New function.
22337 (tree_predictive_commoning_loop): Return integer value indicating
22338 if loop is unrolled or lcssa form is corrupted.
22339 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22341 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22343 * tree-predcom.c (initialize_root): Delete.
22344 (execute_pred_commoning_chain): Initialize root vars and replace
22345 reference of non-combined chain directly, rather than call above
22348 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22350 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22351 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22353 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22355 * tree-predcom.c (struct chain): New field init_seq.
22356 (release_chain): Release init_seq.
22357 (prepare_initializers_chain): Record intialization stmts in above
22359 (insert_init_seqs): New function.
22360 (tree_predictive_commoning_loop): Call insert_init_seqs.
22362 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22364 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22366 2017-07-28 Richard Biener <rguenther@suse.de>
22368 * match.pd: Remove superfluous :c.
22369 * genmatch.c (simplify::id): Add member.
22370 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22372 (current_id): New global.
22373 (dt_node::parent): Move from ...
22374 (dt_operand::parent): ... here. Add for_id member.
22375 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22376 (decision_tree::find_node): Relax order requirement when
22377 merging DT_TRUE nodes to ones inbetween the current simplify
22378 and the one we try to merge with. Add diagnostic whenever
22379 we need to enforce pattern order by not merging.
22380 (decision_tree::insert): Set current_id.
22381 (decision_tree::print_node): Dump parent node and for_id.
22382 (parser::last_id): Add member.
22383 (parser::push_simplify): Assign unique id.
22384 (parser::parser): Initialize last_id.
22386 2017-07-28 Martin Liska <mliska@suse.cz>
22389 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22390 gimple_build_debug_bind.
22392 2017-07-28 Richard Biener <rguenther@suse.de>
22394 PR tree-optimization/81502
22395 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22397 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22399 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22400 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22401 for BIT_FIELD_REF args.
22402 * fold-const.c (make_bit_field_ref): Likewise.
22403 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22405 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22408 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22409 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22410 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22411 Or it into SANITIZER_UNDEFINED.
22412 * ubsan.c: Include gimple-fold.h and varasm.h.
22413 (ubsan_expand_ptr_ifn): New function.
22414 (instrument_pointer_overflow): New function.
22415 (maybe_instrument_pointer_overflow): New function.
22416 (instrument_object_size): Formatting fix.
22417 (pass_ubsan::execute): Call instrument_pointer_overflow
22418 and maybe_instrument_pointer_overflow.
22419 * internal-fn.c (expand_UBSAN_PTR): New function.
22420 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22421 * sanitizer.def (__ubsan_handle_pointer_overflow,
22422 __ubsan_handle_pointer_overflow_abort): New builtins.
22423 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22424 * internal-fn.def (UBSAN_PTR): New internal function.
22425 * opts.c (sanitizer_opts): Add pointer-overflow.
22426 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22427 * fold-const.c (build_range_check): Compute pointer range check in
22428 integral type if pointer arithmetics would be needed. Formatting
22431 2017-07-28 Martin Liska <mliska@suse.cz>
22434 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22435 parameters that are of a variable-length.
22437 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22439 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22441 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22442 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22443 (CRT_CALL_STATIC_FUNCTION): Likewise.
22444 (ASM_DEFAULT_SPEC): New define.
22445 (ASM_SPEC32): Likewise.
22446 (ASM_SPEC64): Likewise.
22447 (ASM_SPEC_COMMON): Likewise.
22448 (ASM_SPEC): Likewise.
22449 (INVALID_64BIT): Likewise.
22450 (LINK_OS_DEFAULT_SPEC): Likewise.
22451 (LINK_OS_SPEC32): Likewise.
22452 (LINK_OS_SPEC64): Likewise.
22453 (POWERPC_LINUX): Likewise.
22454 (PTRDIFF_TYPE): Likewise.
22455 (RESTORE_FP_PREFIX): Likewise.
22456 (RESTORE_FP_SUFFIX): Likewise.
22457 (SAVE_FP_PREFIX): Likewise.
22458 (SAVE_FP_SUFFIX): Likewise.
22459 (SIZE_TYPE): Likewise.
22460 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22461 (TARGET_64BIT): Likewise.
22462 (TARGET_64BIT): Likewise.
22463 (TARGET_AIX): Likewise.
22464 (WCHAR_TYPE_SIZE): Likewise.
22465 (WCHAR_TYPE): Undefine.
22466 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22467 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22468 (CPP_OS_RTEMS_SPEC): Delete.
22469 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22470 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22472 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22474 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22476 PR tree-optimization/81578
22477 * tree-parloops.c (build_new_reduction): Bail out if
22478 reduction_code isn't one of the standard OpenMP reductions.
22479 Move the details printing after that decision.
22481 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22483 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22484 related to reload_in_progress.
22485 (splat_input_operand): Likewise.
22486 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22488 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22490 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22491 (TARGET_INSTANTIATE_DECLS): Likewise.
22492 (legitimate_indexed_address_p): Delete reload_in_progress code.
22493 (rs6000_debug_legitimate_address_p): Likewise.
22494 (rs6000_eliminate_indexed_memrefs): Likewise.
22495 (rs6000_emit_le_vsx_store): Likewise.
22496 (rs6000_emit_move_si_sf_subreg): Likewise.
22497 (rs6000_emit_move): Likewise.
22498 (register_to_reg_type): Likewise.
22499 (rs6000_pre_atomic_barrier): Likewise.
22500 (rs6000_machopic_legitimize_pic_address): Likewise.
22501 (rs6000_allocate_stack_temp): Likewise.
22502 (rs6000_address_for_fpconvert): Likewise.
22503 (rs6000_address_for_altivec): Likewise.
22504 (rs6000_secondary_memory_needed_rtx): Delete function.
22505 (rs6000_check_sdmode): Likewise.
22506 (rs6000_alloc_sdmode_stack_slot): Likewise.
22507 (rs6000_instantiate_decls): Likewise.
22508 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22509 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22510 Delete reload_in_progress.
22511 (*vec_reload_and_plus_<mptrsize>): Likewise.
22512 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22513 (vsx_div_v2di): Likewise.
22514 (vsx_udiv_v2di): Likewise.
22516 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22518 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22519 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22520 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22521 (rs6000_debug_reg_global): Delete print of LRA status.
22522 (rs6000_option_override_internal): Delete dead LRA related code.
22523 (rs6000_lra_p): Delete function.
22524 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22526 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22528 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22529 * config/riscv/rtems.h: New file.
22531 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22532 Sudakshina Das <sudi.das@arm.com>
22534 * config/aarch64/aarch64.md
22535 (define_split for and<mode>3nr_compare): Move
22536 non aarch64_logical_operand to a register.
22537 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22538 register immediate operand to a register.
22539 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22541 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22543 PR middle-end/81564
22544 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22546 2017-07-27 Richard Biener <rguenther@suse.de>
22548 PR tree-optimization/81573
22549 PR tree-optimization/81494
22550 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22551 multi defuse cycle case.
22553 2017-07-27 Richard Biener <rguenther@suse.de>
22555 PR tree-optimization/81571
22556 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22559 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22561 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22562 earlier and only if MASK_FPU is set. Adjust formatting.
22564 2017-07-27 Martin Liska <mliska@suse.cz>
22566 * opt-functions.awk: Add validation of value of Init.
22567 * optc-gen.awk: Pass new argument.
22569 2017-07-27 Martin Liska <mliska@suse.cz>
22571 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22572 Fix wrong condition.
22574 2017-07-27 Martin Liska <mliska@suse.cz>
22576 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22577 BBs and edges seen by autoFDO.
22579 2017-07-27 Richard Biener <rguenther@suse.de>
22581 PR tree-optimization/81502
22582 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22583 with incompatible but same sized type.
22584 (execute_update_addresses_taken): Likewise.
22586 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22588 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22589 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22591 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22594 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22595 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22596 Change s_operand to memory_operand.
22598 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22600 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22601 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22602 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22603 Emit instructions rather than returning an expression. Handle TFmode
22604 and KFmode by casting to TImode.
22605 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22606 (rs6000_emit_le_vsx_store): Likewise.
22607 * config/rs6000/vsx.md (VSX_TI): New iterator.
22608 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22609 (*vsx_le_undo_permute_<mode>): Likewise.
22610 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22611 emit the split sequence.
22612 (*vsx_le_perm_store_<mode>): Likewise.
22614 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22616 PR tree-optimization/81555
22617 PR tree-optimization/81556
22618 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22619 if true, force CHANGED for the recursive invocation.
22620 (reassociate_bb): Remember original length of ops array, pass
22621 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22623 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22624 attributes for noipa attribute. For naked attribute use
22625 lookup_attribute first before lookup_attribute_spec.
22626 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22628 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22629 for functions with noipa attribute.
22630 (cgraph_externally_visible_p): Return true for functions with noipa
22632 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22633 for functions with noipa attribute.
22634 * doc/extend.texi: Document noipa function attribute.
22635 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22636 also for functions with noipa attribute.
22637 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22639 2017-07-26 Andrew Pinski <apinski@cavium.com>
22641 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22642 vec_unalign_load_cost and vec_unalign_store_cost.
22644 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22646 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22647 -mvsx-small-integer option.
22648 (ISA_3_0_MASKS_IEEE): Likewise.
22649 (OTHER_VSX_VECTOR_MASKS): Likewise.
22650 (POWERPC_MASKS): Likewise.
22651 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22652 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22653 code, only testing for DImode being allowed in non-VSX floating
22655 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22656 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22658 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22659 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22660 TARGET_P8_VECTOR test.
22661 (rs6000_secondary_reload_simple_move): Likewise.
22662 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22663 since TARGET_P9_VECTOR was already tested.
22664 (rs6000_opt_masks): Remove -mvsx-small-integer.
22665 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22666 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22668 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22669 test for TARGET_VEXTRACTUB was used, and that uses
22671 (p9 extract splitter): Likewise.
22672 (vsx_extract_<mode>_di_p9): Likewise.
22673 (vsx_extract_<mode>_store_p9): Likewise.
22674 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22675 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22676 the elimination of TARGET_VSX_SMALL_INTEGER.
22677 (vsx_extract_<mode>_p8): Likewise.
22678 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22679 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22680 (vsx_set_<mode>_p9): Likewise.
22681 (vsx_set_v4sf_p9): Likewise.
22682 (vsx_set_v4sf_p9_zero): Likewise.
22683 (vsx_insert_extract_v4sf_p9): Likewise.
22684 (vsx_insert_extract_v4sf_p9_2): Likewise.
22685 * config/rs6000/rs6000.md (sign extend splitter): Change
22686 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22687 (floatsi<mode>2_lfiwax_mem): Likewise.
22688 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22689 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22690 since a test for TARGET_P9_VECTOR was used.
22691 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22692 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22693 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22694 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22695 TARGET_P8_VECTOR test.
22696 (fix_trunc<mode>si2_stfiwx): Likewise.
22697 (fix_trunc<mode>si2_internal): Likewise.
22698 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22699 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22701 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22702 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22703 TARGET_P8_VECTOR test.
22704 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22705 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22706 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22708 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22709 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22710 since a test for TARGET_P9_VECTOR was used.
22711 (splitter for loading small constants): Likewise.
22713 2017-07-26 Andrew Pinski <apinski@cavium.com>
22715 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22718 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22721 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22722 (fp_valid_at): Likewise.
22724 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22726 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22727 (qdf24xx_addrcost_table): Likewise.
22728 (cortexa57_tunings): Update to use generic_branch_cost.
22729 (cortexa72_tunings): Likewise.
22730 (cortexa73_tunings): Likewise.
22731 (qdf24xx_tunings): Likewise.
22733 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22735 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22736 (thunderx2t99_branch_cost): Likewise.
22737 (cortexa35_tunings): Update to use generic_branch_cost.
22738 (cortexa53_tunings): Likewise.
22739 (cortexa57_tunings): Likewise.
22740 (cortexa72_tunings): Likewise.
22741 (cortexa73_tunings): Likewise.
22742 (thunderx2t99_tunings): Likewise.
22744 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22746 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22747 (sparc_option_override): Honour MASK_FSMULD.
22748 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22749 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22750 * config/sparc/sparc.opt (mfsmuld): New option.
22751 * doc/invoke.texi (mfsmuld): Document option.
22753 2017-07-26 Marek Polacek <polacek@redhat.com>
22755 PR middle-end/70992
22756 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22758 2017-07-26 Richard Biener <rguenther@suse.de>
22760 * gimple-match-head.c (do_valueize): Return OP if valueize
22762 (get_def): New helper to get at the def stmt of a SSA name
22763 if valueize allows.
22764 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22765 do_valueize to get at the def stmt.
22766 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22768 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22770 PR middle-end/46932
22771 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22773 2017-07-26 Martin Liska <mliska@suse.cz>
22776 * function.c (expand_function_start): Make expansion of
22777 nonlocal_goto_save_area after parm_birth_insn.
22779 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22781 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22782 from all CPU target flags enable members.
22784 2017-07-26 Richard Biener <rguenther@suse.de>
22786 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22787 (decision_tree::gen): Make 'type' const.
22788 (write_predicate): Likewise.
22790 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22792 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22793 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22794 (rs6000_option_override_internal): Likewise.
22795 (rs6000_expand_vector_set): Likewise.
22796 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22797 (TARGET_UPPER_REGS_SF): Likewise.
22798 (TARGET_UPPER_REGS_DI): Likewise.
22799 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22800 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22801 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22802 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22803 (Splitters for DI constants in Altivec registers): Likewise.
22804 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22805 (vsx_set_v4sf_p9): Likewise.
22806 (vsx_set_v4sf_p9_zero): Likewise.
22807 (vsx_insert_extract_v4sf_p9): Likewise.
22808 (vsx_insert_extract_v4sf_p9_2): Likewise.
22810 2017-07-25 Carl Love <cel@us.ibm.com>
22812 * doc/extend.texi: Update the built-in documentation file for the
22813 existing built-in functions
22814 vector signed char vec_cnttz (vector signed char);
22815 vector unsigned char vec_cnttz (vector unsigned char);
22816 vector signed short vec_cnttz (vector signed short);
22817 vector unsigned short vec_cnttz (vector unsigned short);
22818 vector signed int vec_cnttz (vector signed int);
22819 vector unsigned int vec_cnttz (vector unsigned int);
22820 vector signed long long vec_cnttz (vector signed long long);
22821 vector unsigned long long vec_cnttz (vector unsigned long long);
22823 2017-07-25 Andrew Pinski <apinski@cavium.com>
22825 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22826 accesses where the use is for the first operand of a BIT_INSERT.
22828 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22831 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22832 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22834 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22836 * config/i386/gstabs.h: Delete.
22837 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22839 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22841 * config/i386/i386.c (ix86_decompose_address): Do not check for
22842 register RTX when looking at index_reg or base_reg.
22843 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22845 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22847 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22848 to update EH info here.
22850 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22852 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22854 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22856 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22858 2017-07-25 Torsten Duwe <duwe@suse.de>
22860 * common.opt: Introduce -fpatchable-function-entry
22861 command line option, and its variables function_entry_patch_area_size
22862 and function_entry_patch_area_start.
22863 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22864 including a two-value parser.
22865 * target.def (print_patchable_function_entry): New target hook.
22866 * targhooks.h (default_print_patchable_function_entry): New function.
22867 * targhooks.c (default_print_patchable_function_entry): Likewise.
22868 * toplev.c (process_options): Switch off IPA-RA if
22869 patchable function entries are being generated.
22870 * varasm.c (assemble_start_function): Look at the
22871 patchable-function-entry command line switch and current
22872 function attributes and maybe generate NOP instructions by
22873 calling the print_patchable_function_entry hook.
22874 * doc/extend.texi: Document patchable_function_entry attribute.
22875 * doc/invoke.texi: Document -fpatchable_function_entry
22876 command line option.
22877 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22879 * doc/tm.texi: Re-generate.
22881 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22884 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22885 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22887 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22889 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22891 2017-07-25 Richard Biener <rguenther@suse.de>
22893 PR tree-optimization/81455
22894 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22895 not walk in cycles when looking for guards.
22897 2017-07-25 Richard Biener <rguenther@suse.de>
22899 PR tree-optimization/81529
22900 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22901 when optimizing backedge uses.
22903 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22905 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22907 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22908 to dl_section_ref. On AIX, append an expression to subtract
22909 the size of the section length to dl_section_ref.
22911 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22913 * configure.ac: If any of the config.* scripts fail, exit 1.
22914 * configure: Regenerate.
22916 2017-07-25 Richard Biener <rguenther@suse.de>
22918 PR middle-end/81546
22919 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22920 of immediate uses, be more verbose on errors.
22922 2017-07-25 Richard Biener <rguenther@suse.de>
22924 PR tree-optimization/81510
22925 * tree-vect-loop.c (vect_is_simple_reduction): When the
22926 reduction stmt is not inside the loop bail out.
22928 2017-07-25 Richard Biener <rguenther@suse.de>
22930 PR tree-optimization/81303
22931 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22932 profitability check against LOOP_VINFO_NITERSM1.
22934 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22936 * domwalk.c (cmp_bb_postorder): Simplify.
22937 (sort_bbs_postorder): New function. Use it...
22938 (dom_walker::walk): ...here to optimize common cases.
22940 2017-07-25 Martin Liska <mliska@suse.cz>
22943 * ipa-visibility.c (function_and_variable_visibility): Make the
22944 redirection just on target that supports aliasing.
22945 Fix GNU coding style.
22947 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22950 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22952 * config/arm/rtems.h: Likewise.
22953 * config/bfin/rtems.h: Likewise.
22954 * config/i386/rtemself.h: Likewise.
22955 * config/lm32/rtems.h: Likewise.
22956 * config/m32c/rtems.h: Likewise.
22957 * config/m68k/rtemself.h: Likewise.
22958 * config/microblaze/rtems.h: Likewise.
22959 * config/mips/rtems.h: Likewise.
22960 * config/moxie/rtems.h: Likewise.
22961 * config/nios2/rtems.h: Likewise.
22962 * config/powerpcspe/rtems.h: Likewise.
22963 * config/rs6000/rtems.h: Likewise.
22964 * config/rtems.h: Likewise.
22965 * config/sh/rtems.h: Likewise.
22966 * config/sh/rtemself.h: Likewise.
22967 * config/sparc/rtemself.h: Likewise.
22969 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22972 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22973 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22974 * tree-ssa-structalias.c (alias_get_name): Same.
22976 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22979 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22980 instructions if no du chain is found.
22982 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22984 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22986 2017-07-25 Richard Biener <rguenther@suse.de>
22988 PR middle-end/81505
22989 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22992 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22994 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22995 upper-regs options.
22996 (ISA_2_7_MASKS_SERVER): Likewise.
22997 (ISA_3_0_MASKS_IEEE): Likewise.
22998 (OTHER_P8_VECTOR_MASKS): Likewise.
22999 (OTHER_VSX_VECTOR_MASKS): Likewise.
23000 (POWERPC_MASKS): Likewise.
23001 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23002 duplicate list of options.
23003 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23004 explicit -mupper-regs options.
23005 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23006 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23007 alias for -mupper-regs-df.
23008 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23009 (rs6000_init_hard_regno_mode_ok): Likewise.
23010 (rs6000_option_override_internal): Likewise.
23011 (rs6000_opt_masks): Likewise.
23012 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23013 options in terms of whether -mvsx or -mpower8-vector was used.
23014 (TARGET_UPPER_REGS_DI): Likewise.
23015 (TARGET_UPPER_REGS_SF): Likewise.
23016 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23017 -mupper-regs-* options.
23019 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23021 * passes.c (emergency_dump_function): Print some empty lines and a
23022 header before the RTL dump.
23024 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23026 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23028 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23031 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23032 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23034 2017-07-24 Carl Love <cel@us.ibm.com>
23036 * config/rs6000/rs6000-c.c: Add support for built-in functions
23037 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23038 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23039 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23040 vec_extract_fp_from_shortl): Add defines for the two builtins.
23041 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23042 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23044 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23045 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23046 * doc/extend.texi: Update the built-in documentation file for the
23047 new built-in function.
23049 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23052 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23054 * doc/generic.texi: Likewise.
23055 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23056 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23058 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23060 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23061 (aarch64_mls_elt_merge<mode>): Likewise.
23063 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23065 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23066 having __cxa_atexit.
23068 2017-07-23 Michael Collison <michael.collison@arm.com>
23070 * config/arm/arm.c (arm_option_override): Deprecate
23071 use of -mstructure-size-boundary.
23072 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23073 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23075 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23078 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23079 Reduce cost estimate for direct moves.
23081 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23084 * config/i386/i386.c (ix86_option_override_internal): Disable
23085 BMI, BMI2 and TBM instructions for -m16.
23087 2017-07-21 Carl Love <cel@us.ibm.com>
23089 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23090 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23091 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23092 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23093 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23094 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23095 VMULOSW): New enum "unspec" values.
23096 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23097 altivec_vmulosw): New patterns.
23098 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23099 VMULOSW): Add definitions.
23101 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23103 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23104 (qdf24xx): Likewise.
23105 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23106 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23107 (AARCH64_FL_V8_1): Renumber.
23108 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23109 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23110 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23111 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23112 rdma to feature modifiers list.
23114 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23116 PR middle-end/56727
23117 * ipa-visibility (function_and_variable_visibility): Convert
23118 recursive PLT call to direct call if appropriate.
23120 2017-07-21 Andrew Pinski <apinski@cavium.com>
23122 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23123 operand 1 to see if the types precision matches.
23124 * fold-const.c (operand_equal_p): Likewise.
23126 2017-07-21 Richard Biener <rguenther@suse.de>
23128 PR tree-optimization/81303
23129 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23130 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23131 (vect_peeling_hash_get_lowest_cost): Adjust.
23132 (vect_enhance_data_refs_alignment): Likewise. Use
23133 vect_get_peeling_costs_all_drs to compute the penalty for no
23134 peeling to match up costs.
23136 2017-07-21 Richard Biener <rguenther@suse.de>
23138 PR tree-optimization/81500
23139 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23140 we didn't identify a reduction path.
23142 2017-07-21 Tom de Vries <tom@codesourcery.com>
23143 Cesar Philippidis <cesar@codesourcery.com>
23145 PR gcov-profile/81442
23146 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23149 2017-07-21 Tom de Vries <tom@codesourcery.com>
23152 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23154 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23155 nvptx_override_options_after_change.
23157 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23159 * dwarf2out.c (output_file_names): Avoid double testing for
23160 dwarf_version >= 5.
23162 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23164 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23166 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23168 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23170 (try_crossjump_to_edge): Do not punt on partitioned functions.
23172 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23174 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23175 Put all BBs reachable only via paths crossing cold region to cold
23177 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23179 2016-07-21 Richard Biener <rguenther@suse.de>
23181 PR tree-optimization/81303
23182 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23183 into account prologue and epilogue iterations when raising
23184 min_profitable_iters to sth at least covering one vector iteration.
23186 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23188 * config/arm/arm.c (arm_test_cpu_arch_dat):
23191 2017-07-20 Nathan Sidwell <nathan@acm.org>
23193 Remove TYPE_METHODS.
23194 * tree.h (TYPE_METHODS): Delete.
23195 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23196 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23197 (dbxout_type_methods): Scan TYPE_FIELDS.
23198 (dbxout_type): Don't check TYPE_METHODS here.
23199 * function.c (use_register_for_decl): Always ignore register for
23200 class types when not optimizing.
23201 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23202 * tree.c (free_lang_data_in_type): Stitch out member functions and
23203 templates from TYPE_FIELDS.
23204 (build_distinct_type_copy, verify_type_variant,
23205 verify_type): Member fns are on TYPE_FIELDS.
23206 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23207 * tree-pretty-print.c (dump_generic_node): Likewise.
23209 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23212 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23213 V2TImode and V4TImode.
23214 (ix86_expand_vector_extract): Likewise.
23215 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23216 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23217 (ssescalarmode): Handle V4TImode and V2TImode.
23218 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23219 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23220 (VEXTRACTI128_MODE): New mode iterator.
23221 (splitter for *vec_extractv?ti first element): New.
23222 (VEC_INIT_MODE): New mode iterator.
23223 (vec_init<mode>): Consolidate 3 expanders into one using
23224 VEC_INIT_MODE mode iterator.
23226 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23228 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23229 non_spilled_static_chain_regno_p.
23231 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23233 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23235 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23237 * bb-reorder.c (connect_traces): Allow copying of blocks within
23240 2017-07-20 Richard Biener <rguenther@suse.de>
23242 * gimple.h (gimple_phi_result): Add gphi * overload.
23243 (gimple_phi_result_ptr): Likewise.
23244 (gimple_phi_arg): Likewise. Adjust index assert to only
23245 allow actual argument accesses rather than all slots available
23247 (gimple_phi_arg_def): Add gphi * overload.
23248 * tree-phinodes.c (make_phi_node): Initialize only actual
23250 (resize_phi_node): Clear memory not covered by old node,
23251 do not initialize excess argument slots.
23252 (reserve_phi_args_for_new_edge): Initialize new argument slot
23255 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23257 PR tree-optimization/81388
23259 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23261 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23262 by removing computation of may_be_zero.
23264 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23265 Tom de Vries <tom@codesourcery.com>
23267 PR middle-end/81030
23268 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23269 when gimple level profile disagrees with what RTL expander did.
23271 2017-07-20 Richard Biener <rguenther@suse.de>
23273 PR tree-optimization/61171
23274 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23275 (vect_analyze_stmt): Add slp instance parameter.
23276 (vectorizable_reduction): Likewise.
23277 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23278 (vect_is_simple_reduction): Deal with chains not detected
23279 as SLP reduction chain, specifically not properly associated
23280 chains containing a mix of plus/minus.
23281 (get_reduction_op): Remove.
23282 (get_initial_defs_for_reduction): Simplify, pass in whether
23283 this is a reduction chain, pass in the SLP node for the PHIs.
23284 (vect_create_epilog_for_reduction): Get the SLP instance as
23286 (vectorizable_reduction): Get the SLP instance as arg.
23287 During analysis remember the SLP node with the PHIs in the
23288 instance. Simplify getting at the vectorized reduction PHIs.
23289 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23290 through SLP instance.
23291 (vect_slp_analyze_operations): Likewise.
23292 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23293 (vect_transform_stmt): Likewise.
23295 2017-07-20 Tom de Vries <tom@codesourcery.com>
23297 PR tree-optimization/81489
23298 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23299 read of phi arg location to before loop that modifies phi.
23301 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23303 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23306 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23308 PR middle-end/81331
23309 * except.c (execute): Fix ordering issue.
23311 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23313 PR rtl-optimization/81423
23314 * combine.c (make_compound_operation_int): Don't try to optimize
23315 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23317 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23319 PR rtl-optimization/81423
23320 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23321 with a constant that is -1 in the truncated to mode.
23323 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23325 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23326 (determine_unlikely_bbs): ... here.
23327 * predict.h (propagate_unlikely_bbs_forward): Declare.
23328 * cfgexpand.c (pass_expand::execute): Use it.
23329 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23331 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23332 propagate_unlikely_bbs_forward.
23334 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23336 PR middle-end/81331
23337 * except.c (maybe_add_nop_after_section_switch): New function.
23340 2017-07-19 Tom de Vries <tom@codesourcery.com>
23342 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23344 2017-07-19 Tom de Vries <tom@codesourcery.com>
23346 * gimple.h (gimple_phi_arg): Make assert more strict.
23348 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23350 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23351 * config/rs6000/mmintrin.h: New file.
23352 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23354 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23356 PR tree-optimization/81346
23357 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23359 2017-07-19 Tom de Vries <tom@codesourcery.com>
23361 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23363 2017-07-19 Tom de Vries <tom@codesourcery.com>
23365 * config/nvptx/nvptx-modes.def: Add V2DImode.
23366 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23367 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23368 (nvptx_output_mov_insn): Handle lack of mov.b128.
23369 (nvptx_print_operand): Handle 'H' and 'L' codes.
23370 (nvptx_vector_mode_supported): Allow V2DImode.
23371 (nvptx_preferred_simd_mode): New function.
23372 (nvptx_data_alignment): New function.
23373 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23374 nvptx_preferred_simd_mode.
23375 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23377 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23379 2017-07-19 Tom de Vries <tom@codesourcery.com>
23381 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23382 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23383 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23384 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23385 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23386 (mov<VECIM>_insn): New define_insn.
23387 (define_expand "mov<VECIM>): New define_expand.
23389 2017-07-19 Tom de Vries <tom@codesourcery.com>
23391 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23393 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23395 PR tree-optimization/81346
23396 * fold-const.h (fold_div_compare, range_check_type): Declare.
23397 * fold-const.c (range_check_type): New function.
23398 (build_range_check): Use range_check_type.
23399 (fold_div_compare): No longer static, rewritten into
23400 a match.pd helper function.
23401 (fold_comparison): Don't call fold_div_compare here.
23402 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23403 as helper function.
23405 2017-07-19 Nathan Sidwell <nathan@acm.org>
23407 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23408 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23409 * tree.c (find_decls_types_r, verify_type): Use
23410 TYPE_{MIN,MAX}_VALUE_RAW.
23411 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23412 (hash_tree): Likewise.
23413 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23415 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23418 2017-07-18 Tom de Vries <tom@codesourcery.com>
23420 PR middle-end/81464
23421 * omp-expand.c (expand_omp_for_static_chunk): Handle
23422 equal-argument loop exit phi.
23424 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23427 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23428 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23429 operand 2 predicate.
23430 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23431 operand 2 predicate.
23432 (ror,rol -> rorx splitters): Use const_int_operand as
23433 operand 2 predicate.
23435 2017-06-18 Richard Biener <rguenther@suse.de>
23437 PR tree-optimization/81410
23438 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23439 the gap in the ! slp_perm SLP case after each group.
23441 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23443 PR middle-end/81463
23444 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23447 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23449 PR middle-end/81462
23450 * predict.c (set_even_probabilities): Cleanup; do not affect
23451 probabilities that are already known.
23452 (combine_predictions_for_bb): Call even when count is set.
23454 2017-07-18 Nathan Sidwell <nathan@acm.org>
23456 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23459 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23462 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23463 optimization for loop niter analysis.
23465 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23468 * config/avr/avr.c (avr_optimize_casesi): Don't use
23469 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23471 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23473 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23474 body_cost_vec from _vect_peel_extended_info.
23475 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23476 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23479 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23481 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23483 2017-07-18 Richard Biener <rguenther@suse.de>
23485 PR tree-optimization/80620
23486 PR tree-optimization/81403
23487 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23488 info when re-using a VN table entry.
23490 2017-07-18 Richard Biener <rguenther@suse.de>
23492 PR tree-optimization/81418
23493 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23494 vectype_in. Verify that with lane-reducing reduction operations
23495 we have a single def-use cycle.
23497 2017-07-17 Carl Love <cel@us.ibm.com>
23499 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23501 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23502 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23503 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23504 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23505 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23506 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23507 VMULOSW): New enum "unspec" values.
23508 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23509 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23510 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23511 altivec_vmulosw): New patterns.
23512 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23513 VMULOSW): Add definitions.
23515 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23517 * config/alpha/alpha.c: Include predict.h.
23519 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23521 * tree-vrp.c (compare_assert_loc): Fix comparison function
23522 to return predictable results.
23524 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23526 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23528 (subdi3): Likewise.
23529 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23530 * doc/invoke.texi (mexpand-adddi): Update text.
23532 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23534 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23535 that also clobbers the CC register. The old expand code is moved
23537 (*arc_clzsi2): ... here.
23538 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23539 the CC register. The old expand code is moved to ...
23540 (arc_ctzsi2): ... here.
23542 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23544 * config/arc/arc.opt (mindexed-loads): Use initial value
23545 TARGET_INDEXED_LOADS_DEFAULT.
23546 (mauto-modify-reg): Use initial value
23547 TARGET_AUTO_MODIFY_REG_DEFAULT.
23548 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23549 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23550 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23551 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23553 2017-07-17 Martin Liska <mliska@suse.cz>
23556 * opts.c (finish_options): Do not allow -fgnu-tm
23557 w/ -fsanitize={kernel-,}address. Say sorry.
23559 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23562 * tree-loop-distribution.c (classify_partition): Only assert on
23563 numer of iterations.
23564 (merge_dep_scc_partitions): Delete prameter. Update function call.
23565 (distribute_loop): Remove code handling loop with unknown niters.
23566 (pass_loop_distribution::execute): Skip loop with unknown niters.
23568 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23571 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23572 function sort_partitions_by_post_order.
23574 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23576 PR tree-optimization/81374
23577 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23578 the max index of basic blocks, rather than number of basic blocks.
23580 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23582 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23584 (arc_legitimate_pic_operand_p): Likewise.
23585 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23587 (arc_needs_pcl_p): Likewise.
23588 (arc_legitimate_pc_offset_p): Likewise.
23589 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23590 function is also used in constrains.md.
23591 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23592 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23593 PLUS. Only return true/false in known cases, otherwise assert.
23594 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23595 is already called in arc_legitimate_constant_p.
23596 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23598 (LEGITIMATE_PIC_OPERAND_P): Use
23599 arc_raw_symbolic_reference_mentioned_p function.
23600 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23605 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23606 Andrew Burgess <andrew.burgess@embecosm.com>
23608 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23609 (arc_return_address_register): New function.
23610 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23611 (arc_handle_fndecl_attribute): Add naked attribute.
23612 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23613 (TARGET_WARN_FUNC_RETURN): Likewise.
23614 (arc_allocate_stack_slots_for_args): New function.
23615 (arc_warn_func_return): Likewise.
23616 (machine_function): Change type fn_type.
23617 (arc_compute_function_type): Consider new naked function type,
23618 change function return type.
23619 (arc_must_save_register): Adapt to handle new
23620 arc_compute_function_type's return type.
23621 (arc_expand_prologue): Likewise.
23622 (arc_expand_epilogue): Likewise.
23623 (arc_return_address_regs): Delete.
23624 (arc_return_address_register): New function.
23625 (arc_epilogue_uses): Use above function.
23626 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23627 (arc_function_type): Change encoding, add naked type.
23628 (ARC_INTERRUPT_P): Change to handle the new encoding.
23629 (ARC_FAST_INTERRUPT_P): Likewise.
23630 (ARC_NORMAL_P): Define.
23631 (ARC_NAKED_P): Likewise.
23632 (arc_compute_function_type): Delete prototype.
23633 * config/arc/arc.md (in_ret_delay_slot): Use
23634 arc_return_address_register function.
23635 (simple_return): Likewise.
23636 (p_return_i): Likewise.
23638 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23640 PR tree-optimization/81428
23641 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23642 can't be built for those types.
23644 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23646 Remove stuff dead since r239246.
23648 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23649 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23650 (avr_inform_devices): Remove dead stuff.
23652 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23654 * config/arm/arm_neon.h: Fix softp typo.
23656 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23658 PR tree-optimization/81365
23659 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23660 aggregate moves onto bb predecessor edges, make sure there are no
23661 loads that could alias the lhs in between the start of bb and the
23664 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23667 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23668 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23669 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23671 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23673 PR tree-optimization/81396
23674 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23675 (init_symbolic_number): Initialize it to 1.
23676 (perform_symbolic_merge): Add n_ops from both operands into the new
23678 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23679 without base_addr as useless if they need more than one operation.
23680 (bswap_replace): Handle !bswap case for NULL base_addr.
23682 2017-07-17 Tom de Vries <tom@codesourcery.com>
23685 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23688 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23690 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23691 conditional builtin define __FIX_LEON3FT_B2BST.
23693 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23695 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23696 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23699 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23701 PR rtl-optimization/81424
23702 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23703 to remove potential trapping from operands if -fnon-call-exceptions.
23705 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23707 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23708 profile_proability for scalling.
23709 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23711 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23713 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23715 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23717 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23718 fixpoint arithmetics.
23720 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23722 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23723 fixpoint arithmetics.
23725 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23727 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23728 fixpoint arithmetics.
23730 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23732 * profile-count.h (profile_probability::from_reg_br_prob_note,
23733 profile_probability::to_reg_br_prob_note): New functions.
23734 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23735 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23736 * predict.c (probability_reliable_p): Update.
23737 (edge_probability_reliable_p): Update.
23738 (br_prob_note_reliable_p): Update.
23739 (invert_br_probabilities): Update.
23740 (add_reg_br_prob_note): New function.
23741 (combine_predictions_for_insn): Update.
23742 * asan.c (asan_clear_shadow): Update.
23743 * cfgbuild.c (compute_outgoing_frequencies): Update.
23744 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23745 (update_br_prob_note): Update.
23746 (rtl_verify_edges): Update.
23747 (purge_dead_edges): Update.
23748 (fixup_reorder_chain): Update.
23749 * emit-rtl.c (try_split): Update.
23750 * ifcvt.c (cond_exec_process_insns): Update.
23751 (cond_exec_process_if_block): Update.
23752 (dead_or_predicable): Update.
23753 * internal-fn.c (expand_addsub_overflow): Update.
23754 (expand_neg_overflow): Update.
23755 (expand_mul_overflow): Update.
23756 * loop-doloop.c (doloop_modify): Update.
23757 * loop-unroll.c (compare_and_jump_seq): Update.
23758 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23759 * predict.h: Update.
23760 * reorg.c (mostly_true_jump): Update.
23762 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23763 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23764 * config/arc/arc.c: (emit_unlikely_jump): Update.
23765 * config/arm/arm.c: (emit_unlikely_jump): Update.
23766 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23767 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23768 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23769 (ix86_print_operand): Update.
23770 (ix86_split_fp_branch): Update.
23771 (predict_jump): Update.
23772 * config/ia64/ia64.c (ia64_print_operand): Update.
23773 * config/mmix/mmix.c (mmix_print_operand): Update.
23774 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23775 (rs6000_expand_split_stack_prologue): Update.
23776 * config/rs6000/rs6000.c: Update.
23777 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23778 (s390_expand_vec_movstr): Update.
23779 (s390_expand_cs_tdsi): Update.
23780 (s390_expand_split_stack_prologue): Update.
23781 * config/sh/sh.c (sh_print_operand): Update.
23782 (expand_cbranchsi4): Update.
23783 (expand_cbranchdi4): Update.
23784 * config/sparc/sparc.c (output_v9branch): Update.
23785 * config/spu/spu.c (get_branch_target): Update.
23786 (ea_load_store_inline): Update.
23787 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23788 * config/tilepro/tilepro.c: Update.
23790 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23792 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23793 (gimplify_save_expr): Likewise.
23795 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23797 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23799 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23801 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23803 * timevar.def (TV_IPA_FNSUMMARY): Define.
23805 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23807 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23808 to back store errata sensitive sequence from being generated.
23809 (sqrtdf2_fix): Likewise.
23811 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23813 * tree-ssa-threadupdate.c (compute_path_counts,
23814 update_joiner_offpath_counts): Use profile_probability.
23816 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23819 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23821 * config/arm/arm-c.c (arm_cpu_builtins): Define
23822 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23824 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23826 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23827 array entries to represent __ieee128 versions of the
23828 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23829 scalar_extract_sig, and scalar_insert_exp built-in functions.
23830 (altivec_resolve_overloaded_builtin): Add special case handling
23831 for the __builtin_scalar_insert_exp function, as represented by
23832 the P9V_BUILTIN_VEC_VSIEDP constant.
23833 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23834 exponent support for __ieee128 argument.
23835 (VSESQP): Add scalar extract signature support for __ieee128
23837 (VSTDCNQP): Add scalar test negative support for __ieee128
23839 (VSIEQP): Add scalar insert exponent support for __int128 argument
23840 with __ieee128 result.
23841 (VSIEQPF): Add scalar insert exponent support for __ieee128
23842 argument with __ieee128 result.
23843 (VSTDCQP): Add scalar test data class support for __ieee128
23845 (VSTDCNQP): Add overload support for scalar test negative with
23846 __ieee128 argument.
23847 (VSTDCQP): Add overload support for scalar test data class
23848 __ieee128 argument.
23849 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23850 UNSPEC_VSX_SXSIGDP.
23851 (UNSPEC_VSX_SIEXPQP): New constant.
23852 (xsxexpqp): New insn for VSX scalar extract exponent quad
23854 (xsxsigqp): New insn for VSX scalar extract significand quad
23856 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23857 precision with floating point argument.
23858 (xststdcqp): New expand for VSX scalar test data class quad
23860 (xststdcnegqp): New expand for VSX scalar test negative quad
23862 (xststdcqp): New insn to match expansions for VSX scalar test data
23863 class quad precision and VSX scalar test negative quad precision.
23864 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23865 special case operand checking to enforce that second operand of
23866 VSX scalar test data class with quad precision argument is a 7-bit
23868 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23869 prototypes and descriptions of __ieee128 versions of
23870 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23871 scalar_test_data_class, and scalar_test_neg built-in functions.
23873 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23875 PR tree-optimization/81162
23876 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23877 replace a negate with an add.
23879 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23881 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23883 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23885 * config/arm/arm-c.c (arm_cpu_builtins): Define
23886 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23888 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23890 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23891 (armv8-r): Set ARM Cortex-R52 as default CPU.
23892 * config/arm/arm-tables.opt: Regenerate.
23893 * config/arm/arm-tune.md: Regenerate.
23894 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23896 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23897 extension for -mcpu=cortex-r52.
23899 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23901 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23902 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23903 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23904 (fp-armv8): Define it as FP_ARMv8 only.
23905 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23906 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23908 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23909 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23910 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23911 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23912 than TARGET_FPU_ARMV8.
23913 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23914 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23915 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23917 * config/arm/neon.md (neon_vrint): Likewise.
23918 (neon_vcvt): Likewise.
23919 (neon_<fmaxmin_op><mode>): Likewise.
23920 (<fmaxmin><mode>3): Likewise.
23921 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23922 * config/arm/predicates.md (arm_cond_move_operator): Check against
23923 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23925 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23927 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23928 to top of function.
23930 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23932 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23933 loop in comment with memset.
23935 2017-07-14 Martin Liska <mliska@suse.cz>
23937 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23938 * dwarf2out.c (is_java): Remove the function.
23939 (output_pubname): Remove usage of the function.
23940 (lower_bound_default): Remove usage of DW_LANG_Java.
23941 (gen_compile_unit_die): Likewise.
23942 * gcc.c: Remove compiler defaults for .java and .zip files.
23943 * gimple-expr.c (remove_suffix): Change as there's no longer
23944 extension than 4-letter one.
23945 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23946 (gimplify_save_expr): Likewise.
23947 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23948 as it's possible even for other languages than Java.
23949 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23950 * lto-opts.c (lto_write_options): Remove reference to Java.
23951 * opts.c (strip_off_ending): Update file extension handling.
23952 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23953 * tree-eh.c (lower_resx): Likewise.
23954 * tree.c (free_lang_data_in_type): Remove dead code.
23955 (find_decls_types_r): Likewise.
23956 (build_common_builtin_nodes): Remove Java from a comment.
23957 (verify_type): Remove dead code.
23958 * varasm.c (assemble_external): Remove Java from a comment.
23960 2017-07-14 Martin Liska <mliska@suse.cz>
23962 * opts.c (finish_options): Add quotes.
23963 (common_handle_option): Likewise.
23965 2017-07-14 Martin Liska <mliska@suse.cz>
23967 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23968 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23969 Remove N_SO_PASCAL.
23970 * dwarf2out.c (lower_bound_default): Do not handle
23972 (gen_compile_unit_die): Likewise.
23973 * gcc.c: Remove default extension binding for GNU Pascal.
23974 * stmt.c: Remove Pascal language from a comment.
23975 * xcoffout.c: Likewise.
23977 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23980 * diagnostic-show-locus.c (fixit_cmp): New function.
23981 (layout::layout): Sort m_fixit_hints.
23982 (column_range::column_range): Assert that the values are valid.
23983 (struct char_span): New struct.
23984 (correction::overwrite): New method.
23985 (struct source_line): New struct.
23986 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23987 calls in terms of classes source_line and char_span, and
23988 correction::overwrite.
23989 (selftest::test_overlapped_fixit_printing_2): New function.
23990 (selftest::diagnostic_show_locus_c_tests): Call it.
23992 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23994 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23995 early if there is no lhs.
23997 2017-07-13 Martin Liska <mliska@suse.cz>
23999 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24000 (gen_reference_type_die): Likewise.
24001 * stor-layout.c: Remove Pascal-related comment.
24003 2017-07-13 Martin Liska <mliska@suse.cz>
24005 * opts.c (finish_options): Add quotes to error messages.
24006 (parse_sanitizer_options): Likewise.
24008 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24010 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24012 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24014 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24016 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24018 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24020 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24022 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24025 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24026 provides the hardware capability bits, define the macro
24027 __BUILTIN_CPU_SUPPORTS__.
24028 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24029 if GLIBC does not provide the hardware capability bits. Add a
24030 gcc_unreachable call if the built-in cpu function is neither
24031 __builtin_cpu_is nor __builtin_cpu_supports.
24032 (rs6000_get_function_versions_dispatcher): Change the warning
24033 that an old GLIBC is used which does not export the capability
24034 bits to be an error.
24035 * doc/extend.texi (target_clones attribute): Document the
24036 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24037 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24038 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24039 the macros defined by GCC if the newer GLIBC is available.
24041 2017-07-12 Jeff Law <law@redhat.com>
24043 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24044 remaining includes slightly.
24045 * config/riscv/riscv-builtins.c: Include profile-count.h.
24047 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24050 * config/avr/avr.c (avr_set_current_function): In diagnostic
24051 messages: Quote keywords and (parts of) identifiers.
24052 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24055 2017-07-12 Carl Love <cel@us.ibm.com>
24057 * config/rs6000/rs6000-c.c: Add support for built-in functions
24058 vector bool char vec_revb (vector bool char);
24059 vector bool short vec_revb (vector short char);
24060 vector bool int vec_revb (vector bool int);
24061 vector bool long long vec_revb (vector bool long long);
24062 * doc/extend.texi: Update the built-in documentation file for the
24063 new built-in functions.
24065 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24067 * config/s390/s390.md: Remove movcc splitter.
24069 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24071 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24072 load/store on condition.
24074 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24077 * config/avr/avr.c (avr_encode_section_info)
24078 [progmem && !TREE_READONLY]: Error if progmem object needs
24081 2017-07-11 Michael Collison <michael.collison@arm.com>
24083 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24086 2017-07-11 Carl Love <cel@us.ibm.com>
24088 * config/rs6000/rs6000-c.c: Add support for builtins
24089 vector unsigned int vec_parity_lsbb (vector signed int);
24090 vector unsigned int vec_parity_lsbb (vector unsigned int);
24091 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24092 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24093 vector unsigned long long vec_parity_lsbb (vector signed long long);
24094 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24095 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24096 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24097 * doc/extend.texi: Update the built-in documentation file for the
24098 new built-in functions.
24100 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24102 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24103 (layout::m_primary_loc): New field.
24104 (layout::layout): Initialize new field. Move location filtering
24105 logic from here to...
24106 (layout::maybe_add_location_range): ...this new method. Add
24107 support for filtering to just the lines already specified by other
24109 (layout::will_show_line_p): New method.
24110 (gcc_rich_location::add_location_if_nearby): New method.
24111 (selftest::test_add_location_if_nearby): New test function.
24112 (selftest::diagnostic_show_locus_c_tests): Call it.
24113 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24116 2017-07-11 Tom de Vries <tom@codesourcery.com>
24118 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24119 (bb_first_real_insn): New function.
24120 (nvptx_single): Add extra initialization of broadcasted condition
24123 2017-07-11 Nathan Sidwell <nathan@acm.org>
24125 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24127 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24129 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24130 Binutils doc as TEXI will mess them up.
24131 * doc/invoke.texi (AVR Options): Same here.
24133 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24135 * config/sparc/sparc.opt (mfix-ut700): New option.
24136 (mfix-gr712rc): Likewise.
24137 (sparc_fix_b2bst): New variable.
24138 * doc/invoke.texi (SPARC options): Document them.
24139 (ARM options): Fix warnings.
24140 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24141 instructions to prevent sequences that can trigger the store-store
24142 errata for certain LEON3FT processors.
24143 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24144 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24145 * config/sparc/sparc.md (fix_b2bst): New attribute.
24146 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24148 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24151 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24153 (*rsqrtsf2_sse): Ditto.
24155 (div<mode>3): Macroize insn from divdf3 and divsf3
24156 using MODEF mode iterator.
24158 2017-07-10 Martin Sebor <msebor@redhat.com>
24160 PR tree-optimization/80397
24161 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24162 instead of testing for equality to INTEGER_TYPE.
24164 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24166 * config.gcc: Remove uclibc from arc target spec.
24168 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24170 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24172 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24175 * lto-wrapper.c (remove_option): New function.
24176 (merge_and_complain): Merge PIC/PIE options more realistically.
24178 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24180 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24184 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24185 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24186 * config.in: Regenerate.
24187 * configure: Regenerate.
24188 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24189 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24190 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24191 (TARGET_GASISR_PROLOGUES): ...target mask.
24192 * common/config/avr/avr-common.c
24193 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24194 Set -mgas-isr-prologues.
24195 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24196 INSERT_PASS_BEFORE for it.
24197 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24198 * config/avr/avr.c (avr_option_override)
24199 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24200 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24201 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24202 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24203 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24205 (make_avr_pass_pre_proep): New function.
24206 (emit_push_sfr) <treg>: Add argument to function and use it
24207 instead of TMP_REG.
24208 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24209 and set machine->gasisr.yes.
24210 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24211 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24212 __gcc_isr.n_pushed to .L__stack_usage.
24213 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24214 (avr_asm_final_postscan_insn): ...this new static function.
24215 * config/avr/avr.h (machine_function)
24216 <is_no_gccisr, use_L__stack_usage>: New fields.
24217 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24218 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24219 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24220 (gasisr, *gasisr): New expander and insn.
24221 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24222 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24223 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24225 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24227 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24230 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24232 Move jump-tables out of .text again.
24235 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24236 (ASM_OUTPUT_ADDR_VEC): New function.
24237 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24238 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24239 INSN_ADDRESSes as asm comment.
24240 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24241 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24242 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24243 * config/avr/avr.md (*tablejump): Adjust comment.
24244 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24245 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24247 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24248 (avr_output_addr_vec): New proto.
24249 (avr_log_t) <insn_addresses>: New field.
24251 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24254 * config/i386/i386.c (ix86_function_arg_advance): Set
24255 outgoing_args_on_stack to true if there are outgoing arguments
24257 (ix86_function_arg): Likewise.
24258 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24259 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24260 * config/i386/i386.h (machine_function): Add
24261 outgoing_args_on_stack.
24263 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24265 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24266 supporting pthreds.
24267 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24269 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24271 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24272 (REAL_H): Remove $(MACHMODE_H).
24273 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24275 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24276 $(MACHMODE_H) and double-int.h.
24277 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24279 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24282 2017-07-07 Andrew Pinski <apinski@cavium.com>
24284 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24285 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24287 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24289 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24290 Add warning if GCC was not configured to link against a GLIBC that
24291 exports the hardware capability bits.
24292 (make_resolver_func): Make resolver function private and not a
24293 COMDAT function. Create the name with clone_function_name instead
24294 of make_unique_name.
24297 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24298 correct operand in doing the split.
24300 2017-07-07 Carl Love <cel@us.ibm.com>
24302 * config/rs6000/rs6000-c: Add support for built-in function
24303 vector unsigned short vec_pack_to_short_fp32 (vector float,
24305 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24306 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24307 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24308 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24309 (convert_4f32_8i16): Add define_expand.
24310 * doc/extend.texi: Update the built-in documentation file for the
24311 new built-in function.
24313 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24315 * config/sparc/m8.md: New file.
24316 * config/sparc/sparc.md: Include m8.md.
24318 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24320 * config/sparc/sparc.opt: New option -mvis4b.
24321 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24322 (sparc_option_override): Handle VIS4B.
24323 (enum sparc_builtins): Define
24324 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24325 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24326 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24327 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24328 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24329 (check_constant_argument): New function.
24330 (sparc_vis_init_builtins): Define builtins
24331 __builtin_vis_dictunpack{8,16,32},
24332 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24333 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24334 __builtin_vis_fpcmpde{8,16,32}shl and
24335 __builtin_vis_fpcmpur{8,16,32}shl.
24336 (sparc_expand_builtin): Check that the constant operands to
24337 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24338 constant and in range.
24339 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24341 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24342 (SPARC_IMM5_P): Likewise.
24343 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24344 (enabled): Handle vis4b.
24345 (UNSPEC_DICTUNPACK): New unspec.
24346 (UNSPEC_FPCMPSHL): Likewise.
24347 (UNSPEC_FPUCMPSHL): Likewise.
24348 (UNSPEC_FPCMPDESHL): Likewise.
24349 (UNSPEC_FPCMPURSHL): Likewise.
24350 (cpu_feature): New CPU feature `vis4b'.
24351 (dictunpack{8,16,32}): New insns.
24352 (FPCSMODE): New mode iterator.
24353 (fpcscond): New code iterator.
24354 (fpcsucond): Likewise.
24355 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24356 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24357 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24358 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24359 * config/sparc/constraints.md: Define constraints `q' for unsigned
24360 2-bit integer constants and `t' for unsigned 5-bit integer
24362 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24364 (imm5_operand_dictunpack16): Likewise.
24365 (imm5_operand_dictunpack32): Likewise.
24366 (imm2_operand): Likewise.
24367 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24368 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24369 ditunpack* and fpcmp*shl builtins.
24371 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24373 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24374 * config.in: Add HAVE_AS_SPARC6 define.
24375 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24377 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24379 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24380 (CPP_CPU_SPEC): Handle m8.
24381 (ASM_CPU_SPEC): Likewise.
24382 * config/sparc/sparc-opts.h (enum processor_type): Add
24384 * config/sparc/sparc.c (m8_costs): New struct.
24385 (sparc_option_override): Handle TARGET_CPU_m8.
24386 (sparc32_initialize_trampoline): Likewise.
24387 (sparc64_initialize_trampoline): Likewise.
24388 (sparc_issue_rate): Likewise.
24389 (sparc_register_move_cost): Likewise.
24390 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24391 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24392 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24393 (CPP_CPU_SPEC): Handle M8.
24394 (ASM_CPU_SPEC): Likewise.
24395 (AS_M8_FLAG): Define.
24396 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24397 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24398 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24400 * configure: Regenerate.
24401 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24404 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24406 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24408 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24409 ("*movdi_insn_sp32"): Do not set v3pipe.
24410 ("*movsi_insn"): Likewise.
24411 ("*movdi_insn_sp64"): Likewise.
24412 ("*movsf_insn"): Likewise.
24413 ("*movdf_insn_sp32"): Likewise.
24414 ("*movdf_insn_sp64"): Likewise.
24415 ("*zero_extendsidi2_insn_sp64"): Likewise.
24416 ("*sign_extendsidi2_insn"): Likewise.
24417 ("*mov<VM32:mode>_insn"): Likewise.
24418 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24419 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24420 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24421 ("<vlop:code><VL:mode>3"): Likewise.
24422 ("*not_<vlop:code><VL:mode>3"): Likewise.
24423 ("*nand<VL:mode>_vis"): Likewise.
24424 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24425 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24426 ("one_cmpl<VL:mode>2"): Likewise.
24427 ("faligndata<VM64:mode>_vis"): Likewise.
24428 ("alignaddrsi_vis"): Likewise.
24429 ("alignaddrdi_vis"): Likweise.
24430 ("alignaddrlsi_vis"): Likewise.
24431 ("alignaddrldi_vis"): Likewise.
24432 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24433 ("bmaskdi_vis"): Likewise.
24434 ("bmasksi_vis"): Likewise.
24435 ("bshuffle<VM64:mode>_vis"): Likewise.
24436 ("cmask8<P:mode>_vis"): Likewise.
24437 ("cmask16<P:mode>_vis"): Likewise.
24438 ("cmask32<P:mode>_vis"): Likewise.
24439 ("pdistn<P:mode>_vis"): Likewise.
24440 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24442 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24444 * config/sparc/sparc.md ("subtype"): New insn attribute.
24445 ("*wrgsr_sp64"): Set insn subtype.
24446 ("*rdgsr_sp64"): Likewise.
24447 ("alignaddrsi_vis"): Likewise.
24448 ("alignaddrdi_vis"): Likewise.
24449 ("alignaddrlsi_vis"): Likewise.
24450 ("alignaddrldi_vis"): Likewise.
24451 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24452 ("fexpand_vis"): Likewise.
24453 ("fpmerge_vis"): Likewise.
24454 ("faligndata<VM64:mode>_vis"): Likewise.
24455 ("bshuffle<VM64:mode>_vis"): Likewise.
24456 ("cmask8<P:mode>_vis"): Likewise.
24457 ("cmask16<P:mode>_vis"): Likewise.
24458 ("cmask32<P:mode>_vis"): Likewise.
24459 ("fchksm16_vis"): Likewise.
24460 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24461 ("fmean16_vis"): Likewise.
24462 ("fp<plusminus_insn>64_vis"): Likewise.
24463 ("<plusminus_insn>v8qi3"): Likewise.
24464 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24465 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24466 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24467 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24468 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24469 ("*movqi_insn"): Likewise.
24470 ("*movhi_insn"): Likewise.
24471 ("*movsi_insn"): Likewise.
24472 ("movsi_pic_gotdata_op"): Likewise.
24473 ("*movdi_insn_sp32"): Likewise.
24474 ("*movdi_insn_sp64"): Likewise.
24475 ("movdi_pic_gotdata_op"): Likewise.
24476 ("*movsf_insn"): Likewise.
24477 ("*movdf_insn_sp32"): Likewise.
24478 ("*movdf_insn_sp64"): Likewise.
24479 ("*zero_extendhisi2_insn"): Likewise.
24480 ("*zero_extendqihi2_insn"): Likewise.
24481 ("*zero_extendqisi2_insn"): Likewise.
24482 ("*zero_extendqidi2_insn"): Likewise.
24483 ("*zero_extendhidi2_insn"): Likewise.
24484 ("*zero_extendsidi2_insn_sp64"): Likewise.
24485 ("ldfsr"): Likewise.
24486 ("prefetch_64"): Likewise.
24487 ("prefetch_32"): Likewise.
24488 ("tie_ld32"): Likewise.
24489 ("tie_ld64"): Likewise.
24490 ("*tldo_ldub_sp32"): Likewise.
24491 ("*tldo_ldub1_sp32"): Likewise.
24492 ("*tldo_ldub2_sp32"): Likewise.
24493 ("*tldo_ldub_sp64"): Likewise.
24494 ("*tldo_ldub1_sp64"): Likewise.
24495 ("*tldo_ldub2_sp64"): Likewise.
24496 ("*tldo_ldub3_sp64"): Likewise.
24497 ("*tldo_lduh_sp32"): Likewise.
24498 ("*tldo_lduh1_sp32"): Likewise.
24499 ("*tldo_lduh_sp64"): Likewise.
24500 ("*tldo_lduh1_sp64"): Likewise.
24501 ("*tldo_lduh2_sp64"): Likewise.
24502 ("*tldo_lduw_sp32"): Likewise.
24503 ("*tldo_lduw_sp64"): Likewise.
24504 ("*tldo_lduw1_sp64"): Likewise.
24505 ("*tldo_ldx_sp64"): Likewise.
24506 ("*mov<VM32:mode>_insn"): Likewise.
24507 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24508 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24510 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24512 * config/sparc/sparc.md ("type"): New insn type viscmp.
24513 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24515 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24516 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24517 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24518 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24520 ("n7_vis_logical_11cycle"): Likewise.
24521 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24522 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24523 * config/sparc/niagara.md ("niag_vis"): Likewise.
24524 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24525 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24527 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24529 * config/sparc/sparc.md: New instruction type `bmask'.
24530 (bmaskdi_vis): Use the `bmask' type.
24531 (bmasksi_vis): Likewise.
24532 * config/sparc/ultra3.md (us3_array): Likewise.
24533 * config/sparc/niagara7.md (n7_array): Likewise.
24534 * config/sparc/niagara4.md (n4_array): Likewise.
24535 * config/sparc/niagara2.md (niag2_vis): Likewise.
24536 (niag3_vis): Likewise.
24537 * config/sparc/niagara.md (niag_vis): Likewise.
24539 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24541 * ipa-comdats.c: Remove optimize check from gate.
24542 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24543 for functions not optimized.
24544 (ipa_fn_summary_read): Skip optimize check.
24545 (ipa_fn_summary_write): Likewise.
24546 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24548 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24550 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24551 (check_callers): Give up on caller which is not optimized.
24552 (inline_small_functions): Likewise.
24553 (ipa_inline): Do not give up when not optimizing.
24554 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24555 away unoptimizes cdtors.
24556 (whole_program_function_and_variable_visibility): Do
24557 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24558 * ipa.c (process_references): Do not check optimize.
24559 (symbol_table::remove_unreachable_nodes): Update optimize check.
24560 (set_writeonly_bit): Update optimize check.
24561 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24562 (pass_ipa_single_use::gate): Remove.
24564 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24566 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24567 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24568 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24569 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24570 permute_load, permute_store, adjust_extract, adjust_splat,
24571 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24572 replace_swap_with_copy, dump_swap_insn_table,
24573 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24574 recombine_lvx_pattern, recombine_stvx_pattern,
24575 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24576 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24577 to file rs6000-p8swap.c.
24578 * config/rs6000/rs6000-p8swap.c: New file.
24579 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24580 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24581 and rs6000*-*-* targets.
24583 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24585 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24587 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24589 * lto-wrapper.c (merge_and_complain): Do not merge
24590 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24591 fsigned_zeros, ftrapping_math, fwrapv.
24592 (append_compiler_options): Do not track these options.
24593 (append_linker_options): Likewie
24595 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24597 * cgraphunit.c (cgraph_node::finalize_function): When
24598 !flag_toplevel_reorde set no_reorder flag.
24599 (varpool_node::finalize_decl): Likewise.
24600 (symbol_table::compile): Drop no toplevel reorder path.
24602 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24604 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24605 edges; zero probability is not better than uninitialized.
24607 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24609 * asan.h (asan_sanitize_allocas_p): Declare.
24610 * asan.c (asan_sanitize_allocas_p): New function.
24611 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24612 (handle_builtin_alloca): Likewise.
24613 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24614 if !asan_sanitize_allocas_p.
24615 * params.def (asan-instrument-allocas): Add new option.
24616 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24617 * opts.c (common_handle_option): Disable allocas sanitization for
24620 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24622 * asan.c: Include gimple-fold.h.
24623 (get_last_alloca_addr): New function.
24624 (handle_builtin_stackrestore): Likewise.
24625 (handle_builtin_alloca): Likewise.
24626 (asan_emit_allocas_unpoison): Likewise.
24627 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24628 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24629 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24630 (instrument_builtin_call): Pass gimple iterator to
24631 get_mem_refs_of_builtin_call.
24632 (last_alloca_addr): New global.
24633 * asan.h (asan_emit_allocas_unpoison): Declare.
24634 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24635 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24636 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24637 if function calls alloca.
24638 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24639 * gimple-fold.h (replace_call_with_value): Declare.
24640 * internal-fn.c: Include asan.h.
24641 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24642 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24644 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24646 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24647 (C_SELFTEST_FLAGS): New.
24648 (CPP_SELFTEST_FLAGS): New.
24649 (SELFTEST_DEPS): New, from deps of s-selftest.
24650 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24651 (CPP_SELFTEST_DEPS): New.
24652 (selftest): Add dependency on s-selftest-c++.
24653 (s-selftest): Rename to...
24654 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24655 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24656 than SELFTEST_FLAGS.
24657 (selftest-gdb): Rename to...
24658 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24660 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24661 (selftest-valgrind): Rename to...
24662 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24664 (selftest-valgrind): Reintroduce as an alias for
24665 selftest-c-valgrind.
24666 (s-selftest-c++): New.
24667 (selftest-c++-gdb): New.
24668 (selftest-c++-valgrind): New.
24670 2017-07-06 Olivier Hainque <hainque@adacore.com>
24672 * gcc.c (process_command): When deciding if undefined variables
24673 should be ignored when processing specs, accept "gcc -v" as well.
24675 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24677 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24678 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24680 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24682 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24683 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24684 * config/arm/arm-tables.opt: Regenerate.
24685 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24687 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24689 2017-07-06 Carl Love <cel@us.ibm.com>
24691 * ChangeLog: Clean up from mid air collision
24693 2017-07-06 Carl Love <cel@us.ibm.com>
24695 * config/rs6000/rs6000-c.c: Add support for built-in functions
24696 vector signed int vec_subc (vector signed int, vector signed int);
24697 vector signed __int128 vec_subc (vector signed __int128,
24698 vector signed __int128);
24699 vector unsigned __int128 vec_subc (vector unsigned __int128,
24700 vector unsigned __int128);
24701 vector signed int vec_sube (vector signed int, vector signed int,
24702 vector signed int);
24703 vector unsigned int vec_sube (vector unsigned int,
24704 vector unsigned int,
24705 vector unsigned int);
24706 vector signed __int128 vec_sube (vector signed __int128,
24707 vector signed __int128,
24708 vector signed__int128);
24709 vector unsigned __int128 vec_sube (vector unsigned __int128,
24710 vector unsigned __int128,
24711 vector unsigned __int128);
24712 vector signed int vec_subec (vector signed int, vector signed int,
24713 vector signed int);
24714 vector unsigned int vec_subec (vector unsigned int,
24715 vector unsigned int,
24716 vector unsigned int);
24717 vector signed __int128 vec_subec (vector signed __int128,
24718 vector signed __int128,
24719 vector signed__int128);
24720 vector unsigned __int128 vec_subec (vector unsigned __int128,
24721 vector unsigned __int128,
24722 vector unsigned __int128);
24723 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24724 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24725 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24726 BU_ALTIVEC_OVERLOAD_X definitions.
24727 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24728 * doc/extend.texi: Update the built-in documentation file for the new
24729 built-in functions.
24731 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24734 * diagnostic-show-locus.c (layout::layout): Use start and finish
24735 spelling location for the start and finish of each range.
24736 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24737 Add unused aspect param.
24738 * input.c (expand_location_1): Add "aspect" param, and use it
24739 to access the correct part of the location.
24740 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24742 (expand_location_to_spelling_point): Likewise.
24743 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24744 param, and pass it to expand_location_1.
24746 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24748 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24749 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24750 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24751 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24752 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24753 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24754 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24755 _mm_maskz_getmant_ss): New intrinsics.
24756 (__builtin_ia32_getexpss128_mask): Changed to ...
24757 __builtin_ia32_getexpss128_round ... this.
24758 (__builtin_ia32_getexpsd128_mask): Changed to ...
24759 __builtin_ia32_getexpsd128_round ... this.
24760 * config/i386/i386-builtin-types.def
24761 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24762 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24763 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24764 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24765 __builtin_ia32_getmantss_mask_round): New builtins.
24766 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24767 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24768 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24769 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24770 * config/i386/sse.md
24771 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24772 avx512f_sgetexp<mode><mask_scalar_name>
24773 <round_saeonly_scalar_name> ... this.
24774 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24775 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24776 vgetexp<ssescalarmodesuffix>
24777 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24778 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24779 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24780 avx512f_vgetmant<mode><mask_scalar_name>
24781 <round_saeonly_scalar_name> ... this.
24782 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24783 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24784 vgetmant<ssescalarmodesuffix>
24785 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24786 %0<mask_scalar_operand4>, %1, %2
24787 <round_saeonly_scalar_mask_op4>, %3} ... this.
24788 * config/i386/subst.md (mask_scalar_operand4,
24789 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24790 round_saeonly_scalar_nimm_predicate): New subst attributes.
24792 2017-07-06 Julia Koval <julia.koval@intel.com>
24794 * config/i386/i386.c (ix86_erase_embedded_rounding):
24795 Remove code for old rounding pattern.
24797 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24799 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24801 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24803 * doc/sourcebuild.texi (Test Directives, Variants of
24804 dg-require-support): Add documentation for dg-require-stack-check.
24806 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24808 * config/i386/subst.md (mask_scalar, round_scalar,
24809 round_saeonly_scalar): New meta-templates.
24810 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24811 round_scalar_mask_operand3, round_scalar_mask_op3,
24812 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24813 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24814 round_saeonly_scalar_constraint,
24815 round_saeonly_scalar_prefix): New subst attribute.
24816 * config/i386/sse.md
24817 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24818 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24819 <round_scalar_name> ... this.
24820 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24821 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24822 <round_scalar_name> ... this.
24823 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24824 <sse>_vm<code><mode>3<mask_scalar_name>
24825 <round_saeonly_scalar_name> ... this.
24826 (v<plusminus_mnemonic><ssescalarmodesuffix>
24827 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24828 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24829 v<plusminus_mnemonic><ssescalarmodesuffix>
24830 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24831 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24832 (v<multdiv_mnemonic><ssescalarmodesuffix>
24833 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24834 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24835 v<multdiv_mnemonic><ssescalarmodesuffix>
24836 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24837 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24838 (v<maxmin_float><ssescalarmodesuffix>
24839 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24840 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24841 v<maxmin_float><ssescalarmodesuffix>
24842 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24843 %0<mask_scalar_operand3>, %1, %<iptr>2
24844 <round_saeonly_scalar_mask_op3>} ... this.
24846 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24848 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24849 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24851 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24852 Alan Hayward <alan.hayward@arm.com>
24853 David Sherwood <david.sherwood@arm.com>
24855 * combine.c (simplify_if_then_else): Remove "enum" before
24857 * compare-elim.c (can_eliminate_compare): Likewise.
24858 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24860 (aarch64_lookup_simd_builtin_type): Likewise.
24861 (aarch64_simd_builtin_type): Likewise.
24862 (aarch64_init_simd_builtin_types): Likewise.
24863 (aarch64_simd_expand_args): Likewise.
24864 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24866 (aarch64_reverse_mask): Likewise.
24867 (aarch64_simd_emit_reg_reg_move): Likewise.
24868 (aarch64_gen_adjusted_ldpstp): Likewise.
24869 (aarch64_ccmp_mode_to_code): Likewise.
24870 (aarch64_operands_ok_for_ldpstp): Likewise.
24871 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24872 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24874 (aarch64_min_divisions_for_recip_mul): Likewise.
24875 (aarch64_reassociation_width): Likewise.
24876 (aarch64_get_condition_code_1): Likewise.
24877 (aarch64_simd_emit_reg_reg_move): Likewise.
24878 (aarch64_simd_attr_length_rglist): Likewise.
24879 (aarch64_reverse_mask): Likewise.
24880 (aarch64_operands_ok_for_ldpstp): Likewise.
24881 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24882 (aarch64_gen_adjusted_ldpstp): Likewise.
24883 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24885 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24886 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24887 (arm_lookup_simd_builtin_type): Likewise.
24888 (arm_simd_builtin_type): Likewise.
24889 (arm_init_simd_builtin_types): Likewise.
24890 (arm_expand_builtin_args): Likewise.
24891 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24892 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24893 (ft32_setup_incoming_varargs): Likewise.
24894 (ft32_function_arg): Likewise.
24895 (ft32_function_arg_advance): Likewise.
24896 (ft32_pass_by_reference): Likewise.
24897 (ft32_arg_partial_bytes): Likewise.
24898 (ft32_valid_pointer_mode): Likewise.
24899 (ft32_addr_space_pointer_mode): Likewise.
24900 (ft32_addr_space_legitimate_address_p): Likewise.
24901 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24903 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24904 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24905 (iamcu_alignment): Likewise.
24906 (canonicalize_vector_int_perm): Likewise.
24907 (ix86_noce_conversion_profitable_p): Likewise.
24908 (ix86_mpx_bound_mode): Likewise.
24909 (ix86_operands_ok_for_move_multiple): Likewise.
24910 * config/microblaze/microblaze-protos.h
24911 (microblaze_expand_conditional_branch_reg): Likewise.
24912 * config/microblaze/microblaze.c
24913 (microblaze_expand_conditional_branch_reg): Likewise.
24914 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24916 (rs6000_reassociation_width): Likewise.
24917 (rs6000_invalid_binary_op): Likewise.
24918 (fusion_p9_p): Likewise.
24919 (emit_fusion_p9_load): Likewise.
24920 (emit_fusion_p9_store): Likewise.
24921 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24923 (riscv_hard_regno_mode_ok_p): Likewise.
24924 (riscv_address_insns): Likewise.
24925 (riscv_split_symbol): Likewise.
24926 (riscv_legitimize_move): Likewise.
24927 (riscv_function_value): Likewise.
24928 (riscv_hard_regno_nregs): Likewise.
24929 (riscv_expand_builtin): Likewise.
24930 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24931 (riscv_build_integer): Likewise.
24932 (riscv_split_integer): Likewise.
24933 (riscv_legitimate_constant_p): Likewise.
24934 (riscv_cannot_force_const_mem): Likewise.
24935 (riscv_regno_mode_ok_for_base_p): Likewise.
24936 (riscv_valid_base_register_p): Likewise.
24937 (riscv_valid_offset_p): Likewise.
24938 (riscv_valid_lo_sum_p): Likewise.
24939 (riscv_classify_address): Likewise.
24940 (riscv_legitimate_address_p): Likewise.
24941 (riscv_address_insns): Likewise.
24942 (riscv_load_store_insns): Likewise.
24943 (riscv_force_binary): Likewise.
24944 (riscv_split_symbol): Likewise.
24945 (riscv_force_address): Likewise.
24946 (riscv_legitimize_address): Likewise.
24947 (riscv_move_integer): Likewise.
24948 (riscv_legitimize_const_move): Likewise.
24949 (riscv_legitimize_move): Likewise.
24950 (riscv_address_cost): Likewise.
24951 (riscv_subword): Likewise.
24952 (riscv_output_move): Likewise.
24953 (riscv_canonicalize_int_order_test): Likewise.
24954 (riscv_emit_int_order_test): Likewise.
24955 (riscv_function_arg_boundary): Likewise.
24956 (riscv_pass_mode_in_fpr_p): Likewise.
24957 (riscv_pass_fpr_single): Likewise.
24958 (riscv_pass_fpr_pair): Likewise.
24959 (riscv_get_arg_info): Likewise.
24960 (riscv_function_arg): Likewise.
24961 (riscv_function_arg_advance): Likewise.
24962 (riscv_arg_partial_bytes): Likewise.
24963 (riscv_function_value): Likewise.
24964 (riscv_pass_by_reference): Likewise.
24965 (riscv_setup_incoming_varargs): Likewise.
24966 (riscv_print_operand): Likewise.
24967 (riscv_elf_select_rtx_section): Likewise.
24968 (riscv_save_restore_reg): Likewise.
24969 (riscv_for_each_saved_reg): Likewise.
24970 (riscv_register_move_cost): Likewise.
24971 (riscv_hard_regno_mode_ok_p): Likewise.
24972 (riscv_hard_regno_nregs): Likewise.
24973 (riscv_class_max_nregs): Likewise.
24974 (riscv_memory_move_cost): Likewise.
24975 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24976 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24977 (rl78_addr_space_address_mode): Likewise.
24978 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24980 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24981 (rs6000_reassociation_width): Likewise.
24982 (rs6000_invalid_binary_op): Likewise.
24983 (fusion_p9_p): Likewise.
24984 (emit_fusion_p9_load): Likewise.
24985 (emit_fusion_p9_store): Likewise.
24986 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24987 (ok_for_simple_move_operands): Likewise.
24988 (ok_for_simple_move_strict_operands): Likewise.
24989 (ok_for_simple_arith_logic_operands): Likewise.
24990 (visium_legitimize_reload_address): Likewise.
24991 (visium_select_cc_mode): Likewise.
24992 (output_cbranch): Likewise.
24993 (visium_split_double_move): Likewise.
24994 (visium_expand_copysign): Likewise.
24995 (visium_expand_int_cstore): Likewise.
24996 (visium_expand_fp_cstore): Likewise.
24997 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24998 (visium_function_arg): Likewise.
24999 (visium_function_arg_advance): Likewise.
25000 (visium_libcall_value): Likewise.
25001 (visium_setup_incoming_varargs): Likewise.
25002 (visium_legitimate_constant_p): Likewise.
25003 (visium_legitimate_address_p): Likewise.
25004 (visium_legitimize_address): Likewise.
25005 (visium_secondary_reload): Likewise.
25006 (visium_register_move_cost): Likewise.
25007 (visium_memory_move_cost): Likewise.
25008 (prepare_move_operands): Likewise.
25009 (ok_for_simple_move_operands): Likewise.
25010 (ok_for_simple_move_strict_operands): Likewise.
25011 (ok_for_simple_arith_logic_operands): Likewise.
25012 (visium_function_value_1): Likewise.
25013 (rtx_ok_for_offset_p): Likewise.
25014 (visium_legitimize_reload_address): Likewise.
25015 (visium_split_double_move): Likewise.
25016 (visium_expand_copysign): Likewise.
25017 (visium_expand_int_cstore): Likewise.
25018 (visium_expand_fp_cstore): Likewise.
25019 (visium_split_cstore): Likewise.
25020 (visium_select_cc_mode): Likewise.
25021 (visium_split_cbranch): Likewise.
25022 (output_cbranch): Likewise.
25023 (visium_print_operand_address): Likewise.
25024 * expmed.c (flip_storage_order): Likewise.
25025 * expmed.h (emit_cstore): Likewise.
25026 (flip_storage_order): Likewise.
25027 * genrecog.c (validate_pattern): Likewise.
25028 * hsa-gen.c (gen_hsa_addr): Likewise.
25029 * internal-fn.c (expand_arith_overflow): Likewise.
25030 * ira-color.c (allocno_copy_cost_saving): Likewise.
25031 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25032 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25033 (process_invariant_for_inheritance): Likewise.
25034 * lra-eliminations.c (move_plus_up): Likewise.
25035 * omp-low.c (lower_oacc_reductions): Likewise.
25036 * simplify-rtx.c (simplify_subreg): Likewise.
25037 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25038 (TARGET_CHKP_BOUND_MODE): Likewise..
25039 * targhooks.c (default_chkp_bound_mode): Likewise.
25040 (default_setup_incoming_vararg_bounds): Likewise.
25041 * targhooks.h (default_chkp_bound_mode): Likewise.
25042 (default_setup_incoming_vararg_bounds): Likewise.
25043 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25044 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25045 (have_whole_vector_shift): Likewise.
25046 * tree-vect-stmts.c (vectorizable_load): Likewise.
25047 * doc/tm.texi: Regenerate.
25049 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25051 Graceful degrade if Binutils PR21472 is not available.
25054 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25055 .rodata in flash test fails.
25056 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25057 * confgure: Regenerate.
25058 * config.in: Regenerate.
25059 * config/avr/avr.c (avr_asm_named_section)
25060 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25061 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25062 (avr_asm_init_sections): Same.
25064 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25066 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25067 (fma<VH:mode>4_intrinsic): Likewise.
25068 (*fmsub<VCVTF:mode>4): Likewise.
25069 (*fmsub<VH:mode>4_intrinsic): Likewise.
25071 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25074 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25075 Don't depend on "optimize > 0".
25076 (out_movhi_r_mr, out_movqi_mr_r): Same.
25077 (out_movhi_mr_r, out_movqi_r_mr): Same.
25078 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25079 io_address_operand on "optimize > 0".
25081 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25083 * tree-loop-distribution.c: Add general explanantion on the pass.
25084 (generate_loops_for_partition): Mark distributed loop.
25085 (pg_add_dependence_edges): New parameter. Handle alias data
25086 dependence specially and record it in the parameter if asked.
25087 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25088 (init_partition_graph_vertices, add_partition_graph_edge): New.
25089 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25090 (free_partition_graph_vdata, build_partition_graph): New.
25091 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25092 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25093 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25094 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25095 (version_for_distribution_p, finalize_partitions): New.
25096 (distribute_loop): Handle alias data dependence specially. Factor
25097 out loop fusion code as functions and call these functions.
25099 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25101 * tree-loop-distribution.c (classify_partition): New parameter and
25102 better handle reduction statement.
25103 (rdg_build_partitions): Revise comment.
25104 (distribute_loop): Compute statements in all partitions and pass it
25105 to classify_partition.
25107 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25109 * tree-loop-distribution.c (enum partition_type): New.
25110 (struct partition): New field type.
25111 (partition_merge_into): Add parameter. Update partition type.
25112 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25113 (build_rdg_partition_for_vertex): Compute partition type.
25114 (rdg_build_partitions): Dump partition type.
25115 (distribute_loop): Update calls to partition_merge_into.
25117 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25119 * tree-loop-distribution.c (struct ddr_hasher): New.
25120 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25122 (classify_partition): Call get_data_dependence.
25123 (pg_add_dependence_edges): Ditto.
25124 (distribute_loop): Release data dependence hash table.
25126 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25128 * tree-loop-distribution.c (ref_base_address): Delete.
25129 (similar_memory_accesses): Rename ...
25130 (share_memory_accesses): ... to this. Check if partitions access
25131 the same memory reference.
25132 (distribute_loop): Call share_memory_accesses.
25134 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25136 * tree-loop-distribution.c (struct partition): New field recording
25137 its data reference.
25138 (partition_alloc, partition_free): Init and release data refs.
25139 (partition_merge_into): Merge data refs.
25140 (build_rdg_partition_for_vertex): Collect data refs for partition.
25141 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25143 (distribute_loop): Remve data refs from vertice data of partition
25146 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25148 * tree-loop-distribution.c (params.h): Include header file.
25149 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25150 (datarefs_vec): New global var.
25151 (create_rdg_vertices): Use datarefs_vec directly.
25152 (free_rdg): Don't free data references.
25153 (build_rdg): Update use. Don't free data references.
25154 (distribute_loop): Compute global variable for data references.
25155 Bail out if there are too many data references.
25157 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25159 * tree-loop-distribution.c (loop_nest): New global var.
25160 (build_rdg): Use loop directly, rather than loop nest.
25161 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25163 (distribute_loop): Compute global variable loop nest. Update use.
25165 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25167 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25168 (partition_merge_into): New parameter. Dump reason for fusion.
25169 (distribute_loop): Update use of partition_merge_into.
25171 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25173 * tree-loop-distribution.c (bb_top_order_index): New.
25174 (bb_top_order_index_size, bb_top_order_cmp): New.
25175 (stmts_from_loop): Use topological order.
25176 (pass_loop_distribution::execute): Compute and release topological
25177 order for basic blocks.
25179 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25181 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25184 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25186 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25187 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25188 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25189 * internal-fn.def (LOOP_DIST_ALIAS): New.
25190 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25191 (fold_loop_internal_call): ... this.
25192 (vect_loop_dist_alias_call): New function.
25193 (set_uid_loop_bbs): Call fold_loop_internal_call.
25194 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25197 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25200 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25201 Require dead FLAGS_REG at the beginning of a peephole.
25203 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25206 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25207 arguments in the call to __builtin_ia32_sbb_u32.
25208 (_subborrow_u64): Swap _X and _Y arguments in the call to
25209 __builtin_ia32_sbb_u64.
25211 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25214 * tree-vrp.c (compare_assert_loc): Turn into a function template
25215 with stable template parameter. Only test if a->e is NULL,
25216 !a->e == !b->e has been verified already. Use e == NULL or
25217 e != NULL instead of e or ! e tests. If stable is true, don't use
25218 iterative_hash_expr, on the other side allow a or b or both NULL
25219 and sort the NULLs last.
25220 (process_assert_insertions): Sort using compare_assert_loc<false>
25221 instead of compare_assert_loc, later sort using
25222 compare_assert_loc<true> before calling process_assert_insertions_for
25223 in a loop. Use break instead of continue once seen NULL pointer.
25225 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25227 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25228 Cortex-R7 and Cortex-R8 processors.
25230 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25232 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25233 uninitialized while src is not.
25235 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25237 * common/config/arm/arm-common.c: Adjust include path for
25239 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25240 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25241 (arm-cpu-data.h): Likewise.
25242 (arm-cpu-cdata.h): Likewise.
25243 * config/arm/arm-cpu.h: Delete.
25244 * config/arm/arm-cpu-cdata.h: Delete.
25245 * config/arm/arm-cpu-data.h: Delete.
25247 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25249 * config/arm/arm-cpus.in (cortex-a55): New.
25250 (cortex-a75): Likewise.
25251 (cortex-a75.cortex-a55): Likewise.
25252 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25254 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25255 * config/arm/arm-cpu-cdata.h: Regenerate.
25256 * config/arm/arm-cpu-data.h: Regenerate.
25257 * config/arm/arm-cpu.h: Regenerate.
25258 * config/arm/arm-tables.opt: Regenerate.
25259 * config/arm/arm-tune.md: Regenerate.
25261 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25263 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25265 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25267 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25270 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25272 PR tree-optimization/81292
25273 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25274 full_string_p, also call adjust_related_strinfos if the adjustment
25275 is simple, otherwise invalidate related strinfos.
25277 2017-07-04 Martin Liska <mliska@suse.cz>
25280 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25281 newly created variable as DECL_IGNORED_P.
25283 2017-07-04 Martin Liska <mliska@suse.cz>
25286 * ipa-inline.c (inline_small_functions):
25287 Use xstrdup_for_dump.
25289 2017-07-04 Tom de Vries <tom@codesourcery.com>
25291 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25293 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25296 * config/darwin.c (darwin_function_switched_text_sections):
25299 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25301 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25303 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25305 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25307 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25309 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25310 min_profitable_iters, and th as inclusive lower bounds.
25311 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25312 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25313 for min_profitable_iters and min_profitable_estimate.
25314 (vect_transform_loop): Treat th as an inclusive lower bound.
25315 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25317 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25320 * config/darwin.c (darwin_function_switched_text_sections):
25321 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25322 in two pieces, and suppress the use of buf.
25324 2017-07-03 Nathan Sidwell <nathan@acm.org>
25326 * hash-table.h (hash_table_mod1): Fix indentation.
25328 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25330 PR middle-end/81290
25331 * predict.c (force_edge_cold): Be more careful about propagation
25333 * profile-count.h (profile_probability::guessed,
25334 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25336 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25338 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25340 * doc/invoke.texi (rcpc architecture extension): Document it.
25342 2017-07-03 Richard Biener <rguenther@suse.de>
25344 PR tree-optimization/60510
25345 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25346 the scalar reduction PHI and use it.
25347 (vectorizable_reduction): Properly guard the single_defuse_cycle
25348 path for non-SLP reduction chains where we cannot use it.
25349 Rework reduc_def/index and vector type deduction. Rework
25350 vector operand gathering during reduction op code-gen.
25351 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25352 chains dissolve the chain and leave it to non-SLP reduction
25355 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25357 * tree-data-ref.h (dr_alignment): Declare.
25358 * tree-data-ref.c (dr_alignment): New function.
25359 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25360 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25362 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25364 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25366 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25367 and base_misalignment fields.
25368 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25369 * tree-data-ref.c: Include builtins.h.
25370 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25371 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25372 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25373 * tree-vect-data-refs.c: Include tree-cfg.h.
25374 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25375 fields instead of calculating an alignment here.
25376 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25377 innermost_loop_behavior fields.
25379 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25381 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25383 (DR_STEP_ALIGNMENT): New macro.
25384 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25385 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25386 (create_data_ref): Print it.
25387 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25388 to tell whether the step preserves vector (mis)alignment.
25389 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25390 Move the check for an integer step and generalise to all INTEGER_CST.
25391 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25392 Print the outer step alignment.
25394 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25396 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25397 with offset_alignment.
25398 (DR_ALIGNED_TO): Delete.
25399 (DR_OFFSET_ALIGNMENT): New macro.
25400 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25401 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25402 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25403 (create_data_ref): Likewise.
25404 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25405 (vect_analyze_data_refs): Likewise.
25406 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25407 creating dummy innermost behavior.
25409 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25411 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25412 with a "innermost_loop_behavior *" and refeence tree.
25413 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25414 (create_data_ref): Update call accordingly.
25415 * tree-predcom.c (find_looparound_phi): Likewise.
25417 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25419 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25420 fields with dr_wrt_vec_loop.
25421 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25422 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25423 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25424 (vect_dr_behavior): New function.
25425 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25426 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25427 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25428 track whether the step preserves the misalignment.
25429 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25430 Use vect_dr_behavior.
25431 (vect_setup_realignment): Update call accordingly.
25432 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25433 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25434 call to vect_create_addr_base_for_vector_ref.
25435 (vect_create_cond_for_align_checks): Likewise.
25436 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25437 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25438 (vect_recog_mask_conversion_pattern): Likewise.
25439 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25440 (new_stmt_vec_info): Remove redundant zeroing.
25442 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25444 * common/config/arm/arm-common.c (arm_be8_option): New function.
25445 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25446 (ISA_ARMv6): Add isa_bit_be8.
25447 * config/arm/arm.h (arm_be8_option): Add prototype.
25448 (BE8_SPEC_FUNCTION): New define.
25449 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25450 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25451 (mlittle-endian): Similarly.
25452 (mbe8, mbe32): New options.
25453 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25454 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25456 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25458 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25460 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25462 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25463 (cleanup_tree_cfg_bb): Use it.
25464 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25466 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25468 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25471 * loop-doloop.c (add_test): Update profile.
25473 2017-07-03 Martin Liska <mliska@suse.cz>
25476 * sanopt.c (rewrite_usage_of_param): New function.
25477 (sanitize_rewrite_addressable_params): Likewise.
25478 (pass_sanopt::execute): Call rewrite_usage_of_param.
25480 2017-07-03 Richard Biener <rguenther@suse.de>
25482 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25483 back to using VIEW_CONVERT_EXPR.
25485 2017-07-03 Martin Liska <mliska@suse.cz>
25488 * doc/extend.texi: Document when a resolver function is
25489 generated for target_clones.
25491 2017-07-03 Martin Liska <mliska@suse.cz>
25493 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25494 and shadow memory of auto variables which are subject of
25495 use-after-scope sanitization.
25496 (asan_expand_mark_ifn): Add do set only when is_poison.
25498 2016-07-03 Richard Biener <rguenther@suse.de>
25500 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25502 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25503 (vectorizable_reduction): Perform reduction PHI creation when
25504 visiting a reduction PHI and adjust and simplify code generation
25505 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25506 (vect_transform_loop): Visit reduction PHIs.
25507 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25508 defs into the SLP tree.
25509 (vect_build_slp_tree): Reduction defs terminate the recursion.
25510 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25512 (vect_get_vec_defs_for_stmt_copy): Export.
25513 (vect_get_vec_defs): Likewise.
25514 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25516 (vect_get_vec_defs_for_stmt_copy): Declare.
25517 (vect_get_vec_defs): Likewise.
25519 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25521 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25522 parameter with a "loop" parameter and use it instead of the
25523 loop containing DR_STMT. Don't check simple_iv when doing
25524 BB analysis. Describe the two analysis modes in the comment.
25526 2017-07-03 Tom de Vries <tom@codesourcery.com>
25528 PR tree-optimization/69468
25529 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25530 (find_same_succ_bb): Handle ignore_edge_flags.
25532 2017-07-03 Tom de Vries <tom@codesourcery.com>
25534 PR tree-optimization/81192
25535 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25537 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25539 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25541 2017-07-03 Tom de Vries <tom@codesourcery.com>
25543 PR tree-optimization/81192
25544 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25545 BB_SAME_SUCC (bb) == NULL.
25547 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25549 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25552 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25554 * dumpfile.c: Include profile-count.h
25555 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25557 (insert_cond_bb): Update profile.
25558 * tree-cfg.h (insert_cond_bb): Update prototype.
25559 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25560 * tree-dump.c: Do not include tree-cfg.
25562 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25564 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25566 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25568 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25571 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25573 * tree-complex.c (expand_complex_div_wide): update profile.
25575 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25576 Alan Hayward <alan.hayward@arm.com>
25577 David Sherwood <david.sherwood@arm.com>
25579 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25580 (CORETYPES_H): New define.
25581 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25582 (insn-modes-inline.h, s-modes-inline-h): New rules.
25583 (generated_files): Add insn-modes-inline.h.
25584 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25585 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25586 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25587 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25588 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25589 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25590 (build/gencodes.o, build/genconditions.o): Likewise.
25591 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25592 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25593 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25594 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25595 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25596 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25597 * coretypes.h: Include everything up to real.h for generators.
25598 Include insn-modes.h first. Include wide-int-print.h after
25599 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25600 * machmode.h: Don't include insn-modes.h here.
25601 * function-tests.c: Remove includes of signop.h, machmode.h,
25602 double-int.h and wide-int.h.
25604 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25606 * optc-save-gen.awk: Likewise.
25607 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25608 * godump.c: Remove include of wide-int-print.h.
25609 * pretty-print.h: Likewise.
25610 * wide-int-print.cc: Likewise.
25611 * wide-int.cc: Likewise.
25612 * hash-map-tests.c: Remove include of signop.h.
25613 * hash-set-tests.c: Likewise.
25614 * rtl-tests.c: Likewise.
25615 * mkconfig.sh: Remove include of machmode.h.
25616 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25618 (emit_insn_modes_inline_h): ...this new function. Emit the code
25619 into an insn-modes-inline.h header file, adding appropriate
25620 include guards and end comments.
25621 (emit_insn_modes_c_header): Remove include of machmode.h.
25622 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25624 (main): Handle -i flag and call emit_insn_modes_inline_h when
25627 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25629 * tree-ssa-strlen.c (strinfo): Rename the length field to
25630 nonzero_chars. Add a full_string_p field.
25631 (compare_nonzero_chars, zero_length_string_p): New functions.
25632 (get_addr_stridx): Add an offset_out parameter.
25633 Use compare_nonzero_chars.
25634 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25635 (new_strinfo): Update after above changes to strinfo.
25636 (set_endptr_and_length): Set full_string_p.
25637 (get_string_length): Update after above changes to strinfo.
25638 (unshare_strinfo): Update call to new_strinfo.
25639 (maybe_invalidate): Likewise.
25640 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25641 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25642 as a uhwi instead of an shwi. Update after above changes to
25643 strinfo and new_strinfo.
25644 (zero_length_string): Assert that chainsi contains full strings.
25645 Use zero_length_string_p. Update call to new_strinfo.
25646 (adjust_related_strinfos): Update after above changes to strinfo.
25647 Copy full_string_p from origsi.
25648 (adjust_last_stmt): Use zero_length_string_p.
25649 (handle_builtin_strlen): Update after above changes to strinfo and
25650 new_strinfo. Install the lhs as the string length if the previous
25651 entry didn't describe a full string.
25652 (handle_builtin_strchr): Update after above changes to strinfo
25654 (handle_builtin_strcpy): Likewise.
25655 (handle_builtin_strcat): Likewise.
25656 (handle_builtin_malloc): Likewise.
25657 (handle_pointer_plus): Likewise.
25658 (handle_builtin_memcpy): Likewise. Track nonzero characters
25659 that aren't necessarily followed by a nul terminator.
25660 (handle_char_store): Likewise.
25662 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25664 PR tree-optimization/80769
25665 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25666 for malloc and calloc. Document the new invariant that all related
25667 strinfos have delayed lengths or none do.
25668 (verify_related_strinfos): Move earlier in file.
25669 (set_endptr_and_length): New function, split out from...
25670 (get_string_length): ...here. Also set the lengths of related
25672 (zero_length_string): Assert that chainsi has known (rather than
25674 (adjust_related_strinfos): Likewise.
25676 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25678 PR tree-optimization/81136
25679 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25680 assert that two references with the same misalignment have the same
25681 compile-time misalignment if those compile-time misalignments
25684 2017-07-01 Andi Kleen <ak@linux.intel.com>
25686 * print-tree.c (print_node): Print all attributes.
25688 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25690 * cfg.c (scale_bbs_frequencies): New function.
25691 * cfg.h (scale_bbs_frequencies): Declare it.
25692 * cfgloopanal.c (single_likely_exit): Cleanup.
25693 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25695 (scale_loop_profile): Likewise.
25696 (loop_version): Likewise.
25697 (create_empty_loop_on_edge): Update.
25698 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25699 scale_loop_frequencies, scale_loop_profile, loopify,
25700 loop_version): Update prototypes.
25701 * modulo-sched.c (sms_schedule): Update.
25702 * predict.c (unlikely_executed_edge_p): Also check probability.
25703 (probably_never_executed_edge_p): Fix typo.
25704 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25705 * tree-parloops.c (gen_parallel_loop): Update.
25706 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25707 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25708 * tree-ssa-loop-split.c (split_loop): Update.
25709 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25710 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25711 (vect_loop_versioning): Update.
25712 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25714 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25716 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25718 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25720 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25721 to keep profile consistent.
25723 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25725 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25726 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25727 * profile-count.h (max_safe_multiplier): Make unsigned.
25728 (profile_count::guessed_zero): New.
25730 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25732 * bb-reorder.c (fix_up_crossing_landing_pad,
25733 fix_crossing_conditional_branches): Use make_single_succ_edge
25734 to keep profile consistent.
25736 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25738 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25741 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25744 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25745 the right scopes, make sure cond_jump isn't preserved between multiple
25746 iterations. Search for fallthru edge whenever there are 3+ edges and
25747 use find_fallthru_edge for it.
25749 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25751 Patch by Alexander Monakov <amonakov@ispras.ru>
25752 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25753 probabilities consistently.
25755 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25757 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25758 emit_cmp_and_jump_insns.
25760 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25763 * tree-inline.c (expand_call_inline): Combine profile statuses.
25765 2017-06-30 Andrew Pinski <apinski@cavium.com>
25767 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25768 fold_stmt returned true.
25770 2017-06-30 Nathan Sidwell <nathan@acm.org>
25772 * ggc.h (empty_string): Delete.
25773 * cfgexpand.c (expand_asm_stmt): Use plain "".
25774 * optabs.c (expand_asm_memory_barrier): Likewise.
25775 * stringpool.c (empty_string): Delete.
25776 (digit_vector, digit_string): Delete.
25777 (ggc_alloc_string): Use plain "", don't optimize single digit
25778 strings. Use ggc_alloc_atomic.
25780 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25782 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25783 comparison set and one other set, use the cost of the non-comparison
25786 2017-06-30 Nathan Sidwell <nathan@acm.org>
25788 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25791 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25793 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25794 loops. Remove now unneeded calls to gimple_switch_set_label() that
25795 just set removed labels to NULL_TREE.
25797 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25799 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25800 (set_range_info): ...here. Only call set_range_info_raw if domain
25802 (set_nonzero_bits): Call set_range_info_raw.
25803 * tree-ssanames.h (set_range_info_raw): New.
25805 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25808 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25809 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25810 of nonimmediate_operand and <store_mask_constraint> instead of m
25811 for the input operand. For V8FI iterator, always split if input
25812 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25813 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25814 <store_mask_predicate> instead of register_operand and
25815 <store_mask_constraint> instead of v for the input operand. Make
25816 sure both operands aren't MEMs for if not <mask_applied>.
25818 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25820 * lto-wrapper.c (copy_file) Close both file descriptors before
25823 2017-06-30 Martin Liska <mliska@suse.cz>
25826 * multiple_target.c (create_dispatcher_calls): Make ifunc
25827 also for function that don't have calls or are not referenced.
25829 2017-06-30 Richard Biener <rguenther@suse.de>
25831 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25832 analyze the first scalar stmt. Move vector type computation
25833 for the BB case here from ...
25834 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25835 live operation processing in the SLP case properly.
25837 2017-06-30 Richard Biener <rguenther@suse.de>
25839 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25841 2017-06-30 Martin Liska <mliska@suse.cz>
25844 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25845 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25847 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25849 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25852 2017-06-30 Richard Biener <rguenther@suse.de>
25854 PR tree-optimization/81249
25855 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25856 condition reduction result to original scalar type.
25858 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25860 * profile-count.h (enum profile_quality): Fix typos and whitespace
25863 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25865 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25866 type for branch probabilities.
25868 2017-06-29 Julian Brown <julian@codesourcery.com>
25869 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25871 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25872 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25873 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25874 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25876 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25878 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25879 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25880 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25881 CC usage from generic code to here.
25882 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25883 CC usage into the target macros.
25885 2017-06-29 Maya Rashish <coypu@sdf.org>
25887 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25890 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25892 * arm/arm-builtins.c: Include profile-count.h
25893 * except.c (sjlj_emit_function_enter): Use
25894 profile_probability::unlikely.
25896 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25898 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25899 and tocrel_offset be pointer args rather than implicitly using
25901 (legitimate_constant_pool_address_p, rs6000_emit_move,
25902 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25903 tocrel_offset and use in toc_relative_expr_p call.
25904 (print_operand, print_operand_address): Use static tocrel_base_oac
25905 and tocrel_offset_oac.
25906 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25909 2017-06-29 Maya Rashish <coypu@sdf.org>
25911 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25913 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25915 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25916 objects, take into account only the alignment of 'op0' and 'mode1' if
25919 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25921 * ccmp.c (ccmp_tree_comparison_p): New function.
25922 (ccmp_candidate_p): Update to use above function.
25923 (get_compare_parts): New function.
25924 (expand_ccmp_next): Update to use new functions.
25925 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25927 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25928 take mode as argument.
25929 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25930 * expr.c (expand_expr_real_1): Pass mode as argument.
25932 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25934 * combine.c (combine_instructions): Print insns to dump_file, together
25937 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25939 * asan.c (asan_emit_stack_protection): Update.
25940 (create_cond_insert_point): Update.
25941 * auto-profile.c (afdo_propagate_circuit): Update.
25942 * basic-block.h (struct edge_def): Turn probability to
25943 profile_probability.
25944 (EDGE_FREQUENCY): Update.
25945 * bb-reorder.c (find_traces_1_round): Update.
25946 (better_edge_p): Update.
25947 (sanitize_hot_paths): Update.
25948 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25949 (make_single_succ_edge): Update.
25950 (check_bb_profile): Update.
25951 (dump_edge_info): Update.
25952 (update_bb_profile_for_threading): Update.
25953 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25954 probabilitycount to 0.
25955 * cfgbuild.c (compute_outgoing_frequencies): Update.
25956 * cfgcleanup.c (try_forward_edges): Update.
25957 (outgoing_edges_match): Update.
25958 (try_crossjump_to_edge): Update.
25959 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25960 (expand_gimple_tailcall): Update.
25961 (construct_init_block): Use make_single_succ_edge.
25962 (construct_exit_block): Use make_single_succ_edge.
25963 * cfghooks.c (verify_flow_info): Update.
25964 (redirect_edge_succ_nodup): Update.
25965 (split_edge): Update.
25966 (account_profile_record): Update.
25967 * cfgloopanal.c (single_likely_exit): Update.
25968 * cfgloopmanip.c (scale_loop_profile): Update.
25969 (set_zero_probability): Remove.
25970 (duplicate_loop_to_header_edge): Update.
25971 * cfgloopmanip.h (loop_version): Update prototype.
25972 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25973 (force_nonfallthru_and_redirect): Update.
25974 (update_br_prob_note): Update.
25975 (rtl_verify_edges): Update.
25976 (purge_dead_edges): Update.
25977 (rtl_lv_add_condition_to_bb): Update.
25978 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25979 * cgraphunit.c (init_lowered_empty_function): Update.
25980 (cgraph_node::expand_thunk): Update.
25981 * cilk-common.c: Include profile-count.h
25982 * dojump.c (inv): Remove.
25983 (jumpifnot): Update.
25984 (jumpifnot_1): Update.
25985 (do_jump_1): Update.
25987 (do_jump_by_parts_greater_rtx): Update.
25988 (do_compare_rtx_and_jump): Update.
25989 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25990 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25991 * dwarf2cfi.c: Include profile-count.h
25992 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25993 (sjlj_emit_dispatch_table): Likewise.
25994 * explow.c: Include profile-count.h
25995 * expmed.c (emit_store_flag_force): Update.
25996 (do_cmp_and_jump): Update.
25997 * expr.c (compare_by_pieces_d::generate): Update.
25998 (compare_by_pieces_d::finish_mode): Update.
25999 (emit_block_move_via_loop): Update.
26000 (store_expr_with_bounds): Update.
26001 (store_constructor): Update.
26002 (expand_expr_real_2): Update.
26003 (expand_expr_real_1): Update.
26004 * expr.h (try_casesi, try_tablejump): Update prototypes.
26005 * gimple-pretty-print.c (dump_probability): Update.
26006 (dump_profile): New.
26007 (dump_gimple_label): Update.
26008 (dump_gimple_bb_header): Update.
26009 * graph.c (draw_cfg_node_succ_edges): Update.
26010 * hsa-gen.c (convert_switch_statements): Update.
26011 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26012 (find_if_case_1): Update.
26013 (find_if_case_2): Update.
26014 * internal-fn.c (expand_arith_overflow_result_store): Update.
26015 (expand_addsub_overflow): Update.
26016 (expand_neg_overflow): Update.
26017 (expand_mul_overflow): Update.
26018 (expand_vector_ubsan_overflow): Update.
26019 * ipa-cp.c (good_cloning_opportunity_p): Update.
26020 * ipa-split.c (split_function): Use make_single_succ_edge.
26021 * ipa-utils.c (ipa_merge_profiles): Update.
26022 * loop-doloop.c (add_test): Update.
26023 (doloop_modify): Update.
26024 * loop-unroll.c (compare_and_jump_seq): Update.
26025 (unroll_loop_runtime_iterations): Update.
26026 * lra-constraints.c (lra_inheritance): Update.
26027 * lto-streamer-in.c (input_cfg): Update.
26028 * lto-streamer-out.c (output_cfg): Update.
26029 * mcf.c (adjust_cfg_counts): Update.
26030 * modulo-sched.c (sms_schedule): Update.
26031 * omp-expand.c (expand_omp_for_init_counts): Update.
26032 (extract_omp_for_update_vars): Update.
26033 (expand_omp_ordered_sink): Update.
26034 (expand_omp_for_ordered_loops): Update.
26035 (expand_omp_for_generic): Update.
26036 (expand_omp_for_static_nochunk): Update.
26037 (expand_omp_for_static_chunk): Update.
26038 (expand_cilk_for): Update.
26039 (expand_omp_simd): Update.
26040 (expand_omp_taskloop_for_outer): Update.
26041 (expand_omp_taskloop_for_inner): Update.
26042 * omp-simd-clone.c (simd_clone_adjust): Update.
26043 * optabs.c (expand_doubleword_shift): Update.
26044 (expand_abs): Update.
26045 (emit_cmp_and_jump_insn_1): Update.
26046 (expand_compare_and_swap_loop): Update.
26047 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26048 * predict.c (predictable_edge_p): Update.
26049 (edge_probability_reliable_p): Update.
26050 (set_even_probabilities): Update.
26051 (combine_predictions_for_insn): Update.
26052 (combine_predictions_for_bb): Update.
26053 (propagate_freq): Update.
26054 (estimate_bb_frequencies): Update.
26055 (force_edge_cold): Update.
26056 * profile-count.c (profile_count::dump): Add missing space into dump.
26057 (profile_count::debug): Add newline.
26058 (profile_count::differs_from_p): Explicitly convert to unsigned.
26059 (profile_count::stream_in): Update.
26060 (profile_probability::dump): New member function.
26061 (profile_probability::debug): New member function.
26062 (profile_probability::differs_from_p): New member function.
26063 (profile_probability::differs_lot_from_p): New member function.
26064 (profile_probability::stream_in): New member function.
26065 (profile_probability::stream_out): New member function.
26066 * profile-count.h (profile_count_quality): Rename to ...
26067 (profile_quality): ... this one.
26068 (profile_probability): New.
26069 (profile_count): Update.
26070 * profile.c (compute_branch_probabilities): Update.
26071 * recog.c (peep2_attempt): Update.
26072 * sched-ebb.c (schedule_ebbs): Update.
26073 * sched-rgn.c (find_single_block_region): Update.
26074 (compute_dom_prob_ps): Update.
26075 (schedule_region): Update.
26076 * sel-sched-ir.c (compute_succs_info): Update.
26077 * stmt.c (struct case_node): Update.
26078 (do_jump_if_equal): Update.
26079 (get_outgoing_edge_probs): Update.
26080 (conditional_probability): Update.
26081 (emit_case_dispatch_table): Update.
26082 (expand_case): Update.
26083 (expand_sjlj_dispatch_table): Update.
26084 (emit_case_nodes): Update.
26085 * targhooks.c: Update.
26086 * tracer.c (better_p): Update.
26087 (find_best_successor): Update.
26088 * trans-mem.c (expand_transaction): Update.
26089 * tree-call-cdce.c: Update.
26090 * tree-cfg.c (gimple_split_edge): Upate.
26091 (move_sese_region_to_fn): Upate.
26092 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26093 * tree-eh.c (lower_resx): Upate.
26094 (cleanup_empty_eh_move_lp): Upate.
26095 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26096 * tree-inline.c (copy_edges_for_bb): Update.
26097 (copy_cfg_body): Update.
26098 * tree-parloops.c (gen_parallel_loop): Update.
26099 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26100 (gimple_gen_time_profiler): Update.
26101 * tree-ssa-dce.c (remove_dead_stmt): Update.
26102 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26103 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26104 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26105 (unloop_loops): Update.
26106 (try_peel_loop): Update.
26107 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26108 * tree-ssa-loop-split.c (connect_loops): Update.
26109 (split_loop): Update.
26110 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26111 (hoist_guard): Update.
26112 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26113 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26114 (value_replacement): Update.
26115 * tree-ssa-reassoc.c (branch_fixup): Update.
26116 * tree-ssa-tail-merge.c (replace_block_by): Update.
26117 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26118 (create_edge_and_update_destination_phis): Update.
26119 (compute_path_counts): Update.
26120 (recompute_probabilities): Update.
26121 (update_joiner_offpath_counts): Update.
26122 (freqs_to_counts_path): Update.
26123 (duplicate_thread_path): Update.
26124 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26125 (struct switch_conv_info): Update.
26126 (gen_inbound_check): Update.
26127 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26128 (vect_do_peeling): Update.
26129 (vect_loop_versioning): Update.
26130 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26131 (optimize_mask_stores): Update.
26132 * ubsan.c (ubsan_expand_null_ifn): Update.
26133 * value-prof.c (gimple_divmod_fixed_value): Update.
26134 (gimple_divmod_fixed_value_transform): Update.
26135 (gimple_mod_pow2): Update.
26136 (gimple_mod_pow2_value_transform): Update.
26137 (gimple_mod_subtract): Update.
26138 (gimple_mod_subtract_transform): Update.
26139 (gimple_ic): Update.
26140 (gimple_stringop_fixed_value): Update.
26141 (gimple_stringops_transform): Update.
26142 * value-prof.h: Update.
26144 2017-06-29 Carl Love <cel@us.ibm.com>
26146 * config/rs6000/rs6000-c.c: Add support for built-in functions
26147 vector signed int vec_signed (vector float);
26148 vector signed long long vec_signed (vector double);
26149 vector signed int vec_signed2 (vector double, vector double);
26150 vector signed int vec_signede (vector double);
26151 vector signed int vec_signedo (vector double);
26152 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26153 instruction generator.
26154 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26155 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26156 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26158 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26159 vunsignede_v2df): Add define_expands.
26160 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26161 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26162 VEC_UNSIGNEDO): Add definitions.
26163 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26164 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26165 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26166 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26167 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26168 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26169 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26170 vec_unsignede, vec_unsignedo): Add builtin defines.
26171 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26173 * doc/extend.texi: Update the built-in documentation file for the
26174 new built-in functions.
26176 2017-06-29 Richard Biener <rguenther@suse.de>
26178 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26179 reduction chains to LOOP_VINFO_REDUCTIONS.
26180 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26181 SLP reductions after processing reduction chains.
26183 2017-06-29 Nathan Sidwell <nathan@acm.org>
26185 * builtins.c (fold_builtin_FUNCTION): Use
26186 lang_hooks.decl_printable_name.
26188 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26190 PR middle-end/81194
26191 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26192 with only one label.
26193 * stmt.c (expand_case): Assert NCASES is greater than one.
26195 2017-06-29 Richard Biener <rguenther@suse.de>
26197 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26199 (group_case_labels): Likewise.
26200 (find_taken_edge): Push sanity checking on val to workers...
26201 (find_taken_edge_cond_expr): ... here
26202 (find_taken_edge_switch_expr): ... and here, handle cases
26203 with just a default label.
26204 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26205 (group_case_labels): Likewise.
26206 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26207 group_case_labels does anything cleanup the CFG again.
26209 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26211 PR tree-optimization/81196
26212 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26213 exit condition comparing two IVs.
26215 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26217 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26218 profile to the dummy entry at the end of the list of architectures.
26219 * config/arm/arm-cpu-cdata.h: Regenerated.
26221 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26222 Michael Collison <michael.collison@arm.com>
26225 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26227 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26228 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26229 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26230 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26231 of shift when the shift amount is masked with constant equal to
26232 the size of the mode.
26233 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26236 2017-06-29 Martin Liska <mliska@suse.cz>
26238 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26240 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26243 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26246 (edge_growth_cache_entry::edge_growth_cache_entry): New
26248 (reset_edge_growth_cache): Update to use constructor.
26250 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26252 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26253 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26254 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26256 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26258 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26259 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26261 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26263 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26264 (*-linux-uclibc*): Add t-uclibc tmake_file.
26265 * config/t-musl: New.
26266 * config/t-uclibc: New.
26268 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26270 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26272 (gen_comm_data): Emit architectural setting of arch_prof.
26273 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26275 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26276 (armv8-m.base, armv8-m.main): Likewise.
26277 * arm-protos.h (arm_build_target): Add profile field.
26278 (arch_option): Likewise.
26279 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26281 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26282 arm_active_target.profile.
26284 2017-06-28 Richard Biener <rguenther@suse.de>
26286 PR middle-end/81227
26287 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26288 TYPE_OVERFLOW_WRAPS.
26289 * match.pd (negate_expr_p): Likewise.
26290 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26291 fold_build2, not fold_binary.
26293 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26295 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26296 Convert memory address to Pmode.
26297 (aarch64_print_operand): Assert MEM operands are always Pmode.
26299 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26302 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26303 Remove redundant if.
26304 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26305 * config/arm/aarch-common-protos.h
26306 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26307 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26309 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26312 * multiple_target.c (create_dispatcher_calls): Set the default
26313 clone to be static, not public.
26315 2017-06-28 Richard Biener <rguenther@suse.de>
26317 * tree-vect-loop.c (vectorizable_reduction): Move special
26318 cond reduction IV var creation ...
26319 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26320 parameter. Use STMT_VINFO_VECTYPE.
26321 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26324 2017-06-28 Martin Liska <mliska@suse.cz>
26327 * ipa-visibility.c (non_local_p): Handle visibility.
26329 2017-06-28 Martin Liska <mliska@suse.cz>
26332 * common.opt: Add IntegerRange to various options.
26333 * opt-functions.awk (integer_range_info): New function.
26334 * optc-gen.awk: Add integer_range_info to cl_options struct.
26335 * opts-common.c (decode_cmdline_option): Handle
26336 CL_ERR_INT_RANGE_ARG.
26337 (cmdline_handle_error): Likewise.
26338 * opts.c (print_filtered_help): Show valid interval in
26339 when --help is provided.
26340 * opts.h (struct cl_option): Add range_min and range_max fields.
26341 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26343 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26345 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26346 (x * C EQ/NE y * C): New transformation.
26348 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26350 * genmultilib (combination_space): Accept '+' in option names.
26352 2017-06-28 Martin Liska <mliska@suse.cz>
26355 * asan.c (instrument_derefs): Bail out inner references
26356 that are hard register variables.
26358 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26361 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26362 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26364 2017-06-28 Richard Biener <rguenther@suse.de>
26366 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26367 (vect_get_slp_defs): Adjust.
26368 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26370 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26372 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26373 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26374 (vectorizable_reduction): Adjust.
26375 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26377 (vect_get_slp_defs): Likewise.
26378 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26379 (vectorizable_bswap): Adjust.
26380 (vectorizable_call): Likewise.
26381 (vectorizable_conversion): Likewise.
26382 (vectorizable_assignment): Likewise.
26383 (vectorizable_shift): Likewise.
26384 (vectorizable_operation): Likewise.
26385 (vectorizable_store): Likewise.
26386 (vectorizable_condition): Likewise.
26387 (vectorizable_comparison): Likewise.
26389 2017-06-28 Michael Collison <michael.collison@arm.com>
26392 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26394 (arm_gen_movmemqi): Removed unused variable 'i'.
26395 Convert 'for' loop into 'while' loop.
26396 (arm_expand_prologue): Remove last unnecessary set of insn.
26397 (thumb_pop): Remove unused variable 'pushed_words'.
26398 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26400 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26402 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26403 * config/s390/s390-protos.h: Add prototype of
26404 s390_rel_address_ok_p.
26405 * config/s390/s390.c (s390_got_symbol): New function.
26406 (s390_rel_address_ok_p): New function.
26407 (legitimize_pic_address): Use s390_rel_address_ok_p.
26408 (s390_load_got): Use s390_got_symbol.
26409 (s390_option_override): Issue error if
26410 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26411 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26413 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26415 2017-06-27 Andrew Pinski <apinski@cavium.com>
26417 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26418 (X * copysign (1.0, X)): New pattern.
26419 (X * copysign (1.0, -X)): New pattern.
26420 (copysign (-1.0, CST)): New pattern.
26422 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26424 * genmultilib (combination_space): Remove variable.
26425 Validate reuse rules against regular expression for any sequence
26426 of multilib options in any order.
26428 2017-06-27 Michael Collison <michael.collison@arm.com>
26430 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26431 call aarch64_split_simd_combine.
26432 * (aarch64_combine_internal<mode>): Delete pattern.
26433 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26434 Allow register and subreg operands.
26436 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26438 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26439 specific need, just fallback on defaults.
26440 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26442 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26443 Olivier Hainque <hainque@adacore.com>
26445 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26447 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26448 targets. Pick a default if no particular attempt applied.
26449 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26452 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26454 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26455 (x86_64-wrs-vxworks7): Likewise.
26457 2017-06-27 Marek Polacek <polacek@redhat.com>
26460 * ubsan.c (instrument_null): Check get_base_address's result for null.
26462 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26464 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26466 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26468 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26469 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26470 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26471 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26472 New function types.
26473 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26474 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26475 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26476 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26477 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26478 BUILT_IN_FEUPDATEENV): New builtins.
26479 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26480 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26481 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26482 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26484 (builtin_structptr_types): Adjust size.
26485 * tree.c (builtin_structptr_types): Add four entries.
26487 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26488 Olivier Hainque <hainque@adacore.com>
26490 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26491 (TLS_SYM): New local macro, forcing reference to __tls__ on
26492 link command lines for VxWorks 7 RTPs, triggering initialization
26494 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26495 OS features TLS support, true for RTPs on VxWorks 7.
26496 * config/vxworks.c (vxworks_override_options): Setup emutls
26499 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26501 * predict.c (test_prediction_value_range): Use -1U instead of -1
26502 to avoid narrowing conversion warning.
26503 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26504 to avoid narrowing conversion warning.
26505 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26507 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26509 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26511 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26512 64bit configurations.
26513 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26514 (SIZE_TYPE): Likewise.
26515 * config/vxworks.c (vxworks_emutls_var_fields): Use
26516 long_unsigned_type_node instead of unsigned_type_node as the offset
26517 field type, which is "pointer" mode in emutls.c.
26519 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26522 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26524 PR middle-end/81207
26525 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26526 gimple_vuse copying separately from gimple_vdef copying.
26528 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26530 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26533 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26534 Olivier Hainque <hainque@adacore.com>
26536 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26537 for all vxworks7 targets.
26538 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26539 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26540 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26541 variations for VX6/VX7 and 32/64bits later on in ...
26542 (VXWORKS_LIB_SPEC): Leverage new macros.
26543 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26544 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26546 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26548 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26549 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26551 2017-06-26 Carl Love <cel@us.ibm.com>
26553 * config/rs6000/rs6000-c.c: Add support for built-in functions
26554 vector bool char vec_reve (vector bool char);
26555 vector signed char vec_reve (vector signed char);
26556 vector unsigned char vec_reve (vector unsigned char);
26557 vector bool int vec_reve (vector bool int);
26558 vector signed int vec_reve (vector signed int);
26559 vector unsigned int vec_reve (vector unsigned int);
26560 vector bool long long vec_reve (vector bool long long);
26561 vector signed long long vec_reve (vector signed long long);
26562 vector unsigned long long vec_reve (vector unsigned long long);
26563 vector bool short vec_reve (vector bool short);
26564 vector signed short vec_reve (vector signed short);
26565 vector double vec_reve (vector double);
26566 vector float vec_reve (vector float);
26567 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26568 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26569 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26570 (altivec_vreve): New pattern.
26571 * config/rs6000/altivec.h (vec_reve): New define.
26572 * doc/extend.texi (vec_rev): Update the built-in documentation file
26573 for the new built-in functions.
26575 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26577 PR tree-optimization/71815
26578 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26580 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26582 (slsr_process_phi): Likewise.
26583 (replace_uncond_cands_and_profitable_phis): Don't replace a
26584 multiply candidate with a stride of 1 (copy or cast).
26585 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26587 (lowest_cost_path): Likewise.
26588 (total_savings): Likewise.
26590 2017-06-26 Richard Biener <rguenther@suse.de>
26593 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26594 Use def_builtin_pure for all gather builtins.
26596 2017-06-26 Richard Biener <rguenther@suse.de>
26598 PR tree-optimization/81203
26599 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26600 non-dominating BBs.
26602 2017-06-26 Marek Polacek <polacek@redhat.com>
26605 * doc/invoke.texi: Document -Wmultistatement-macros.
26607 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26609 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26610 arm_neon_ok_no_float_abi effective target.
26612 2017-06-26 Richard Biener <rguenther@suse.de>
26614 PR tree-optimization/80928
26615 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26616 (copy_bbs): Set BB_DUPLICATED flag early.
26617 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26619 (execute_on_shrinking_pred): Likewise.
26620 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26621 BB_DUPLICATED blocks.
26622 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26623 iterate over all PHIs considering removal of *gsi.
26625 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26627 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26630 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26632 * config/rs6000/rs6000-string.c: (expand_block_clear,
26633 do_load_for_compare, select_block_compare_mode,
26634 compute_current_alignment, expand_block_compare,
26635 expand_strncmp_align_check, expand_strn_compare,
26636 expand_block_move, rs6000_output_load_multiple)
26637 Move functions related to string/block move/compare
26638 to a separate file.
26639 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26640 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26641 for this function which is now used in two files.
26642 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26643 * config.gcc: Add rs6000-string.o to extra_objs for
26644 targets powerpc*-*-* and rs6000*-*-*.
26646 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26649 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26650 32-bit, since indexed is not valid for DImode.
26651 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26652 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26653 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26654 (define_peephole2 for Altivec d-form store): Likewise.
26657 * multiple_target.c (create_dispatcher_calls): Only create the
26658 dispatcher call if the function is the default clone of a
26659 versioned function.
26661 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26663 PR middle-end/80902
26664 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26665 a call, force the call to not be a tail call.
26667 2017-06-23 Jeff Law <law@redhat.com>
26669 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26672 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26674 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26675 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26676 handling for early expansion of vector shifts (sl,sr,sra,rl).
26677 (builtin_function_type): Add vector shift right instructions
26678 to the unsigned argument list.
26680 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26682 rtl-optimizatoin/79286
26683 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26684 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26685 trap. PIC register plus a const unspec without offset can never trap.
26687 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26689 * tree.h (builtin_structptr_type): New type.
26690 (builtin_structptr_types): Declare new array.
26691 * tree.c (builtin_structptr_types): New array.
26692 (free_lang_data, build_common_tree_nodes): Use it.
26694 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26697 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26700 2017-06-22 Matt Turner <mattst88@gmail.com>
26702 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26703 Lake models to skylake case. Assume skylake for unknown
26704 models with clflushopt.
26706 2017-06-22 Jeff Law <law@redhat.com>
26708 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26709 frame sizes that do not satisfy aarch64_uimm12_shift.
26711 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26713 * profile-count.h (apply_probability,
26714 apply_scale, probability_in): Fix checks for zero.
26716 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26718 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26719 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26721 2016-06-22 Richard Biener <rguenther@suse.de>
26723 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26724 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26725 REDUC_MAX_EXPR support.
26726 (vectorizable_reduction): Likewise.
26727 (vect_create_epilog_for_reduction): Likewise.
26729 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26731 * match.pd (A / (1 << B) -> A >> B): New.
26732 * generic-match-head.c: Include optabs-tree.h.
26733 * gimple-match-head.c: Likewise.
26734 * optabs-tree.h (target_supports_op_p): New.
26735 * optabs-tree.c (target_supports_op_p): New.
26737 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26739 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26740 $gcc_cv_ld --help output.
26741 (gcc_cv_ld_demangle): Likewise.
26742 (gcc_cv_ld_eh_frame_hdr): Likewise.
26743 (gcc_cv_ld_pie): Likewise.
26744 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26745 (gcc_cv_ld_buildid): Likewise.
26746 (gcc_cv_ld_sysroot): Likewise.
26747 (ld_bndplt_support): Likewise.
26748 (ld_pushpopstate_support): Likewise.
26749 * configure: Regenerate.
26750 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26752 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26755 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26756 operands indexes to avoid gap between operands and match_dups.
26758 2017-06-21 Andrew Pinski <apinski@cavium.com>
26760 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26761 Increment Arith_shift and Arith_shift_reg by 1.
26762 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26764 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26765 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26766 (aarch64_strip_extend): Add new argument and test for it.
26767 (aarch64_cheap_mult_shift_p): New function.
26768 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26769 add a cost if it is true.
26770 Update calls to aarch64_strip_extend.
26771 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26773 2017-06-21 Andrew Pinski <apinski@cavium.com>
26775 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26777 (thunderxt88): Likewise.
26778 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26779 (thunderx_prefetch_tune): New variable.
26780 (thunderx2t99_prefetch_tune): Update for the correct values.
26781 (thunderxt88_tunings): New variable.
26782 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26783 generic_prefetch_tune.
26784 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26786 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26788 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26789 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26790 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26791 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26792 (aarch64_atomic_cas<mode>, GPI): Likewise.
26794 2017-06-21 Martin Liska <mliska@suse.cz>
26796 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26797 statements on cold and hot labels.
26798 * predict.c (tree_estimate_probability_bb): Remove the
26799 prediction from this place.
26801 2017-06-21 Martin Liska <mliska@suse.cz>
26803 PR tree-optimization/79489
26804 * gimplify.c (maybe_add_early_return_predict_stmt): New
26806 (gimplify_return_expr): Call the function.
26807 * predict.c (tree_estimate_probability_bb): Remove handling
26809 * predict.def: Update comment about early return predictor.
26810 * gimple-predict.h (is_gimple_predict): New function.
26811 * predict.def: Change default value of early return to 66.
26812 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26814 * passes.def: Put pass_strip_predict_hints to the beginning of
26817 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26819 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26820 FUNCTION_DECL declarations.
26821 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26823 (dwaf2out_decl): Likewise.
26824 * godump.c (go_early_global_decl): Skip call to the real debug hook
26825 for FUNCTION_DECL declarations.
26826 * passes.c (rest_of_decl_compilation): Skip call to the
26827 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26828 -fdump-go-spec is passed.
26830 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26832 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26833 Reorder for compactness.
26834 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26835 (def_builtin_pure, def_builtin_pure2): New functions.
26836 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26838 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26840 * match.pd (nop_convert): New predicate.
26841 ((A +- CST1) +- CST2): Allow some NOP conversions.
26843 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26846 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26847 with ctors/dtors if GOVD_SHARED is set.
26849 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26851 * config/aarch64/aarch64.md (movti_aarch64):
26852 Emit mov rather than orr.
26853 (movtf_aarch64): Likewise.
26854 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26855 Emit mov rather than orr.
26857 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26859 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26860 Swap alternatives, make integer dup more expensive.
26862 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26864 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26865 Return true for non-tls symbols.
26867 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26869 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26870 (cortex-a75): Likewise.
26871 (cortex-a75.cortex-a55): Likewise.
26872 * config/aarch64/aarch64-tune.md: Regenerate.
26873 * doc/invoke.texi (-mtune): Document new values for -mtune.
26875 2017-06-21 Tom de Vries <tom@codesourcery.com>
26877 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26878 stack_size feature.
26879 (Effective-Target Keywords, Other attributes): Suggest using
26880 dg-add-options stack_size feature to get stack limit in stack_size
26881 effective target documentation.
26883 2017-06-21 Julian Brown <julian@codesourcery.com>
26884 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26886 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26887 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26888 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26890 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26891 attribute type list for neon_multiply.
26892 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26893 attribute type list for neon_multiply.
26894 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26895 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26896 attribute type list for neon_multiply.
26897 * config/arm/types.md (crypto_pmull): Add.
26898 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26899 attribute type list.
26901 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26903 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26906 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26908 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26909 to make sure not to dereference a NULL cost_classes_ptr pointer.
26911 2017-06-20 Carl Love <cel@us.ibm.com>
26913 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26914 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26915 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26916 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26917 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26918 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26919 VMULOSW): New enum "unspec" values.
26920 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26921 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26922 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26923 altivec_vmulosw): New patterns.
26924 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26925 VMULOSW): Add definitions.
26927 2017-06-20 Julia Koval <julia.koval@intel.com>
26929 * config/i386/i386.c: Fix rounding expand for new pattern.
26930 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26932 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26934 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26935 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26937 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26939 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26942 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26944 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26945 architecture, then by implementer ID.
26946 * config/aarch64/aarch64-tune.md: Regenerate.
26948 2017-06-20 Richard Biener <rguenther@suse.de>
26950 PR middle-end/81097
26951 * fold-const.c (split_tree): Fold to type before negating.
26953 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26955 * diagnostic-show-locus.c
26956 (selftest::test_fixit_deletion_affecting_newline): New function.
26957 (selftest::diagnostic_show_locus_c_tests): Call it.
26959 2017-06-20 Andreas Schwab <schwab@suse.de>
26962 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26965 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26967 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26968 __ARM_FEATURE_COPROC according to support.
26970 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26972 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26973 Rewritten to avoid overflow for > 32-bit pointers.
26976 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26977 by removing enum keyword.
26978 (ubsan_type_descriptor): Likewise. Formatting fix.
26981 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26982 splitter): Require TARGET_SSE2 in the condition.
26984 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26987 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26988 for doing vector set of SFmode on ISA 3.0.
26989 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26990 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26992 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26993 SFmode value into a V4SF variable that was extracted from another
26994 V4SF variable without converting the element to double precision
26995 and back to single precision vector format.
26996 (vsx_insert_extract_v4sf_p9_2): Likewise.
26998 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27000 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27001 in UWHI to avoid undefined overflow.
27004 * ubsan.h (enum ubsan_encode_value_phase): New.
27005 (ubsan_encode_value): Change second argument to
27006 enum ubsan_encode_value_phase with default value of
27007 UBSAN_ENCODE_VALUE_GENERIC.
27008 * ubsan.c (ubsan_encode_value): Change second argument to
27009 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27010 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27011 create_tmp_var_raw instead of create_tmp_var and use a
27013 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27014 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27015 ubsan_encode_value callers.
27018 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27019 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27020 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27022 2017-06-19 Richard Biener <rguenther@suse.de>
27024 PR middle-end/81118
27025 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27026 estimates if we changed anything.
27028 2017-06-19 Richard Biener <rguenther@suse.de>
27030 PR tree-optimization/80887
27031 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27032 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27033 simplified lookups, then reset mprts_hook.
27034 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27036 (try_to_simplify): Likewise.
27038 2017-06-19 Martin Liska <mliska@suse.cz>
27041 * gimplify.c (gimplify_switch_expr):
27042 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27044 2017-06-19 Martin Liska <mliska@suse.cz>
27046 * doc/install.texi: Document that PGO runs in 4 stages.
27048 2017-06-19 Martin Liska <mliska@suse.cz>
27051 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27052 to dispatcher function name.
27053 * multiple_target.c (replace_function_decl): New function.
27054 (create_dispatcher_calls): Redirect both edges and references.
27056 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27058 * profile-count.c (profile_count::dump): Dump quality.
27059 (profile_count::differs_from_p): Update for unsigned val.
27060 * profile-count.h (profile_count_quality): New enum.
27061 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27063 2017-06-19 Richard Biener <rguenther@suse.de>
27065 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27066 struct function as arg.
27067 (estimate_numbers_of_iterations): Export overload with loop arg.
27068 (free_numbers_of_iterations_estimates_loop): Use an overload of
27069 free_numbers_of_iterations_estimates instead.
27070 * tree-cfg.c (remove_bb): Adjust.
27071 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27072 * tree-parloops.c (gen_parallel_loop): Likewise.
27073 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27075 (tree_unroll_loops_completely): Likewise.
27076 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27077 Use an overload instead and export.
27078 (estimated_loop_iterations): Adjust.
27079 (max_loop_iterations): Likewise.
27080 (likely_max_loop_iterations): Likewise.
27081 (estimate_numbers_of_iterations): Take struct function as arg
27083 (loop_exits_before_overflow): Adjust.
27084 (free_numbers_of_iterations_estimates_loop): Use an overload.
27085 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27086 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27088 2017-06-19 Richard Biener <rguenther@suse.de>
27091 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27092 RANGE_EXPR conservatively.
27094 2017-06-16 Carl Love <cel@us.ibm.com>
27096 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27097 definitions for vec_float, vec_float2, vec_floato,
27098 vec_floate built-ins.
27099 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27100 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27102 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27103 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27104 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27105 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27106 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27107 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27108 vec_floato): Add builtin defines.
27109 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27110 Update the built-in documentation file for the new built-in
27113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27115 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27116 (mthumb): Mark as the negative of -marm.
27118 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27120 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27122 (ARM Options, -mtune): Document that this accepts the same
27123 extension options as -mcpu.
27124 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27128 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27129 permitted extensions.
27131 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27133 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27134 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27135 (armv8-m.main): Add option +nodsp.
27136 * config/arm/arm-cpu-cdata.h: Regenerated.
27138 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27140 * config/arm/t-fuchsia: New file.
27141 * config.gcc (arm*-*-fuchsia*): Use it.
27143 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27145 * config/arm/t-symbian: Rewrite for new option infrastructure.
27147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27149 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27150 (MULTILIB_REQUIRED): Likewise.
27152 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27154 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27155 (MULTILIB_RESUE): Likewise.
27156 (MULTILIB_MATCHES): Likewise.
27157 (MULTLIB_REQUIRED): Likewise.
27159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27161 * config/arm/t-rtems: Rewrite for new option framework.
27163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27165 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27166 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27167 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27168 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27169 * config/arm/t-multilib: ... here.
27170 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27171 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27172 armv7-a and armv8*-a when A-profile libraries have not been built.
27173 * config/arm/t-rmprofile: Rewrite.
27175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27177 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27178 with a backslash. Remove the backslash after substituting unescaped
27180 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27184 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27185 the driver to pass through the default setting of -mfloat-abi.
27186 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27188 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27189 (all_feat_combs): New rule.
27190 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27192 * config/arm/t-aprofile: Rewrite.
27194 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27196 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27197 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27198 fpu is not specified by the user/command-line.
27199 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27200 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27201 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27202 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27203 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27204 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27206 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27208 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27209 * config/arm/t-arm-elf: Rewritten.
27211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27213 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27214 have some floating-point instructions.
27215 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27216 (TARGET_MAYBE_HARD_FLOAT): New macro.
27217 * config/arm/arm-builtins.c (arm_init_builtins): Use
27218 TARGET_MAYBE_HARD_FLOAT.
27219 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27221 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27223 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27224 (configargs.h): Include it.
27225 (arm_print_hint_for_fpu_option): New function.
27226 (arm_parse_fpu_option): New function.
27227 (candidate_extension): New class.
27228 (arm_canon_for_multilib): New function.
27229 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27230 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27231 (ARCH_CANONICAL_SPECS): New macro.
27232 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27234 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27236 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27237 are set after handling multilib fragments. Set target_cpu_default2
27240 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27242 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27244 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27245 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27247 (check_arch): Likewise.
27248 * config/arm/arm.c (arm_configure_build_target): Handle
27249 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27250 options in the default.
27252 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27254 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27255 when an option is an alias of another.
27256 * config/arm/parsecpu.awk (optalias): New parser token.
27257 (gen_comm_data): Mark non-alias options as such. Emit entries
27258 for extension aliases.
27259 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27260 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27261 (armv6kz, armv6zk, armv6t2): Likewise.
27262 (armv7): Make vfpv3-d16 an alias.
27263 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27265 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27266 Sort in canonical order.
27267 (armv8-a): Sort in canonical order.
27268 (armv8.1-a, armv8.2-a): Likewise.
27269 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27271 (cortex-a9): Sort in canonical order.
27272 * config/arm/arm.c (selftests.h): Include it.
27273 (arm_test_cpu_arch_data): New function.
27274 (arm_run_self_tests): New function.
27275 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27276 (targetm): Move declaration to the end of the file.
27277 * arm-cpu-cdata.h: Regenerated.
27279 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27281 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27282 call to target_mode_check describing the type of option passed.
27283 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27284 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27285 arm_parse_cpu_option_name to match parameters against list of
27287 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27288 arm_arch_core_flags data structure.
27289 * config/arm/arm-cpu_cdata.h: Regenerated.
27291 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27293 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27295 (arm_print_hint_for_cpu_option): Likewise.
27296 (arm_print_hint_for_arch_option): Likewise.
27297 (arm_parse_cpu_option_name): Likewise.
27298 (arm_parse_arch_option_name): Likewise.
27299 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27300 of entries in the all_fpus list.
27301 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27302 (arm_parse_cpu_option_name): Declare.
27303 (arm_parse_arch_option_name): Declare.
27304 (arm_parse_option_features): Declare.
27305 (arm_intialize_isa): Declare.
27306 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27308 (gen_comm_data): ... here. Make definitions non-static.
27309 * config/arm/arm-cpu-data.h: Regenerated.
27310 * config/arm/arm-cpu-cdata.h: Regenerated.
27312 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27314 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27315 (cpu_arch_extension): New structure.
27316 (cpu_arch_option, arch_option, cpu_option): New structures.
27317 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27318 architecture types.
27319 (gen_data): Generate new format data tables.
27320 * config/arm/arm.c (cpu_tune): New structure.
27321 (cpu_option, processors): Delete.
27322 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27323 (arm_print_hint_for_cpu_option): ... this and ...
27324 (arm_print_hint_for_arch_option): ... this.
27325 (arm_parse_arch_cpu_name): Delete. Replace with ...
27326 (arm_parse_cpu_option_name): ... this and ...
27327 (arm_parse_arch_option_name): ... this.
27328 (arm_unrecognized_feature): Change type of target parameter to
27330 (arm_parse_arch_cpu_features): Delete. Replace with ...
27331 (arm_parse_option_features): ... this.
27332 (arm_configure_build_target): Rework to use new configuration data
27334 (arm_print_tune_info): Rework for new configuration data tables.
27335 * config/arm/arm-cpu-data.h: Regenerated.
27336 * config/arm/arm-cpu.h: Regenerated.
27338 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27340 * Makefile.in (OBJS): Move sbitmap.o from here ...
27341 (OBJS-libcommon): ... to here.
27343 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27345 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27346 (ISA_ALL_CRYPTO): New macro.
27347 (ISA_ALL_SIMD): New macro
27348 (ISA_ALL_FP): New macro.
27349 * config/arm/arm.c (fpu_bitlist): Update initializer.
27350 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27352 (arm9e): Add fpu. Add option for nofp
27353 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27354 (arm926ej-s, arm1026ej-s): Likewise.
27355 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27356 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27357 neon-fp16, neon-vfpv4, nofp and nosimd.
27358 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27359 (cortex-a8): Add fpu. Add option for nofp.
27360 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27361 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27362 (cortex-r4f): Add fpu.
27363 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27364 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27366 (cortex-r8): Likewise.
27367 (cortex-m4): Add fpu. Add option for nofp.
27368 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27369 (cortex-a17.cortex-a7): Likewise.
27370 (cortex-a32): Add fpu. Add options for crypto and nofp.
27371 (cortex-a35, cortex-a53): Likewise.
27372 (cortex-a57): Add fpu. Add option for crypto.
27373 (cortex-a72, cortex-a73): Likewise.
27374 (exynos-m1): Likewise.
27375 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27376 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27377 (cortex-m33): Add fpu. Add option for nofp.
27378 * config/arm/arm-cpu-cdata.h: Regenerated
27379 * config/arm/arm-cpu-data.h: Regenerated.
27381 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27383 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27384 (armv5te, armv5tej): Likewise.
27385 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27386 (armv7): Add options fp and vfpv3-d16.
27387 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27388 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27390 (armv7ve): Likewise.
27391 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27392 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27393 (armv8-a): Add nocrypto option.
27394 (armv8.1-a, armv8.2-a): Likewise.
27395 (armv8-m.main): add options fp, fp.dp and nofp.
27397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27399 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27401 (armv8-a+crc): Delete.
27402 (armv8.1-a): Add options simd, crypto and nofp.
27403 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27404 (armv8.2-a+fp16): Delete.
27405 (armv8-m.main): Add option dsp.
27406 (armv8-m.main+dsp): Delete.
27407 (cortex-a8): Add fpu. Add nofp option.
27408 (cortex-a9): Add fpu. Add nofp and nosimd options.
27409 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27410 link to main cpu and architecture data structures.
27411 (gen_comm_data): Only put isa attributes from the main architecture
27413 (option): New statement for architecture and CPU entries.
27414 * arm.c (struct cpu_option): New structure.
27415 (struct processors): Add entry for options.
27416 (arm_unrecognized_feature): New function.
27417 (arm_parse_arch_cpu_name): Ignore any characters after the first
27419 (arm_parse_arch_cpu_feature): New function.
27420 (arm_configure_build_target): Separate out any CPU and architecture
27421 features and parse separately. Don't error out if -mfpu=auto is
27422 used with only an architecture string.
27423 (arm_print_asm_arch_directives): New function.
27424 (arm_file_start): Call it.
27425 * config/arm/arm-cpu-cdata.h: Regenerated.
27426 * config/arm/arm-cpu-data.h: Likewise.
27427 * config/arm/arm-tables.opt: Likewise.
27429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27431 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27432 assembler when it is not -mfpu=auto.
27434 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27436 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27437 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27438 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27439 (ASM_CPU_SPEC): Rewrite.
27440 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27441 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27442 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27443 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27444 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27445 copied string is NUL-terminated. Also strip any characters prefixed
27447 (arm_rewrite_selected_arch): New function.
27448 (arm_rewrite_march): New function.
27450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27452 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27453 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27454 (march, mcpu, mtune): Convert to string-based options.
27455 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27456 (arm_parse_arch_cpu_name): New function.
27457 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27458 identify selected architecture or CPU.
27459 (arm_option_save): New function.
27460 (TARGET_OPTION_SAVE): Redefine.
27461 (arm_option_restore): Restore string options.
27462 (arm_option_print): Print string options.
27464 2017-06-16 Martin Sebor <msebor@redhat.com>
27466 PR tree-optimization/80933
27467 PR tree-optimization/80934
27468 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27469 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27470 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27471 (gimple_fold_builtin): Call them.
27473 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27475 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27476 as unlikely; update profile.
27478 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27480 * predict.c (force_edge_cold): Handle declaring edges impossible
27483 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27485 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27487 (try_unroll_loop_completely): Fix reporting.
27489 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27491 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27493 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27496 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27497 if given a non-constant argument for an intrinsic which requires a
27500 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27502 * profile.c (compare_freqs): New function.
27503 (branch_prob): Sort edge list.
27504 (find_spanning_tree): Assume that the list is priority sorted.
27506 2017-06-16 Richard Biener <rguenther@suse.de>
27508 PR tree-optimization/81090
27509 * passes.def (pass_record_bounds): Remove.
27510 * tree-pass.h (make_pass_record_bounds): Likewise.
27511 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27512 make_pass_record_bounds): Likewise.
27513 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27514 not free niter estimates at the beginning but at the end.
27515 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27517 2017-06-16 Richard Biener <rguenther@suse.de>
27519 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27520 initializer to workaround ICE in host GCC 4.8.
27522 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27524 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27526 (clone_inlined_nodes): Update.
27528 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27530 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27531 prefetch settings, and enable prefetching by default at -O3.
27533 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27535 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27536 Set flag_prefetch_loop_arrays according to tuning data.
27538 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27540 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27541 New tune structure.
27542 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27543 [Unrelated to main purpose of the patch] Place the pointer field last
27544 to enable type checking errors when tune structure are wrongly merged.
27545 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27546 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27547 (thunderx2t99_prefetch_tune): New tune constants.
27548 (tune_params *_tunings): Update all tunings (no functional change).
27549 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27550 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27551 from tunings structures.
27553 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27556 * ubsan.c (instrument_null): Add T argument, use it instead
27557 of computing it based on IS_LHS.
27558 (instrument_object_size): Likewise.
27559 (pass_ubsan::execute): Adjust instrument_null and
27560 instrument_object_size callers to pass gimple_get_lhs or
27561 gimple_assign_rhs1 result to it. Use instrument_null instead of
27562 calling get_base_address and instrument_mem_ref. Handle
27563 aggregate call arguments for object-size sanitization.
27565 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27567 PR tree-optimization/81089
27568 * tree-vrp.c (is_masked_range_test): Validate operands of
27571 2017-06-15 Martin Sebor <msebor@redhat.com>
27574 * dumpfile.c (dump_register): Avoid calling memset to initialize
27575 a class with a default ctor.
27576 * gcc.c (struct compiler): Remove const qualification.
27577 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27578 * hash-table.h: Ditto.
27579 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27581 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27582 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27584 * params.h (struct param_info): Make struct members non-const.
27585 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27586 with default initialization.
27587 * vec.h (vec_copy_construct, vec_default_construct): New helper
27589 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27590 with vec_copy_construct.
27591 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27592 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27593 * doc/invoke.texi (-Wclass-memaccess): Document.
27595 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27597 * emit-rtl.h (is_leaf): Update comment about local
27598 register allocator.
27600 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27603 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27604 for a variable to have a section before checking if the section has a
27606 Set section to.persistent if persistent attribute is set.
27607 Warn if .persistent attribute is used on an automatic variable.
27609 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27611 PR rtl-optimization/80474
27612 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27614 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27616 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27619 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27621 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27622 (casesi): Emit bounds checking as RTL.
27623 (casesi_internal_mips16_<mode>): Remove bounds checking.
27625 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27627 * config/xtensa/xtensa.c (xtensa_option_override): Append
27628 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27629 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27630 xtensa_doloop_hooks): Define unconditionally.
27631 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27633 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27634 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27635 for it in xtensa_option_override.
27636 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27637 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27639 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27641 * doc/cppopts.texi: Document '-' special value to -MF.
27643 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27645 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27646 (cortex_a53_fconst): Likewise.
27647 (cortex_a53_fpmul): Likewise.
27648 (cortex_a53_f_load_64): Likewise.
27649 (cortex_a53_f_load_many): Likewise.
27650 (cortex_a53_advsimd_alu): Likewise.
27651 (cortex_a53_advsimd_alu_q): Likewise.
27652 (cortex_a53_advsimd_mul): Likewise.
27653 (cortex_a53_advsimd_mul_q): Likewise.
27654 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27655 Add missing fmul, r2f_cvt and fconst cases.
27657 2017-06-14 Richard Biener <rguenther@suse.de>
27659 PR middle-end/81088
27660 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27662 (fold_binary_loc): When associating do not treat pre-existing
27663 TREE_OVERFLOW on literal constants as a reason to allow
27664 TREE_OVERFLOW on associated literal constants.
27666 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27668 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27669 (MASK_FEATURES): New macro.
27670 * config/sparc/sparc.c (sparc_option_override): Remove the special
27671 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27673 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27675 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27676 a division of 0 if non-call exceptions are enabled.
27678 2017-06-14 Andrew Pinski <apinski@cavium.com>
27679 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27682 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27683 Improve vector initialization code gen for only variable case.
27685 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27687 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27689 2017-06-14 Richard Biener <rguenther@suse.de>
27691 PR tree-optimization/81083
27692 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27695 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27697 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27698 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27699 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27700 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27701 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27702 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27704 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27706 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27707 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27709 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27711 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27713 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27715 * config/rs6000/t-rtems: Don't handle SPE.
27717 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27719 * config/rs6000/t-linux: Don't handle SPE.
27721 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27723 * config/rs6000/eabispe.h: Delete file.
27725 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27727 * config/rs6000/t-spe: Delete file.
27729 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27731 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27732 (rs6000_legitimate_offset_address_p): Return false for anything in
27733 V2SImode or V2SFmode.
27735 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27737 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27738 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27739 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27741 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27742 (rs6000_legitimate_offset_address_p): Ditto.
27743 (rs6000_emit_move): Ditto.
27744 (rs6000_init_builtins): Remove V4HI_type_node.
27746 2017-06-13 Martin Liska <mliska@suse.cz>
27749 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27750 (gate_asan): Likewise.
27751 * asan.h (asan_no_sanitize_address_p): Remove the function.
27752 (sanitize_flags_p): New function.
27753 * builtins.def: Fix coding style.
27754 * common.opt: Use renamed enum value.
27755 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27756 * doc/extend.texi: Document no_sanitize attribute.
27757 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27758 to SANITIZE_UNDEFINED_NONDEFAULT.
27759 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27760 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27761 Use sanitize_flags_p.
27762 * gimplify.c (gimplify_function_tree): Likewise.
27763 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27764 * opts.c (parse_no_sanitize_attribute): New function.
27765 (common_handle_option): Use renamed enum value.
27766 * opts.h (parse_no_sanitize_attribute): Declare.
27767 * tree.c (sanitize_flags_p): New function.
27768 * tree.h: Declared here.
27769 * tsan.c: Use sanitize_flags_p.
27770 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27771 (instrument_mem_ref): Likewise.
27772 (instrument_bool_enum_load): Likewise.
27773 (do_ubsan_in_current_function): Remove the function.
27774 (pass_ubsan::execute): Use sanitize_flags_p.
27775 * ubsan.h: Remove do_ubsan_in_current_function
27776 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27777 (dump_function_to_file): Use it here.
27779 2017-06-13 Martin Jambor <mjambor@suse.cz>
27781 PR tree-optimization/80803
27782 PR tree-optimization/81063
27783 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27784 (propagate_subaccesses_across_link): Enqueue subtree whenever
27785 necessary instead of relying on the caller.
27787 2017-06-13 Martin Jambor <mjambor@suse.cz>
27789 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27790 that have a first_link.
27791 (sort_and_splice_var_accesses): Do not check first_link before
27793 (subtree_mark_written_and_enqueue): Likewise.
27794 (propagate_all_subaccesses): Likewise and do not stop at first
27795 parent with a first_link.
27797 2017-06-13 Martin Jambor <mjambor@suse.cz>
27799 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27802 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27804 * match.pd: New pattern.
27806 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27808 * tree-vrp.c (is_masked_range_test): New function.
27809 (register_edge_assert_for): Determine ranges for
27812 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27814 PR tree-optimization/67328
27815 * fold-const.c (maskable_range_p): New function.
27816 (build_range_check): Generate bittests if possible.
27818 2017-06-13 Martin Liska <mliska@suse.cz>
27820 * gimple-pretty-print.c (dump_probability): Add new argument.
27821 (dump_edge_probability): Dump both probability and count.
27822 (dump_gimple_label): Likewise.
27823 (dump_gimple_bb_header): Likewise.
27825 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27828 * config/avr/avr-devices.c: Fix indentation.
27829 * config/avr/gen-avr-mmcu-specs.c: Dito.
27831 2017-06-13 Richard Biener <rguenther@suse.de>
27833 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27834 instead get vector type from stmt_info.
27835 (vectorizable_reduction): Adjust. Remove dead code.
27837 2017-06-13 Richard Biener <rguenther@suse.de>
27839 PR middle-end/81065
27840 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27841 case of C * (x * C2 + C3).
27842 (fold_addr_of_array_ref_difference): Properly fold index difference.
27844 2017-06-12 David S. Miller <davem@davemloft.net>
27847 * config/sparc/sparc.md (return expander): Emit frame blockage if
27848 function uses alloca.
27850 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27852 * combine.c (make_field_assignment): Check len rather than the mode
27853 precision when calling force_to_mode.
27855 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27857 Support multilibs and devices that see flash in RAM address range.
27860 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27861 (avr_mcu_t) <flash_pm_offset>: New field.
27862 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27863 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27864 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27865 (AVR_TINY_PM_OFFSET): Remove macro.
27866 * config/avr/avr.opt (-mshort-calls): New option.
27867 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27868 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27869 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27870 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27871 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27872 instead of avr_arch->have_jmp_call.
27873 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27874 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27875 avr_arch->flash_pm_offset to define.
27876 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27877 new field flash_pm_offset. Add entry for avrxmega3.
27878 (avr_texinfo): Add entry for avrxmega3.
27879 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27880 attiny212, attiny214,
27881 attiny412, attiny414, attiny416, attiny417,
27882 attiny814, attiny816, attiny817,
27883 attiny1614, attiny1616, attiny1617,
27884 attiny3214, attiny3216, attiny3217.
27885 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27886 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27887 (avr_print_operand_address) [AVR_TINY]: Same.
27888 (avr_asm_init_sections) <readonly_data_section>: Only patch
27889 callback if avr_arch->flash_pm_offset = 0.
27890 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27891 for rodata if avr_arch->flash_pm_offset != 0.
27892 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27893 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27894 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27895 (m_options): Append opt_rcall.
27896 (m_dirnames): Append dir_rcall.
27897 * config/avr/t-multilib: Regenerate.
27899 * configure.ac [target=avr]: Check whether avrxmega3 default
27900 linker description file works as needed.
27901 * configure: Regenerate.
27902 * doc/avr-mmcu.texi: Regenerate.
27903 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27904 <__AVR_ARCH__>: Document avrxmega3 and 103.
27905 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27906 <__AVR_SHORT_CALLS__>: Document it.
27907 <__AVR_PM_BASE_ADDRESS__>: Document it.
27908 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27909 (AVR Variable Attributes) <progmem>: Document this is
27910 not needed for avrxmega3.
27911 (AVR Named Address Spaces) <__flash>: Dito.
27913 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27915 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27917 2017-06-12 Doug Rupp <rupp@adacore.com>
27919 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27920 Append vxworks-stdint.h to the tm_file list.
27921 * config/vxworks-stdint.h: New file.
27923 2017-06-12 Martin Liska <mliska@suse.cz>
27925 PR tree-optimization/81041
27926 * tree-profile.c (gimple_gen_ic_func_profiler):
27927 Create an extra BB in profile-generate
27928 (gimple_gen_time_profiler): Likewise.
27930 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27932 PR tree-optimization/81003
27933 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27934 (update_range_test): Use it instead of force_gimple_operand_gsi.
27936 2017-06-12 Richard Biener <rguenther@suse.de>
27938 PR tree-optimization/81053
27939 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27940 with backedge value not defined in loop. Simplify def stmt
27943 2017-06-11 Tom de Vries <tom@codesourcery.com>
27946 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27948 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27949 nvptx_cannot_force_const_mem.
27951 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27953 * opts.c (finish_options): Move test for flag_split_stack after
27954 it has been initialized.
27956 2017-06-11 Jason Merrill <jason@redhat.com>
27958 * tree.h (id_equal): New.
27959 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27960 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27961 instead of strcmp of IDENTIFIER_POINTER.
27963 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27965 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27966 (mark_all_inlined_calls_cdtor): Fix formating.
27967 (inline_transform): Rescale profile before inlining.
27969 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27971 * cgraph.h (cgraph_edge::clone): Update prototype.
27972 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27973 (cgraph_node::create_clone): Update.
27974 (cgraph_node::create_version_clone): Update.
27975 * tree-inline.c (copy_bb): Update.
27976 (expand_call_inline): Update.
27978 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27980 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27981 factored out from ...
27982 (rs6000_emit_prologue): ... here.
27984 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27986 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27987 factored out from ...
27988 (rs6000_emit_prologue): ... here.
27990 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27992 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27994 (handle_missing_profiles): Fix computation of tp_first_run.
27995 (counts_to_freqs): Do not touch freqs when count is 0.
27997 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27999 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28002 2017-06-10 Tom de Vries <tom@codesourcery.com>
28004 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28005 attributes): Document signal effective target.
28007 2017-06-10 Tom de Vries <tom@codesourcery.com>
28009 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28010 Document effective target stack_size.
28012 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28014 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28015 to the edit_context if they can be auto-applied.
28017 2017-06-9 Ian Lance Taylor <iant@golang.org>
28019 * opts.c (finish_options): If -fsplit-stack, disable implicit
28020 -forder-blocks-and-partition.
28021 * doc/invoke.texi (Optimize Options): Document that when using
28022 -fsplit-stack -forder-blocks-and-partition is not implicitly
28025 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28027 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28028 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28029 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28030 * builtins.def (abort, trap, unreachable): Declare cold.
28031 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28032 * tree-core.h (ECF_COLD): New.
28033 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28034 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28036 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28038 * predict.c (unlikely_executed_stmt_p): Cleanup.
28040 2017-06-09 Richard Biener <rguenther@suse.de>
28042 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28043 model if the ref is always written to.
28045 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28047 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28049 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28051 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28054 2017-06-09 Tom de Vries <tom@codesourcery.com>
28057 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28058 "target cannot support label values" when encountering LABEL_REF.
28060 2017-06-09 Martin Liska <mliska@suse.cz>
28062 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28063 (gimple_gen_ic_func_profiler): Emit direct comparison
28064 of __gcov_indirect_call_callee with NULL.
28065 (gimple_gen_time_profiler): Change probability from
28066 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28068 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28070 * profile.c (edge_gcov_counts): Turn to pointer.
28071 (compute_branch_probabilities, compute_branch_probabilities): Update.
28072 (branch_prob): Do not clear edge_gcov_count.
28073 * profile.h (edge_gcov_counts): Turn to pointer.
28074 (edge_gcov_count): Update.
28076 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28078 * gimple.h (gimple_check_failed): Mark cold.
28080 2017-06-09 Richard Biener <rguenther@suse.de>
28082 PR tree-optimization/66623
28083 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28084 refactor check_reduction into two parts, properly computing
28085 whether we have to check reduction validity for outer loop
28088 2017-06-09 Richard Biener <rguenther@suse.de>
28090 PR tree-optimization/79483
28091 * graphite-scop-detection.c (order): New global.
28092 (get_order): Compute bb to order mapping that satisfies code
28093 generation constraints.
28094 (cmp_pbbs): New helper.
28095 (build_scops): Start domwalk at entry block, sort generated
28098 2017-06-09 Richard Biener <rguenther@suse.de>
28100 PR middle-end/81007
28101 * ipa-polymorphic-call.c
28102 (ipa_polymorphic_call_context::restrict_to_inner_class):
28103 Skip FIELD_DECLs with error_mark_node type.
28104 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28107 2017-06-09 Martin Liska <mliska@suse.cz>
28109 * predict.c (struct branch_predictor): New struct.
28110 (test_prediction_value_range): New test.
28111 (predict_c_tests): New function.
28112 * selftest-run-tests.c (selftest::run_tests): Run the function.
28113 * selftest.h: Declare new tests.
28115 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28118 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28119 gen_add3_insn did not fail.
28120 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28121 r0, construct that number in a temporary reg and add that reg to r0.
28122 If asked to put the result in r0 as well, fail.
28124 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28126 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28127 for early expansion of vec_eqv.
28129 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28131 PR middle-end/81005
28132 * ubsan.c (instrument_null): Avoid pointless code temporary.
28133 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28135 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28139 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28142 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28145 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28147 * predict.c (unlikely_executed_edge_p): Move ahead.
28148 (probably_never_executed_edge_p): Use it.
28150 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28152 PR middle-end/79988
28153 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28154 gimple_call_builtin_p call.
28156 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28158 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28159 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28160 rtl_check_failed_type2, rtl_check_failed_code1,
28161 rtl_check_failed_code2, rtl_check_failed_code_mode,
28162 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28163 rtvec_check_failed_bounds, rtl_check_failed_flag,
28164 _fatal_insn_not_found, _fatal_insn): Likewise.
28165 * tree.h (tree_contains_struct_check_failed,
28166 tree_check_failed, tree_not_check_failed,
28167 tree_class_check_failed, tree_range_check_failed,
28168 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28169 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28170 tree_operand_check_failed, omp_clause_check_failed,
28171 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28174 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28176 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28177 flag_branch_probabilities.
28178 * ipa-inline.c (edge_badness): Likewise.
28179 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28180 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28181 * predict.c (maybe_hot_frequency_p): Likewise.
28182 (probably_never_executed): Likewise.
28183 * sched-ebb.c (schedule_ebbs): Likewise.
28184 * sched-rgn.c (find_single_block_region): Likewise.
28185 * tracer.c (tail_duplicate): Likewise.
28187 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28189 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28190 longer requires x_flag_profile_use.
28192 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28194 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28195 instead of flag_reorder_blocks_and_partition.
28196 * dbxout.c (dbxout_function_end): Likewise.
28197 * dwarf2out.c (gen_subprogram_die): Likewise.
28198 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28199 * hw-doloop.c (reorg_loops): Likewise.
28200 * varasm.c (assemble_start_function,
28201 assemble_end_function): Likewise.
28202 (decide_function_section): Do not check for
28203 flag_reorder_blocks_and_partition.
28205 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28207 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28209 (chkp_get_hard_register_fake_addr_expr): Ditto.
28210 (chkp_build_addr_expr): Add check for hard reg case.
28211 (chkp_parse_array_and_component_ref): Ditto.
28212 (chkp_find_bounds_1): Ditto.
28213 (chkp_process_stmt): Don't generate bounds store for
28216 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28218 * predict.c (maybe_hot_bb_p): Do not check profile status.
28219 (maybe_hot_edge_p): Likewise.
28220 (probably_never_executed): Check for zero counts even if profile
28222 (unlikely_executed_edge_p): New function.
28223 (unlikely_executed_stmt_p): New function.
28224 (unlikely_executed_bb_p): New function.
28225 (set_even_probabilities): Use unlikely predicates.
28226 (combine_predictions_for_bb): Likewise.
28227 (predict_paths_for_bb): Likewise.
28228 (predict_paths_leading_to_edge): Likewise.
28229 (determine_unlikely_bbs): New function.
28230 (estimate_bb_frequencies): Use it.
28231 (compute_function_frequency): Use zero counts even if profile is
28233 * profile-count.h: Fix typo.
28235 2017-08-08 Julia Koval <julia.koval@intel.com>
28237 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28238 _mm512_mask_cvtsepi16_storeu_epi8,
28239 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28240 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28241 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28242 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28243 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28244 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28245 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28246 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28247 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28248 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28249 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28250 __builtin_ia32_pmovuswb256mem_mask,
28251 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28252 __builtin_ia32_pmovwb512mem_mask): New builtins.
28254 2017-08-08 Julia Koval <julia.koval@intel.com>
28256 PR target/73350,80862
28257 * config/i386/subst.md (round): Fix round pattern.
28258 * config/i386/i386.c (ix86_erase_embedded_rounding):
28259 Fix erasing rounding for the fixed pattern.
28261 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28263 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28265 2017-06-08 Martin Liska <mliska@suse.cz>
28267 PR gcov-profile/80911
28268 * gcov.c (block_info::block_info): New constructor.
28270 2017-06-07 Carl Love <cel@us.ibm.com>
28272 * config/rs6000/rs6000-c: The return type of the following
28273 built-in functions was implemented as int not long long. Fix sign
28274 of return value for the unsigned version of vec_mulo and vec_mule.
28275 vector unsigned long long vec_bperm (vector unsigned long long,
28276 vector unsigned char)
28277 vector signed long long vec_mule (vector signed int,
28279 vector unsigned long long vec_mule (vector unsigned int,
28280 vector unsigned int)
28281 vector signed long long vec_mulo (vector signed int,
28283 vector unsigned long long vec_mulo (vector unsigned int,
28284 vector unsigned int)
28285 * doc/extend.texi: Fix the documentation for the built-in
28288 2017-06-07 Carl Love <cel@us.ibm.com>
28291 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28294 2017-06-07 Carl Love <cel@us.ibm.com>
28296 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28297 support, Generate doublehv for signed int/float for BE case only.
28299 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28301 * doc/invoke.texi (mcx16): Rewrite.
28303 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28305 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28306 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28307 *mov<mode>_softfloat, and an anonymous splitter): Use
28308 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28310 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28312 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28314 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28315 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28316 (rs6000_debug_reg_global): Adjust.
28317 (rs6000_init_hard_regno_mode_ok): Adjust.
28318 (rs6000_dbx_register_number): Adjust.
28319 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28320 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28321 Remove SPE_ACC and SPEFSCR.
28322 (REG_ALLOC_ORDER): Ditto.
28323 (FRAME_POINTER_REGNUM): Change to 111.
28324 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28325 (REG_CLASS_NAMES): Ditto.
28326 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28327 (REGISTER_NAMES): Ditto.
28328 (ADDITIONAL_REG_NAMES): Ditto.
28329 (rs6000_reg_names): Ditto.
28330 * config/rs6000/rs6000.md: Renumber some register number
28333 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28335 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28337 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28338 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28340 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28341 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28342 Delete the SPE high registers.
28343 (REG_ALLOC_ORDER): Ditto.
28344 (enum reg_class): Remove SPE_HIGH_REGS.
28345 (REG_CLASS_NAMES): Ditto.
28346 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28347 (REGISTER_NAMES): Ditto.
28348 (rs6000_reg_names): Ditto.
28349 * doc/tm.texi.in: Remove SPE as example.
28350 * doc/tm.texi: Regenerate.
28352 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28354 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28355 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28356 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28357 * config/rs6000/rs6000.md (type): Remove "brinc".
28359 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28361 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28362 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28363 * config/rs6000/linuxspe.h: Delete file.
28364 * config/rs6000/rs6000.md: Don't include spe.md.
28365 * config/rs6000/spe.h: Delete file.
28366 * config/rs6000/spe.md: Delete file.
28367 * config/rs6000/t-rs6000: Remove spe.md.
28369 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28371 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28372 (reg_or_none500mem_operand): Delete.
28373 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28374 instead of reg_or_none500mem_operand.
28376 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28378 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28379 handling of SPE flags.
28380 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28382 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28384 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28386 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28387 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28388 paired_divv2sf3): Similar.
28389 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28390 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28391 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28393 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28394 Rename the paired_* instruction patterns.
28395 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28397 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28398 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28399 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28400 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28401 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28402 PAIRED_VECTOR_MODE.
28403 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28404 (spe_func_has_64bit_regs_p): Delete.
28405 (spe_expand_predicate_builtin): Delete.
28406 (spe_expand_evsel_builtin): Delete.
28407 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28408 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28409 (invalid_e500_subreg): Delete.
28410 (rs6000_legitimize_address): Always force_reg op2 as well, for
28411 paired single memory accesses.
28412 (rs6000_member_type_forces_blk): Delete.
28413 (rs6000_spe_function_arg): Delete.
28414 (rs6000_expand_unop_builtin): Delete SPE handling.
28415 (rs6000_expand_binop_builtin): Ditto.
28416 (spe_expand_stv_builtin): Delete.
28417 (bdesc_2arg_spe): Delete.
28418 (spe_expand_builtin): Delete.
28419 (spe_expand_predicate_builtin): Delete.
28420 (spe_expand_evsel_builtin): Delete.
28421 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28422 (spe_init_builtins): Delete.
28423 (spe_func_has_64bit_regs_p): Delete.
28424 (savres_routine_name): Delete "info" parameter. Adjust callers.
28425 (rs6000_emit_stack_reset): Ditto.
28426 (rs6000_dwarf_register_span): Delete.
28427 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28428 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28429 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28431 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28433 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28434 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28435 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28436 mulv2sf3, divv2sf3): Delete expanders.
28438 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28440 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28442 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28444 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28445 * config/rs6000/rs6000.c: Ditto.
28447 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28449 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28450 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28451 comparison_operator.
28453 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28455 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28456 * config/rs6000/rs6000.opt: Ditto.
28457 * config/rs6000/t-rtems: Ditto.
28459 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28461 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28462 TARGET_E500_SINGLE by 0, simplify.
28463 * config/rs6000/rs6000.c: Ditto.
28464 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28465 (spe_build_register_parallel): Delete.
28466 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28467 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28468 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28469 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28470 (E500_CONVERT): Delete.
28471 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28473 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28475 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28476 * config/rs6000/dfp.md: Ditto.
28477 (negdd2, *negdd2_fpr): Merge.
28478 (absdd2, *absdd2_fpr): Merge.
28479 (negtd2, *negtd2_fpr): Merge.
28480 (abstd2, *abstd2_fpr): Merge.
28481 * config/rs6000/e500.h: Delete file.
28482 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28483 TARGET_FPRS by 1 and simplify.
28484 * config/rs6000/rs6000-c.c: Ditto.
28485 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28486 TARGET_DF_SPE by 0.
28487 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28489 * config/rs6000/rs6000.md: Ditto.
28490 (floatdidf2, *floatdidf2_fpr): Merge.
28491 (move_from_CR_gt_bit): Delete.
28492 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28493 (E500_CR_IOR_COMPARE): Delete.
28494 (All patterns that require !TARGET_FPRS): Delete.
28495 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28497 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28499 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28501 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28503 * graphds.c (add_edge): Intitialize edge's attached data.
28504 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28505 pointer parameter. Call pointed function on each edge during
28506 graph traversing. Skip traversing the edge when the function
28508 (graphds_dfs, graphds_scc): Ditto.
28509 (for_each_edge): New parameter. Pass the new parameter to callback
28511 * graphds.h (skip_edge_callback): New function pointer type.
28512 (graphds_dfs, graphds_scc): New function pointer parameter.
28513 (graphds_edge_callback, for_each_edge): New parameter.
28515 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28517 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28518 out code checking if runtime alias check is possible to below ...
28519 Call the new function.
28520 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28521 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28523 2017-06-07 Marek Polacek <polacek@redhat.com>
28526 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28527 TYPE_OVERFLOW_WRAPS check.
28529 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28531 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28532 if versioning is required.
28533 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28534 peeling with the check for versioning.
28536 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28538 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28539 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28540 Set true to new parameter if new ssa variable is defined.
28541 (vect_gen_vector_loop_niters): Refactor. Set range information
28542 for the new vector loop bound variable.
28543 (vect_do_peeling): Ditto.
28545 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28547 * tree-affine.c (ssa.h): Include header file.
28548 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28549 has wrapping overflow behavior.
28551 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28553 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28555 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28557 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28558 (tree_to_aff_combination): ... here.
28560 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28562 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28563 reg_pressure model function.
28564 (ivopts_global_cost_for_size): Delete.
28565 (determine_set_costs, iv_ca_recount_cost): Call new model function
28566 ivopts_estimate_reg_pressure.
28568 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28570 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28571 expensive than udiv. Remove floating point cases from mod.
28573 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28575 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28576 Increase idiv cost.
28578 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28580 * config/aarch64/aarch64.md
28581 (copysignsf3): Fix mask generation.
28583 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28585 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28587 (class dump_manager): Add register_dumps method.
28588 * dumpfile.c: Include langhooks.h.
28589 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28590 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28591 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28592 (dump_manager::dump_register): Start with 512 entries instead of 32.
28593 (dump_manager::register_dumps): New method.
28594 * toplev.c (general_init): Instead of invoking register_dumps
28595 langhook, invoke register_dumps method on the dump manager.
28596 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28599 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28601 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28602 Say that # requires an associated define_split to exist, and that
28603 the define_split must be suitable for use after register allocation.
28605 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28607 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28608 (compute_outgoing_frequencies): Also initialize zero counts.
28609 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28610 around loops; preserve more of profile when nothing changes.
28612 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28614 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28616 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28617 * config/arm/arm-cpu-cdata.h: Regenerate.
28618 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28619 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28620 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28621 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28622 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28624 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28626 * config/arm/t-rmprofile: Likewise.
28627 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28629 2017-06-06 David S. Miller <davem@davemloft.net>
28632 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28633 blockage if function uses alloca.
28635 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28637 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28638 New "uid" fields to hold pretty-print IDs of group and ref.
28639 Memory references are now identified as <group_id>:<ref_id>
28640 instead of using [random] addresses.
28641 (dump_mem_details): Simplify, no functional change.
28642 (dump_mem_ref): Simplify and make output more concise.
28643 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28644 (find_or_create_group): Initialize group uid.
28645 (record_ref): Initialize ref uid. Improve debug output.
28646 (prune_group_by_reuse, should_issue_prefetch_p,)
28647 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28648 (mark_nontemporal_store, determine_loop_nest_reuse):
28649 Improve debug output.
28651 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28653 * dbgcnt.def (prefetch): New debug counter.
28654 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28655 (schedule_prefetches): Stop issueing prefetches if debug counter
28658 2017-06-06 Tom de Vries <tom@codesourcery.com>
28660 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28661 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28663 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28665 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28666 Use aarch64_reg_or_zero predicate for operand 4.
28667 (aarch64_compare_and_swap<mode> define_insn_and_split):
28668 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28669 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28671 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28673 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28674 (arm_compute_save_core_reg_mask): This.
28675 (thumb1_compute_save_reg_mask): Rename into ...
28676 (thumb1_compute_save_core_reg_mask): This.
28677 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28678 (arm_compute_frame_layout): Likewise.
28680 2017-06-06 Richard Biener <rguenther@suse.de>
28682 PR tree-optimization/80974
28683 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28684 keep or clear leaders SSA info.
28686 2017-06-06 Tom de Vries <tom@codesourcery.com>
28688 * config/nvptx/nvptx.c (split_mode_p): New function.
28689 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28691 2017-06-06 Tom de Vries <tom@codesourcery.com>
28693 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28695 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28698 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28701 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28703 * shrink-wrap.c (handle_simple_exit): Update profile.
28704 (try_shrink_wrapping): Upate profile.
28706 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28708 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28709 (tree_guess_outgoing_edge_probabilities): New.
28710 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28711 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28713 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28715 * ipa-split.c (split_function): Initialize return bb profile.
28717 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28719 * profile.c (compute_branch_probabilities): Also initialize
28720 EXIT_BLOCK profile.
28722 2017-06-06 Richard Biener <rguenther@suse.de>
28724 PR tree-optimization/80928
28725 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28726 (vect_analyze_loop_operations): Properly guard analysis for
28728 (vect_transform_loop): Likewise.
28729 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28730 (vect_model_induction_cost): Do not cost for pure SLP.
28731 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28732 of induction in inner loop vectorization.
28733 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28734 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28735 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28737 (vect_analyze_slp_cost_1): Cost induction.
28738 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28739 (vect_get_slp_vect_defs): Likewise.
28740 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28741 (vect_transform_stmt): Handle SLP reductions.
28742 * tree-vectorizer.h (vectorizable_induction): Adjust.
28744 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28746 * config/rs6000/rs6000.c (make_resolver_func): Update
28747 init_lowered_empty_function call.
28749 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28751 * doc/invoke.texi: Document the -fprofile-abs-path option.
28752 * common.opt (fprofile-abs-path): New option.
28753 * gcov-io.h (gcov_write_filename): Declare.
28754 * gcov-io.c (gcov_write_filename): New function.
28755 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28756 * profile.c (output_location): Likewise.
28758 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28760 * shring-wrap.c: Revert accidental commit.
28762 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28764 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28766 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28768 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28770 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28771 profile in callgraph edge.
28772 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28773 (apply_scale): Likewise.
28774 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28775 Also scale profile when inlining function with zero profile.
28776 (initialize_cfun): Update exit block profile even when it is zero.
28777 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28778 when profile is read.
28780 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28782 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28783 (CLONE_*): New constants to define the processors we can generate
28784 code for with the target_clone attribute.
28785 (rs6000_clone_map): New array to identify which clone processors
28786 the current program is running on.
28787 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28788 target_clone attribute.
28789 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28790 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28791 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28792 (cpu_expand_builtin): Add support for target_clone attribute.
28793 (rs6000_valid_attribute_p): Allow "default" attribute.
28794 (get_decl_name): New debug function to simplify printing the
28795 current function name in debugging statements.
28796 (rs6000_clone_priority): New functions to support the target_clone
28797 attribute, and be able to generate code to switch between ISA 2.05
28798 through ISA 3.0 (power6 through power9).
28799 (rs6000_compare_version_priority): Likewise.
28800 (rs6000_get_function_versions_dispatcher): Likewise.
28801 (make_resolver_func): Likewise.
28802 (add_condition_to_bb): Likewise.
28803 (dispatch_function_versions): Likewise.
28804 (rs6000_generate_version_dispatcher_body): Likewise.
28805 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28806 (fusion_gpr_load_p): Fix a spacing issue.
28807 * doc/extend.texi (Common Function Attributes): Document that the
28808 PowerPC supports the target_clone attribute.
28810 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28812 * config/arm/arm.h: explain F symbol found in description of ARM
28813 register allocation in its legend.
28815 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28817 * config/mips/frame-header-opt.c: Include profile-count.h.
28818 * config/riscv/riscv.c: Include profile-count.h
28820 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28822 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28824 (sm_set_flag_if_changed): Add bbs field.
28825 (execute_sm_if_changed_flag_set): Pass BBS.
28826 (execute_sm): Update.
28828 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28830 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28833 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28835 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28836 (peephole2): New peephole2 to emit the above.
28837 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28839 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28841 * config/aarch64/aarch64.c (define_peephole2 above
28842 *sub_<shift>_<mode>): New peephole.
28844 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28846 * config/i386/i386.c (make_resolver_func): Update.
28847 * Makefile.in: Add profile-count.h and profile-count.o
28848 * auto-profile.c (afdo_indirect_call): Update to new API.
28849 (afdo_set_bb_count): Update.
28850 (afdo_propagate_edge): Update.
28851 (afdo_propagate_circuit): Update.
28852 (afdo_calculate_branch_prob): Update.
28853 (afdo_annotate_cfg): Update.
28854 * basic-block.h: Include profile-count.h
28855 (struct edge_def): Turn count to profile_count.
28856 (struct basic_block_def): Likewie.
28857 (REG_BR_PROB_BASE): Move to profile-count.h
28858 (RDIV): Move to profile-count.h
28859 * bb-reorder.c (max_entry_count): Turn to profile_count.
28860 (find_traces): Update.
28861 (rotate_loop):Update.
28862 (connect_traces):Update.
28863 (sanitize_hot_paths):Update.
28864 * bt-load.c (migrate_btr_defs): Update.
28865 * cfg.c (RDIV): Remove.
28866 (init_flow): Use alloc_block.
28867 (alloc_block): Uninitialize count.
28868 (unchecked_make_edge): Uninitialize count.
28869 (check_bb_profile): Update.
28870 (dump_edge_info): Update.
28871 (dump_bb_info): Update.
28872 (update_bb_profile_for_threading): Update.
28873 (scale_bbs_frequencies_int): Update.
28874 (scale_bbs_frequencies_gcov_type): Update.
28875 (scale_bbs_frequencies_profile_count): New.
28876 * cfg.h (update_bb_profile_for_threading): Update.
28877 (scale_bbs_frequencies_profile_count): Declare.
28878 * cfgbuild.c (compute_outgoing_frequencies): Update.
28879 (find_many_sub_basic_blocks): Update.
28880 * cfgcleanup.c (try_forward_edges): Update.
28881 (try_crossjump_to_edge): Update.
28882 * cfgexpand.c (expand_gimple_tailcall): Update.
28883 (construct_exit_block): Update.
28884 * cfghooks.c (verify_flow_info): Update.
28885 (dump_bb_for_graph): Update.
28886 (split_edge): Update.
28887 (make_forwarder_block): Update.
28888 (duplicate_block): Update.
28889 (account_profile_record): Update.
28890 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28891 (get_estimated_loop_iterations): Update.
28892 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28893 (single_likely_exit): Update.
28894 * cfgloopmanip.c (scale_loop_profile): Update.
28896 (set_zero_probability): Update.
28897 (lv_adjust_loop_entry_edge): Update.
28898 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28899 (purge_dead_edges): Update.
28900 (rtl_account_profile_record): Update.
28901 * cgraph.c (cgraph_node::create): Uninitialize count.
28902 (symbol_table::create_edge): Uninitialize count.
28903 (cgraph_update_edges_for_call_stmt_node): Update.
28904 (cgraph_edge::dump_edge_flags): Update.
28905 (cgraph_node::dump): Update.
28906 (cgraph_edge::maybe_hot_p): Update.
28907 * cgraph.h: Include profile-count.h
28908 (create_clone), create_edge, create_indirect_edge): Update.
28909 (cgraph_node): Turn count to profile_count.
28910 (cgraph_edge0: Likewise.
28911 (make_speculative, clone): Update.
28912 (create_edge): Update.
28913 (init_lowered_empty_function): Update.
28914 * cgraphclones.c (cgraph_edge::clone): Update.
28915 (duplicate_thunk_for_node): Update.
28916 (cgraph_node::create_clone): Update.
28917 * cgraphunit.c (cgraph_node::analyze): Update.
28918 (cgraph_node::expand_thunk): Update.
28919 * final.c (dump_basic_block_info): Update.
28920 * gimple-streamer-in.c (input_bb): Update.
28921 * gimple-streamer-out.c (output_bb): Update.
28922 * graphite.c (print_global_statistics): Update.
28923 (print_graphite_scop_statistics): Update.
28924 * hsa-brig.c: Include basic-block.h.
28925 * hsa-dump.c: Include basic-block.h.
28926 * hsa-gen.c (T sum_slice): Update.
28927 (convert_switch_statements):Update.
28928 * hsa-regalloc.c: Include basic-block.h.
28929 * ipa-chkp.c (chkp_produce_thunks): Update.
28930 * ipa-cp.c (struct caller_statistics): Update.
28931 (init_caller_stats): Update.
28932 (gather_caller_stats): Update.
28933 (ipcp_cloning_candidate_p): Update.
28934 (good_cloning_opportunity_p): Update.
28935 (get_info_about_necessary_edges): Update.
28936 (dump_profile_updates): Update.
28937 (update_profiling_info): Update.
28938 (update_specialized_profile): Update.
28939 (perhaps_add_new_callers): Update.
28940 (decide_about_value): Update.
28941 (ipa_cp_c_finalize): Update.
28942 * ipa-devirt.c (struct odr_type_warn_count): Update.
28943 (struct decl_warn_count): Update.
28944 (struct final_warning_record): Update.
28945 (possible_polymorphic_call_targets): Update.
28946 (ipa_devirt): Update.
28947 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28948 * ipa-icf.c (sem_function::merge): Update.
28949 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28950 * ipa-inline.c (compute_uninlined_call_time): Update.
28951 (compute_inlined_call_time): Update.
28952 (want_inline_small_function_p): Update.
28953 (want_inline_self_recursive_call_p): Update.
28954 (edge_badness): Update.
28955 (lookup_recursive_calls): Update.
28956 (recursive_inlining): Update.
28957 (inline_small_functions): Update.
28958 (dump_overall_stats): Update.
28959 (dump_inline_stats): Update.
28960 * ipa-profile.c (ipa_profile_generate_summary): Update.
28961 (ipa_propagate_frequency): Update.
28962 (ipa_profile): Update.
28963 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28964 * ipa-utils.c (ipa_merge_profiles): Update.
28965 * loop-doloop.c (doloop_modify): Update.
28966 * loop-unroll.c (report_unroll): Update.
28967 (unroll_loop_runtime_iterations): Update.
28968 * lto-cgraph.c (lto_output_edge): Update.
28969 (lto_output_node): Update.
28970 (input_node): Update.
28971 (input_edge): Update.
28972 (merge_profile_summaries): Update.
28973 * lto-streamer-in.c (input_cfg): Update.
28974 * lto-streamer-out.c (output_cfg): Update.
28975 * mcf.c (create_fixup_graph): Update.
28976 (adjust_cfg_counts): Update.
28977 (sum_edge_counts): Update.
28978 * modulo-sched.c (sms_schedule): Update.
28979 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28980 * predict.c (maybe_hot_count_p): Update.
28981 (probably_never_executed): Update.
28982 (dump_prediction): Update.
28983 (combine_predictions_for_bb): Update.
28984 (propagate_freq): Update.
28985 (handle_missing_profiles): Update.
28986 (counts_to_freqs): Update.
28987 (rebuild_frequencies): Update.
28988 (force_edge_cold): Update.
28989 * predict.h: Include profile-count.h
28990 (maybe_hot_count_p, counts_to_freqs): UPdate.
28991 * print-rtl-function.c: Do not include cfg.h
28992 * print-rtl.c: Include basic-block.h
28993 * profile-count.c: New file.
28994 * profile-count.h: New file.
28995 * profile.c (is_edge_inconsistent): Update.
28996 (correct_negative_edge_counts): Update.
28997 (is_inconsistent): Update.
28998 (set_bb_counts): Update.
28999 (read_profile_edge_counts): Update.
29000 (compute_frequency_overlap): Update.
29001 (compute_branch_probabilities): Update; Initialize and deinitialize
29003 (branch_prob): Update.
29004 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29005 (edge_gcov_count): New.
29006 (bb_gcov_count): New.
29007 * shrink-wrap.c (try_shrink_wrapping): Update.
29008 * tracer.c (better_p): Update.
29009 * trans-mem.c (expand_transaction): Update.
29010 (ipa_tm_insert_irr_call): Update.
29011 (ipa_tm_insert_gettmclone_call): Update.
29012 * tree-call-cdce.c: Update.
29013 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29014 (gimple_duplicate_sese_tail): Update.
29015 (gimple_account_profile_record): Update.
29016 (execute_fixup_cfg): Update.
29017 * tree-inline.c (copy_bb): Update.
29018 (copy_edges_for_bb): Update.
29019 (initialize_cfun): Update.
29020 (freqs_to_counts): Update.
29021 (copy_cfg_body): Update.
29022 (expand_call_inline): Update.
29023 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29024 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29025 (try_unroll_loop_completely): Update.
29026 (try_peel_loop): Update.
29027 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29028 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29029 * tree-ssa-loop-split.c (connect_loops): Update.
29030 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29031 * tree-ssa-reassoc.c (branch_fixup): Update.
29032 * tree-ssa-tail-merge.c (replace_block_by): Update.
29033 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29034 (compute_path_counts): Update.
29035 (update_profile): Update.
29036 (recompute_probabilities): Update.
29037 (update_joiner_offpath_counts): Update.
29038 (estimated_freqs_path): Update.
29039 (freqs_to_counts_path): Update.
29040 (clear_counts_path): Update.
29041 (ssa_fix_duplicate_block_edges): Update.
29042 (duplicate_thread_path): Update.
29043 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29044 (struct switch_conv_info): Update.
29045 * tree-tailcall.c (decrease_profile): Update.
29046 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29047 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29048 * value-prof.c (check_counter): Update.
29049 (gimple_divmod_fixed_value): Update.
29050 (gimple_mod_pow2): Update.
29051 (gimple_mod_subtract): Update.
29052 (gimple_ic_transform): Update.
29053 (gimple_stringop_fixed_value): Update.
29054 * value-prof.h (gimple_ic): Update.
29056 2017-06-02 Carl Love <cel@us.ibm.com>
29058 * config/rs6000/rs6000-c: Add support for built-in functions
29059 vector double vec_doublee (vector signed int);
29060 vector double vec_doublee (vector unsigned int);
29061 vector double vec_doublee (vector float);
29062 vector double vec_doubleh (vector signed int);
29063 vector double vec_doubleh (vector unsigned int);
29064 vector double vec_doubleh (vector float);
29065 vector double vec_doublel (vector signed int);
29066 vector double vec_doublel (vector unsigned int);
29067 vector double vec_doublel (vector float);
29068 vector double vec_doubleo (vector signed int);
29069 vector double vec_doubleo (vector unsigned int);
29070 vector double vec_doubleo (vector float);.
29071 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29072 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29074 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29075 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29076 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29078 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29079 vec_doublel, vec_doubleh.
29080 * doc/extend.texi: Update the built-in documentation file for the
29081 new built-in functions.
29083 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29086 * ipa-inline-analysis.c (free_growth_caches): Set
29087 edge_removal_hook_holder to NULL after removing it.
29089 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29091 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29092 comparision with zero.
29094 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29095 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29096 for early expansion of vec_min and vec_max builtins.
29097 (builtin_function_type): Add min/max unsigned variants to those
29098 identified as having unsigned arguments.
29100 2017-06-02 Olivier Hainque <hainque@adacore.com>
29102 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29104 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29106 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29107 Use VALL_F16 iterator rather than VALL.
29109 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29111 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29112 Emit CBNZ inside loop when doing a strong exchange and comparing
29113 against zero. Generate the CC flags after the loop.
29115 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29117 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29118 (dl_section_ref): New.
29119 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29120 On AIX, append an expression to subtract the size of the
29121 section length to dl_section_ref.
29123 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29125 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29126 for early expansion of vector absolute builtins.
29128 2017-06-02 Richard Biener <rguenther@suse.de>
29130 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29131 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29133 2017-06-02 Richard Biener <rguenther@suse.de>
29135 PR tree-optimization/80948
29136 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29137 stmt order as well.
29139 2017-06-02 Richard Biener <rguenther@suse.de>
29141 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29143 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29144 for inductions as relevant.
29146 2017-06-02 Richard Biener <rguenther@suse.de>
29148 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29149 (vectorizable_induction): ... this. Remove dead code.
29151 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29153 * builtins. (expand_builtin_alloca): Remove second parameter and
29154 infer its value from the first parameter instead.
29155 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29157 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29159 PR rtl-optimization/80903
29160 * loop-doloop.c (add_test): Unshare sequence.
29162 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29164 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29166 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29168 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29170 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29171 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29172 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29173 (xlogue_layout::get_stub_rtx): Make static.
29174 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29175 (xlogue_layout::compute_stub_managed_regs): Rename to...
29176 (xlogue_layout::count_stub_managed_regs): ...this.
29177 (xlogue_layout::is_stub_managed_reg): New function.
29178 (xlogue_layout::m_stub_names): Rename to...
29179 (xlogue_layout::s_stub_names): ...this, make static.
29180 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29181 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29182 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29183 xlogue_layout::s_stub_names): Instantiate statics.
29184 (stub_managed_regs): Remove.
29185 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29186 (disable_call_ms2sysv_xlogues): Rename to...
29187 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29188 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29190 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29191 change after reload_completed.
29192 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29194 (ix86_expand_prologue): Likewise.
29195 (ix86_expand_epilogue): Likewise.
29196 (ix86_expand_split_stack_prologue): Likewise.
29197 (ix86_compute_frame_layout): Remove frame parameter ...
29198 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29199 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29201 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29202 (ix86_frame): Move from here ...
29203 * config/i386/i386.h (ix86_frame): ... to here.
29204 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29205 complete ix86_frame data structure instead. Remove some_ld_name.
29207 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29209 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29210 symbols that hold a DECL_VALUE_EXPR.
29212 2017-06-01 Martin Jambor <mjambor@suse.cz>
29214 PR tree-optimization/80898
29215 * tree-sra.c (process_subtree_disqualification): Removed.
29216 (disqualify_candidate): Do not acll
29217 process_subtree_disqualification.
29218 (subtree_mark_written_and_enqueue): New function.
29219 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29220 RHS has been disqualified and re-queue LHS if necessary. Apart
29221 from that, ignore disqualified RHS.
29223 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29225 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29226 address fetch for z10 or later.
29228 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29230 * config/arc/arc.md (tst_movb): Add guard when splitting.
29232 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29234 * config/arc/arc.c (arc_can_eliminate): Test against
29235 arc_frame_pointer_needed.
29237 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29239 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29240 to prevent store reordering.
29241 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29242 (type): Add block type.
29243 (stack_tie): Define special instruction to be used in
29246 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29248 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29249 constraint. It is not valid for the pattern.
29250 (noncommutative_binary_comparison): Likewise.
29252 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29254 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29257 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29259 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29260 be used by the reg-alloc.
29262 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29264 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29265 reg-alloc when having mul64 or mul32x16 instructions.
29266 (mulsidi3): Likewise.
29267 (umulsidi3): Likewise.
29268 (mulsi32x16): New pattern.
29269 (mulsi64): Likewise.
29270 (mulsidi64): Likewise.
29271 (umulsidi64): Likewise.
29272 (MUL32x16_REG): Define.
29273 (mul64_600): Use MUL32x16_REG.
29274 (mac64_600): Likewise.
29275 (umul64_600): Likewise.
29276 (umac64_600): Likewise.
29278 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29280 * config/arc/arc.md (mulsi3_700): Make it commutative.
29282 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29284 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29286 (*sign_extendsidi2_insn): Likewise for movstosw.
29288 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29290 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29291 the type of the input discriminant value. Convert the
29292 discriminant value of signedness vary.
29294 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29296 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29297 Add to -Wall section.
29299 2017-06-01 Richard Biener <rguenther@suse.de>
29301 PR middle-end/66313
29302 * fold-const.c (fold_plusminus_mult_expr): If the factored
29303 factor may be zero use a wrapping type for the inner operation.
29304 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29305 and handle moved defs.
29306 (process_assignment): Properly guard the unary op case. Return a
29307 tri-state indicating that moving the stmt before the call may allow
29308 to continue. Pass through to_move.
29309 (find_tail_calls): Handle moving unrelated defs before
29312 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29315 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29316 splitter result in the canonical way.
29318 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29320 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29321 also for 32bit target. Update insn attributes.
29322 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29324 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29326 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29327 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29328 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29329 (_mm_maskz_min_ss): New intrinsics.
29331 2017-05-31 Martin Liska <mliska@suse.cz>
29333 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29334 Change comment style to one we normally use.
29335 (vectorizable_reduction): Likewise.
29336 (vectorizable_induction): Likewise.
29337 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29338 (vectorizable_call): Likewise.
29339 (vectorizable_simd_clone_call): Likewise.
29340 (vectorizable_conversion): Likewise.
29341 (vectorizable_assignment): Likewise.
29342 (vectorizable_shift): Likewise.
29343 (vectorizable_operation): Likewise.
29344 (vectorizable_store): Likewise.
29345 (vectorizable_load): Likewise.
29346 * tree-vectorizer.h: Likewise.
29348 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29350 * passes.c (emergency_dump_function): New.
29351 * tree-pass.h (emergency_dump_function): Declare.
29352 * plugin.c (plugins_internal_error_function): Remove.
29353 * plugin.h (plugins_internal_error_function): Remove declaration.
29354 * toplev.c (internal_error_function): New static function. Use it...
29355 (general_init): ...here.
29357 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29359 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29360 (arc_rtx_costs): Add costs for new patterns.
29361 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29362 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29364 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29366 * tree-ssa-strlen.c (get_next_strinfo): New function.
29367 (get_stridx_plus_constant): Use it.
29368 (zero_length_string): Likewise.
29369 (adjust_related_strinfos): Likewise.
29370 (adjust_last_stmt): Likewise.
29372 2017-05-31 Richard Biener <rguenther@suse.de>
29375 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29376 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29378 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29380 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29381 loop_vinfo argument and use of dependence distance vectors.
29382 Check instead whether the two references differ only in their
29383 initial value and assume that they have the same alignment if the
29384 difference is a multiple of the vector alignment.
29385 (vect_analyze_data_refs_alignment): Update call accordingly.
29387 2017-05-31 Martin Liska <mliska@suse.cz>
29390 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29392 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29394 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29395 (create_intersect_range_checks): Move from ...
29396 * tree-data-ref.c (create_intersect_range_checks_index)
29397 (create_intersect_range_checks): ... to here.
29398 (create_runtime_alias_checks): New function factored from ...
29399 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29400 here. Call above function.
29401 * tree-data-ref.h (create_runtime_alias_checks): New function.
29403 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29405 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29406 segment length for dr_b and compute it in wide_int.
29408 2017-05-31 Richard Biener <rguenther@suse.de>
29410 PR tree-optimization/80906
29411 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29412 and pass through iv_map.
29413 (copy_bb_and_scalar_dependences): Adjust.
29414 (translate_pending_phi_nodes): Likewise.
29415 (copy_loop_close_phi_args): Handle code-generating IVs instead
29418 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29420 * diagnostic-color.c (color_dict): Add "type-diff".
29421 (parse_gcc_colors): Update comment.
29422 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29423 -fdiagnostics-show-template-tree and -fno-elide-type.
29424 (GCC_COLORS): Add type-diff to example.
29426 (-fdiagnostics-show-template-tree): New.
29427 (-fno-elide-type): New.
29428 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29429 the pp_format_decoder callback. Call any m_format_postprocessor's
29431 (pretty_printer::pretty_printer): Initialize
29432 m_format_postprocessor.
29433 (pretty_printer::~pretty_printer): Delete any
29434 m_format_postprocessor.
29435 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29436 (class format_postprocessor): New class.
29437 (struct pretty_printer::format_decoder): Document the new parameters.
29438 (struct pretty_printer::m_format_postprocessor): New field.
29439 * tree-diagnostic.c (default_tree_printer): Update for new
29440 bool and const char ** params.
29441 * tree-diagnostic.h (default_tree_printer): Likewise.
29443 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29445 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29446 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29447 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29448 rs6000_gen_cell_microcode code.
29449 (rs6000_final_prescan_insn): Delete.
29450 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29451 "warn-cell-microcode" entries.
29452 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29453 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29454 throughout. Change cc_reg_not_micro_cr0_operand to
29455 cc_reg_not_cr0_operand throughout.
29456 (*extendhi<mode>2_noload): Delete.
29457 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29458 (mwarn-cell-microcode): Delete.
29459 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29460 -mgen-cell-microcode and -mwarn-cell-microcode.
29462 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29465 * config/i386/constraints.md (Yd): New constraint.
29467 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29468 and (?Yd, r) alternatives. Update insn attributes.
29469 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29470 and (?*Yd, r) alternatives. Update insn attributes.
29471 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29473 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29475 * gimplify.c (gimplify_modify_expr): Don't create a
29476 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29479 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29481 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29483 2017-05-30 Richard Biener <rguenther@suse.de>
29485 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29486 and reduc_def fields.
29487 (STMT_VINFO_REDUC_TYPE): New define.
29488 (STMT_VINFO_REDUC_DEF): Likewise.
29489 (vect_force_simple_reduction): Adjust prototype.
29490 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29491 (vect_is_simple_reduction): Remove check_reduction argument.
29492 (vect_force_simple_reduction): Adjust and set
29493 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29494 (vectorizable_reduction): Do not re-do reduction analysis
29495 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29496 * tree-parloops.c (gather_scalar_reductions): Adjust.
29498 2017-05-30 Richard Biener <rguenther@suse.de>
29500 PR middle-end/80901
29501 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29504 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29506 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29507 Introduce unknown_misalignment parameter and remove vf.
29508 (vect_peeling_hash_get_lowest_cost):
29509 Pass unknown_misalignment parameter.
29510 (vect_enhance_data_refs_alignment):
29511 Fix unsupportable data ref treatment.
29513 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29515 * tree-vect-data-refs.c (vect_get_data_access_cost):
29516 Workaround for SLP handling.
29517 (vect_enhance_data_refs_alignment):
29518 Compute costs for doing no peeling at all, compare to the best
29519 peeling costs so far and avoid peeling if cheaper.
29521 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29523 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29524 Return peeling info and set costs to zero for unlimited cost
29526 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29527 with unknown misalignment. Compute and costs for unknown
29528 misalignment, compare them to the costs for known misalignment
29529 and choose the cheapest for peeling.
29531 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29533 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29534 (vect_get_peeling_costs_all_drs): Create function.
29535 (vect_peeling_hash_get_lowest_cost):
29536 Use vect_get_peeling_costs_all_drs.
29537 (vect_peeling_supportable): Create function.
29538 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29540 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29542 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29543 DR_HAS_NEGATIVE_STEP.
29544 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29545 (vect_enhance_data_refs_alignment): Use.
29546 (vect_duplicate_ssa_name_ptr_info): Use.
29547 * tree-vectorizer.h (dr_misalignment): Use.
29548 (known_alignment_for_access_p): Use.
29550 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29553 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29555 (has_section_name): New function.
29557 2017-05-30 Martin Liska <mliska@suse.cz>
29560 * auto-profile.c (get_function_decl_from_block): Fix
29563 2017-05-30 Richard Biener <rguenther@suse.de>
29565 PR middle-end/80876
29566 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29568 2017-05-30 Martin Liska <mliska@suse.cz>
29570 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29571 * dumpfile.h (struct dump_file_info): Remove ctors.
29573 2017-05-30 Martin Liska <mliska@suse.cz>
29575 * predict.def: Fix GNU coding style.
29577 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29579 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29580 Mark 'to' argument with ATTRIBUTE_UNUSED.
29582 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29584 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29585 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29586 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29589 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29591 * doc/install.texi (Options specification): Restore entry of
29592 --enable-sjlj-exceptions.
29594 2017-05-27 Michael Eager <eager@eagercon.com>
29597 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29599 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29601 * config/microblaze/microblaze.h
29602 (FIXED_REGISTERS): Update in macro.
29603 (CALL_USED_REGISTERS): Update in macro.
29605 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29607 * doc/install.texi: Add links to macOS binary distributions.
29609 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29613 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29615 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29617 2017-05-26 Martin Liska <mliska@suse.cz>
29619 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29621 2017-05-26 Martin Liska <mliska@suse.cz>
29623 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29624 always leading ';; '.
29625 (dump_bb_info): Likewise.
29626 (brief_dump_cfg): Likewise.
29627 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29628 * dumpfile.c: Remove usage of TDF_VERBOSE.
29629 * dumpfile.h (enum dump_kind): Likewise.
29630 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29631 * print-tree.c (debug_verbose): Remove.
29632 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29633 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29634 * tree-diagnostic.c (default_tree_printer): Replace
29635 TDF_DIAGNOSTIC with TDF_SLIM.
29637 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29639 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29640 in parameter loop, rather than loop_vinfo.
29641 (create_intersect_range_checks): Ditto.
29642 (vect_create_cond_for_alias_checks): Update call to above functions.
29644 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29646 PR tree-optimization/80815
29647 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29648 for merging runtime alias checks. Handle negative DR_STEPs.
29650 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29652 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29654 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29655 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29656 out code pruning runtime alias checks.
29657 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29658 factored out from above.
29659 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29661 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29663 (prune_runtime_alias_test_list): New decalaration.
29665 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29667 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29668 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29669 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29670 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29671 (operator==, comp_dr_with_seg_len_pair): Ditto.
29672 (vect_prune_runtime_alias_test_list): Ditto.
29674 2017-05-26 Martin Liska <mliska@suse.cz>
29677 * params.def: Bound partial-inlining-entry-probability param.
29679 2017-05-26 Marek Polacek <polacek@redhat.com>
29682 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29685 2017-05-26 Richard Biener <rguenther@suse.de>
29687 PR tree-optimization/80842
29688 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29691 2017-05-26 Richard Biener <rguenther@suse.de>
29693 PR tree-optimization/80844
29694 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29696 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29698 * doc/md.texi (Machine Constraints): Update x86 family
29699 machine constraints section to match 'config/i386/constraints.md'.
29701 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29703 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29705 2017-05-25 Nathan Sidwell <nathan@acm.org>
29707 * doc/invoke.texi (--enable-languages): Update documentation.
29709 2017-05-25 Martin Liska <mliska@suse.cz>
29711 * dumpfile.c: Add TDF_FOLDING.
29712 * dumpfile.h (enum dump_kind): Likewise.
29713 * genmatch.c (dt_simplify::gen_1): Use it.
29715 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29717 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29719 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29721 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29722 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29724 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29726 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29727 * match.pd (X == C): Rewrite it here.
29728 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29729 with_certain_nonzero_bits2): New predicates.
29730 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29732 2017-05-24 Nathan Sidwell <nathan@acm.org>
29734 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29737 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29740 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29742 * config/powerpcspe: New port. Files are copied from the rs6000
29743 port, with "rs6000" in filenames replaced by "powerpcspe".
29745 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29747 PR rtl-optimization/80754
29748 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29750 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29752 * config/sparc/sparc.md (length): Return the correct value for -mflat
29753 sibcalls to match output_sibcall.
29755 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29759 * config/rs6000/rs6000.c (struct machine_function): Add new field
29761 (rs6000_get_separate_components): Init that field, use it.
29762 (rs6000_components_for_bb): Use the field.
29764 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29766 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29768 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29770 PR middle-end/80823
29771 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29773 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29776 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29777 address against address_operand predicate.
29778 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29780 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29782 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29783 parameters passed indirectly.
29785 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29787 * config/i386/i386.md (*movdi_internal): Remove SSE4
29788 alternative 18 (?r, *v). Update insn attributes.
29789 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29790 Update insn attributes.
29791 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29792 Update insn attributes.
29793 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29794 alternative 1 (r, v). Remove isa attribute.
29795 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29796 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29797 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29799 2017-05-23 Tom de Vries <tom@codesourcery.com>
29801 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29804 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29806 * cgraphunit.c (symbol_table::process_new_functions): Update.
29807 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29808 (inline_generate_summary): Rename to ...
29809 (ipa_fn_summary_generate): ... this one.
29810 (inline_read_summary): Rename to ...
29811 (ipa_fn_summary_read): ... this one.
29812 (inline_write_summary): Rename to ...
29813 (ipa_fn_summary_write): ... this one.
29814 (inline_free_summary): Rename to ...
29815 (ipa_free_fn_summary): ... this one.
29816 (pass_data_local_fn_summary, pass_local_fn_summary,
29817 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29818 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29819 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29820 make_pass_ipa_fn_summary): New.
29821 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29822 inline_write_summary, inline_free_summary): Remove.
29823 (ipa_free_fn_summary) : New.
29824 * ipa-inline.c (ipa_inline): Update.
29825 (pass_ipa_inline): Do not generate summaries.
29826 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29828 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29829 and add pass_ipa_fn_summary.
29830 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29832 (make_pass_inline_parameters): Remove.
29834 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29836 * omp-low.c (struct omp_context): Remove "default_kind" member.
29839 * omp-offload.c (execute_oacc_device_lower): Remove the
29840 parallelism dimensions function attributes for unparallelized
29841 OpenACC kernels constructs.
29843 2017-05-23 Martin Liska <mliska@suse.cz>
29845 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29847 (cgraph_edge::make_speculative): Likewise.
29848 (cgraph_edge::resolve_speculation): Likewise.
29849 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29850 (cgraph_node::dump): Likewise.
29851 * cgraph.h: Likewise.
29852 * cgraphunit.c (analyze_functions): Likewise.
29853 (symbol_table::compile): Likewise.
29854 * ipa-cp.c (print_all_lattices): Likewise.
29855 (determine_versionability): Likewise.
29856 (initialize_node_lattices): Likewise.
29857 (ipcp_verify_propagated_values): Likewise.
29858 (estimate_local_effects): Likewise.
29859 (update_profiling_info): Likewise.
29860 (create_specialized_node): Likewise.
29861 (perhaps_add_new_callers): Likewise.
29862 (decide_about_value): Likewise.
29863 (decide_whether_version_node): Likewise.
29864 (identify_dead_nodes): Likewise.
29865 (ipcp_store_bits_results): Likewise.
29866 * ipa-devirt.c (dump_targets): Likewise.
29867 (ipa_devirt): Likewise.
29868 * ipa-icf.c (sem_item::dump): Likewise.
29869 (sem_function::equals): Likewise.
29870 (sem_variable::equals): Likewise.
29871 (sem_item_optimizer::read_section): Likewise.
29872 (sem_item_optimizer::execute): Likewise.
29873 (congruence_class::dump): Likewise.
29874 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29875 (dump_inline_summary): Likewise.
29876 (estimate_node_size_and_time): Likewise.
29877 (inline_analyze_function): Likewise.
29878 * ipa-inline-transform.c (inline_call): Likewise.
29879 * ipa-inline.c (report_inline_failed_reason): Likewise.
29880 (want_early_inline_function_p): Likewise.
29881 (edge_badness): Likewise.
29882 (update_edge_key): Likewise.
29883 (inline_small_functions): Likewise.
29884 * ipa-profile.c (ipa_profile): Likewise.
29885 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29886 (ipa_make_edge_direct_to_target): Likewise.
29887 (remove_described_reference): Likewise.
29888 (ipa_impossible_devirt_target): Likewise.
29889 (propagate_controlled_uses): Likewise.
29890 (ipa_print_node_params): Likewise.
29891 (ipcp_transform_function): Likewise.
29892 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29893 (propagate_pure_const): Likewise.
29894 * ipa-reference.c (generate_summary): Likewise.
29895 (read_write_all_from_decl): Likewise.
29896 (propagate): Likewise.
29897 (ipa_reference_read_optimization_summary): Likewise.
29898 * ipa-utils.c (ipa_merge_profiles): Likewise.
29899 * ipa.c (walk_polymorphic_call_targets): Likewise.
29900 (symbol_table::remove_unreachable_nodes): Likewise.
29901 (ipa_single_use): Likewise.
29902 * passes.c (execute_todo): Likewise.
29903 * predict.c (drop_profile): Likewise.
29904 * symtab.c (symtab_node::get_dump_name): New function.
29905 (symtab_node::dump_name): Likewise.
29906 (symtab_node::dump_asm_name): Likewise.
29907 (symtab_node::dump_references): Likewise.
29908 (symtab_node::dump_referring): Likewise.
29909 (symtab_node::dump_base): Likewise.
29910 (symtab_node::debug_symtab): Likewise.
29911 * tree-sra.c (convert_callers_for_node): Likewise.
29912 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29913 * value-prof.c (init_node_map): Likewise.
29915 2017-05-23 Martin Liska <mliska@suse.cz>
29917 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29918 and symtab_node::debug_symtab to symbol_table::debug.
29919 * cgraphunit.c (analyze_functions): Use the renamed function.
29920 (symbol_table::compile): Likewise.
29921 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29922 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29923 * passes.c (execute_todo): Likewise.
29924 * symtab.c (symbol_table::dump): New function.
29925 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29927 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29929 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29930 that nonconst implies exec.
29932 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29934 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29935 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29936 (inline_edge_summary_vec): Turn into ...
29937 (ipa_call_summaries): ... this one.
29938 (redirect_to_unreachable, edge_set_predicate,
29939 evaluate_properties_for_edge, inline_summary_alloc,
29940 reset_ipa_call_summary, reset_inline_summary,
29941 inline_summary_t::duplicate): Update.
29942 (inline_edge_duplication_hook): Turn to ...
29943 (ipa_call_summary_t::duplicate): ... this one.
29944 (inline_edge_removal_hook): Turn to ...
29945 (ipa_call_summary_t::remove): ... this one.
29946 (dump_inline_edge_summary): Turn to ...
29947 (dump_ipa_call_summary): ... this one.
29948 (estimate_function_body_sizes): Update.
29949 (inline_update_callee_summaries): Update.
29950 (remap_edge_change_prob): Update.
29951 (remap_edge_summaries): Update.
29952 (inline_merge_summary): Update.
29953 (do_estimate_edge_time): Update.
29954 (inline_generate_summary): Update.
29955 (inline_read_section): Update.
29956 (inline_read_summary): Update.
29957 (inline_free_summary): Update.
29958 * ipa-inline.c (can_inline_edge_p): Update.
29959 (compute_inlined_call_time): Update.
29960 (want_inline_small_function_p): Update.
29961 (edge_badness): Update.
29962 (early_inliner): Update.
29963 * ipa-inline.h (inline_edge_summary): Turn to ...
29964 (ipa_call_summary): ... this one.
29965 (ipa_call_summary_t): New class.
29966 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29967 (ipa_call_summaries): New.
29968 (inline_edge_summary): Remove.
29969 (estimate_edge_growth): Update.
29970 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29971 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29972 * ipa-split.c (execute_split_functions): Update.
29973 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29975 2017-05-23 Tom de Vries <tom@codesourcery.com>
29977 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29978 attributes): Document rdrand effective target.
29980 2017-05-23 Tom de Vries <tom@codesourcery.com>
29982 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29983 attributes): Sort alphabetically.
29985 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29987 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29989 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29992 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29993 V2DF/V2DI splat into two separate patterns, one that handles
29994 registers, and the other that only handles memory. Drop support
29995 for splatting from a GPR on ISA 2.07 and then splitting the
29996 splat into direct move and splat.
29997 (vsx_splat_<mode>_reg): Likewise.
29998 (vsx_splat_<mode>_mem): Likewise.
30000 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30002 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30004 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30006 PR middle-end/80809
30007 * omp-low.c (finish_taskreg_remap): New function.
30008 (finish_taskreg_scan): If unit size of ctx->record_type
30009 is non-constant, unshare the size expression and replace
30010 decls in it with possible outer var refs.
30012 PR middle-end/80809
30013 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30014 GOVD_SHARED rather than GOVD_PRIVATE with it.
30015 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30016 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30018 PR middle-end/80853
30019 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30020 as last argument to build_outer_var_ref for pointer bases of array
30021 section reductions.
30023 2017-05-19 Martin Sebor <msebor@redhat.com>
30025 * print-tree.c (print_node): Print DECL_READ_P flag.
30027 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30029 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30030 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30031 * cgraph.c: Likewise.
30032 * cgraphunit.c: Likewise.
30033 * gengtype.c: Likewise.
30034 * ipa-cp.c: Likewise.
30035 * ipa-devirt.c: Likewise.
30036 * ipa-icf.c: Likewise.
30037 * ipa-predicate.c: Likewise.
30038 * ipa-profile.c: Likewise.
30039 * ipa-prop.c: Likewise.
30040 * ipa-split.c: Likewise.
30042 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30043 edge_predicate_pool, dump_inline_hints,
30044 inline_summary::account_size_time, redirect_to_unreachable,
30045 edge_set_predicate, set_hint_predicate,
30046 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30047 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30048 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30049 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30050 ipa_call_summary_t::remove, initialize_growth_caches,
30051 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30052 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30053 mark_modified, unmodified_parm_1, unmodified_parm,
30054 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30055 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30056 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30057 will_be_nonconstant_predicate, record_modified_bb_info,
30058 get_minimal_bb, record_modified, param_change_prob,
30059 phi_result_unknown_predicate, predicate_for_phi_result,
30060 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30061 estimate_function_body_sizes, compute_inline_parameters,
30062 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30063 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30064 inline_update_callee_summaries, remap_edge_change_prob,
30065 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30066 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30067 inline_analyze_function, inline_summary_t::insert,
30068 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30069 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30070 inline_free_summary): Move to ipa-fnsummary.h
30071 (predicate_t): Remove.
30072 * ipa-fnsummary.c: New file.
30073 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30074 (enum inline_hints_vals, inline_hints, agg_position_info,
30075 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30076 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30077 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30078 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30079 inline_read_summary, inline_write_summary, inline_free_summary,
30080 inline_analyze_function, initialize_inline_failed,
30081 inline_merge_summary, inline_update_overall_summary,
30082 compute_inline_parameters): Move to ipa-fnsummary.h
30083 * ipa-fnsummary.h: New file.
30084 * ipa-inline-transform.h: Include ipa-inline.h.
30085 * ipa-inline.c: LIkewise.
30087 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30089 * ipa-inline.c (edge_badness): Use inlined_time instead of
30090 inline_summaries->get.
30092 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30094 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30096 2017-05-22 Nathan Sidwell <nathan@acm.org>
30098 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30099 (fdump-lang): Document 'raw' option.
30100 * dumpfile.h (TDI_tu): Delete.
30101 * dumpfile.c (dump_files): Remove translation-unit.
30102 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30104 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30106 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30107 command option from $(AWK) call.
30108 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30109 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30110 [FORMAT]: Remove handling of variable.
30111 * config/avr/t-multilib: Regenerate.
30113 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30115 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30117 (dump_inline_summary): Do not print self_time.
30118 (estimate_function_body_sizes): Do not set self_time.
30119 (compute_inline_parameters): Likewise.
30120 (inline_read_section, inline_write_summary): Do not stream self_time.
30121 * ipa-inline.h (inline_summary): Drop self_time.
30123 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30125 * ipa-inline-analysis.c (account_size_time): Rename to ...
30126 (inline_summary::account_size_time): ... this one.
30127 (reset_ipa_call_summary): Turn to ...
30128 (ipa_call_summary::reset): ... this one.
30129 (reset_inline_summary): Turn to ...
30130 (inline_summary::reset): ... this one.
30131 (inline_summary_t::remove): Update.
30132 (inline_summary_t::duplicate): Update.
30133 (ipa_call_summary_t::remove): Update.
30134 (dump_inline_summary): Update.
30135 (estimate_function_body_sizes): Update.
30136 (compute_inline_parameters): Update.
30137 (estimate_node_size_and_time): Update.
30138 (inline_merge_summary): Update.
30139 (inline_update_overall_summary): Update.
30140 (inline_read_section): Update.
30141 (inline_write_summary): Update.
30142 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30143 add account_size_time and reset member functions.
30144 (ipa_call_summary): Add reset function.
30145 * ipa-predicate.h (predicate::operator &): Constify.
30147 2017-05-22 Richard Biener <rguenther@suse.de>
30149 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30151 2017-05-19 Jason Merrill <jason@redhat.com>
30153 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30155 2017-05-19 Marek Polacek <polacek@redhat.com>
30158 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30159 TYPE_OVERFLOW_WRAPS checks.
30161 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30163 * tree-core.h (enum omp_clause_default_kind): Add
30164 "OMP_CLAUSE_DEFAULT_PRESENT".
30165 * tree-pretty-print.c (dump_omp_clause): Handle it.
30166 * gimplify.c (enum gimplify_omp_var_data): Add
30167 "GOVD_MAP_FORCE_PRESENT".
30168 (gimplify_adjust_omp_clauses_1): Map it to
30169 "GOMP_MAP_FORCE_PRESENT".
30170 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30172 * gimplify.c (oacc_default_clause): Clarify.
30174 2017-05-19 Nathan Sidwell <nathan@acm.org>
30176 LANG_HOOK_REGISTER_DUMPS
30177 * toplev.c (general_init): Call register dump lang hook.
30178 * doc/invoke.texi: Document -fdump-lang option family.
30179 * dumpfile.c (dump_files): Remove class dump here.
30180 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30181 * dumpfile.h (tree_dump_index): Remove TDI_class.
30182 * langhooks-def.h (lhd_register_dumps): Declare.
30183 (LANG_HOOKS_REGISTER_DUMPS): Define.
30184 (LANG_HOOKS_INITIALIZER): Add it.
30185 * langhooks.c (lhd_register_dumps): Define.
30186 * langhooks.h (struct lang_hooks): Add register_dumps.
30188 2017-05-19 Nathan Sidwell <nathan@acm.org>
30190 * context.h (context::set_passes): New.
30191 * context.c (context::context): Do not create pass manager.
30192 * toplev.c (general_init): Create pass manager here.
30194 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30196 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30197 use this splitter if two add or or instructions would also work for
30198 the constant we want to generate.
30200 2017-05-19 Richard Biener <rguenther@suse.de>
30203 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30204 predicate evaluation.
30206 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30208 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30210 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30213 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30215 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30216 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30217 (inline_edge_summary_vec): Turn into ...
30218 (ipa_call_summaries): ... this one.
30219 (redirect_to_unreachable, edge_set_predicate,
30220 evaluate_properties_for_edge, inline_summary_alloc,
30221 reset_ipa_call_summary, reset_inline_summary,
30222 inline_summary_t::duplicate): Update.
30223 (inline_edge_duplication_hook): Turn to ...
30224 (ipa_call_summary_t::duplicate): ... this one.
30225 (inline_edge_removal_hook): Turn to ...
30226 (ipa_call_summary_t::remove): ... this one.
30227 (dump_inline_edge_summary): Turn to ...
30228 (dump_ipa_call_summary): ... this one.
30229 (estimate_function_body_sizes): Update.
30230 (inline_update_callee_summaries): Update.
30231 (remap_edge_change_prob): Update.
30232 (remap_edge_summaries): Update.
30233 (inline_merge_summary): Update.
30234 (do_estimate_edge_time): Update.
30235 (inline_generate_summary): Update.
30236 (inline_read_section): Update.
30237 (inline_read_summary): Update.
30238 (inline_free_summary): Update.
30239 * ipa-inline.c (can_inline_edge_p): Update.
30240 (compute_inlined_call_time): Update.
30241 (want_inline_small_function_p): Update.
30242 (edge_badness): Update.
30243 (early_inliner): Update.
30244 * ipa-inline.h (inline_edge_summary): Turn to ...
30245 (ipa_call_summary): ... this one.
30246 (ipa_call_summary_t): New class.
30247 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30248 (ipa_call_summaries): New.
30249 (inline_edge_summary): Remove.
30250 (estimate_edge_growth): Update.
30251 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30252 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30253 * ipa-split.c (execute_split_functions): Update.
30254 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30256 2017-05-19 Richard Biener <rguenther@suse.de>
30258 PR middle-end/80764
30259 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30261 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30263 * config/rs6000/rs6000.c (struct machine_function): Add field
30264 fpr_is_wrapped_separately.
30265 (rs6000_get_separate_components): Use 64 components. Handle the
30266 new FPR components.
30267 (rs6000_components_for_bb): Handle the FPR components.
30268 (rs6000_emit_prologue_components): Handle the FPR components.
30269 (rs6000_emit_epilogue_components): Handle the FPR components.
30270 (rs6000_set_handled_components): Handle the FPR components.
30271 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30272 that are already separately shrink-wrapped.
30273 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30274 that are already separately shrink-wrapped.
30276 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30279 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30282 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30283 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30284 cases where the register allocator uses a move and an offsettable
30285 memory operation to/from a FPR register on ISA 2.06/2.07.
30286 (define_peephole2 for Altivec d-form store): Likewise.
30288 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30291 * config/i386/mmx.md (*mov<mode>_internal): Enable
30292 alternatives 11, 12, 13 and 14 also for 32bit targets.
30293 Remove alternatives 15, 16, 17 and 18.
30294 * config/i386/sse.md (vec_concatv2di): Change
30295 alternative (!x, *y) to (x, ?!*Yn).
30297 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30299 * dumpfile.h (enum dump_kind): Remove stray comma.
30301 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30303 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30304 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30305 predicate::num_conditions
30306 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30307 (CHANGED): turn into predicate::changed.
30308 (agg_position_info): Move to ipa-predicate.h
30309 (add_condition, predicate::add_clause, predicate::operator &=,
30310 predicate::or_with, predicate::evaluate, predicate::probability,
30311 dump_condition, dump_clause, predicate::dump,
30312 predicate::remap_after_duplication, predicate::remap_after_inlining,
30313 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30314 (evaluate_conditions_for_known_args): Update.
30315 (set_cond_stmt_execution_predicate): Update.
30316 * ipa-inline.h: Include ipa-predicate.h
30317 (condition, inline_param_summary, conditions, agg_position_info,
30318 predicate): Move to ipa-predicate.h
30319 * ipa-predicate.c: New file.
30320 * ipa-predicate.h: New file.
30322 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30324 * final.c (leaf_function_p): Check we are not in a sequence.
30326 2017-05-18 Martin Liska <mliska@suse.cz>
30328 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30329 * dumpfile.c (dump_register): Use new enum dump_kind.
30330 (get_dump_file_name): Likewise.
30331 (dump_enable_all): Likewise.
30332 (dump_switch_p_1): Likewise.
30333 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30334 * dumpfile.h (enum dump_kind): New enum type.
30335 (struct dump_file_info): Create constructor and
30336 format fields and comments.
30337 * passes.c (pass_manager::register_one_dump_file):
30339 * statistics.c (statistics_early_init): Likewise.
30340 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30341 TDF_TREE with TDF_SLIM.
30342 (gather_memory_references_ref): Likewise.
30344 2017-05-18 Martin Liska <mliska@suse.cz>
30346 * vec.h (struct vnull): Use it.
30348 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30350 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30351 (true_predicate, false_predicate, true_predicate_p,
30352 false_predicate_p): Remove.
30353 (single_cond_predicate, not_inlined_predicate): Turn to member function
30355 (add_condition): Update.
30356 (add_clause): Turn to...
30357 (predicate::add_clause): ... this one; update; allow passing NULL
30359 (and_predicates): Turn to ...
30360 (predicate::operator &=): ... this one.
30361 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30362 (or_predicates): Turn to ...
30363 (predicate::or_with): ... this one.
30364 (evaluate_predicate): Turn to ...
30365 (predicate::evaluate): ... this one.
30366 (predicate_probability): Turn to ...
30367 (predicate::probability): ... this one.
30368 (dump_condition): Update.
30369 (dump_predicate): Turn to ...
30370 (predicate::dump): ... this one.
30371 (account_size_time): Update.
30372 (edge_set_predicate): Update.
30373 (set_hint_predicate): UPdate.
30374 (evaluate_conditions_for_known_args): Update.
30375 (evaluate_properties_for_edge): Update.
30376 (remap_predicate_after_duplication): Turn to...
30377 (predicate::remap_after_duplication): ... this one.
30378 (remap_hint_predicate_after_duplication): Update.
30379 (inline_summary_t::duplicate): UPdate.
30380 (dump_inline_edge_summary): Update.
30381 (dump_inline_summary): Update.
30382 (set_cond_stmt_execution_predicate): Update.
30383 (set_switch_stmt_execution_predicate): Update.
30384 (compute_bb_predicates): Update.
30385 (will_be_nonconstant_expr_predicate): Update.
30386 (will_be_nonconstant_predicate): Update.
30387 (phi_result_unknown_predicate): Update.
30388 (predicate_for_phi_result): Update.
30389 (array_index_predicate): Update.
30390 (estimate_function_body_sizes): Update.
30391 (estimate_node_size_and_time): Update.
30392 (estimate_ipcp_clone_size_and_time): Update.
30393 (remap_predicate): Rename to ...
30394 (predicate::remap_after_inlining): ... this one.
30395 (remap_hint_predicate): Update.
30396 (inline_merge_summary): Update.
30397 (inline_update_overall_summary): Update.
30398 (estimate_size_after_inlining): Update.
30399 (read_predicate): Rename to ...
30400 (predicate::stream_in): ... this one.
30401 (read_inline_edge_summary): Update.
30402 (write_predicate): Rename to ...
30403 (predicate::stream_out): ... this one.
30404 (write_inline_edge_summary): Update.
30405 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30406 (clause_t): Turn to uint32_t
30407 (predicate): Turn to class; implement constructor and operators
30409 (size_time_entry): Update.
30410 (inline_summary): Update.
30411 (inline_edge_summary): Update.
30413 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30415 * fold-const.c (fold_binary_loc): Move transformation...
30416 * match.pd (C - X CMP X): ... here.
30418 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30420 * config/sparc/sparc.c (sparc_option_override): Set function
30421 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30422 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30424 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30426 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30428 2017-05-18 Marek Polacek <polacek@redhat.com>
30431 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30432 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30433 gimple_assign_load_p.
30435 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30437 PR middle-end/80692
30438 * real.c (do_compare): Give decimal_do_compare preference over
30439 comparing just the signs.
30441 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30443 * doc/md.texi (Canonicalization of Instructions): Describe the
30444 canonical form of instructions that inherently set a condition
30447 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30449 PR middle-end/80775
30450 * tree-cfg.c: Move deletion of unreachable case statements to after
30451 the merging of consecutive case labels.
30453 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30455 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30456 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30457 restoring of callee-saved registers.
30459 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30461 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30462 * config/visium/visium.c (single_set_and_flags): Likewise.
30463 * config/visium/visium.md (Substitutions): Likewise.
30465 2017-05-17 Martin Liska <mliska@suse.cz>
30467 * cfg.c: Introduce dump_flags_t type and
30468 use it instead of int type.
30470 * cfghooks.c: Likewise.
30471 * cfghooks.h (struct cfg_hooks): Likewise.
30472 * cfgrtl.c: Likewise.
30473 * cfgrtl.h: Likewise.
30474 * cgraph.c (cgraph_node::get_body): Likewise.
30475 * coretypes.h: Likewise.
30476 * domwalk.c: Likewise.
30477 * domwalk.h: Likewise.
30478 * dumpfile.c (struct dump_option_value_info): Likewise.
30479 (dump_enable_all): Likewise.
30480 (dump_switch_p_1): Likewise.
30481 (opt_info_switch_p): Likewise.
30482 * dumpfile.h (enum tree_dump_index): Likewise.
30483 (struct dump_file_info): Likewise.
30484 * genemit.c: Likewise.
30485 * generic-match-head.c: Likewise.
30486 * gengtype.c (open_base_files): Likewise.
30487 * gimple-pretty-print.c: Likewise.
30488 * gimple-pretty-print.h: Likewise.
30489 * graph.c (print_graph_cfg): Likewise.
30490 * graphite-scop-detection.c (dot_all_sese): Likewise.
30491 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30492 * loop-unroll.c (report_unroll): Likewise.
30493 * passes.c (pass_manager::register_one_dump_file): Likewise.
30494 * print-tree.c: Likewise.
30495 * statistics.c: Likewise.
30496 * tree-cfg.c: Likewise.
30497 * tree-cfg.h: Likewise.
30498 * tree-dfa.c: Likewise.
30499 * tree-dfa.h: Likewise.
30500 * tree-dump.c (dump_function): Likewise.
30501 * tree-dump.h (struct dump_info): Likewise.
30502 * tree-pretty-print.c: Likewise.
30503 * tree-pretty-print.h: Likewise.
30504 * tree-ssa-live.c: Likewise.
30505 * tree-ssa-live.h: Likewise.
30506 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30507 * tree-vect-loop.c: Likewise.
30508 * tree-vect-slp.c: Likewise.
30510 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30511 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30513 PR tree-optimization/80457
30514 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30515 of all arguments to a statement as scalar_to_vec operations.
30516 (vectorizable_call): Adjust call to vect_model_simple_cost for
30518 (vectorizable_conversion): Likewise.
30519 (vectorizable_assignment): Likewise.
30520 (vectorizable_shift): Likewise.
30521 (vectorizable_operation): Likewise.
30522 (vectorizable_comparison): Likewise.
30523 (vect_is_simple_cond): Record the def types for operands.
30524 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30525 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30526 for statement argument count.
30528 2017-05-16 Carl Love <cel@us.ibm.com>
30530 * config/rs6000/rs6000-c: Add support for built-in functions
30531 vector unsigned long long vec_bperm (vector unsigned long long,
30532 vector unsigned char)
30533 vector signed long long vec_mule (vector signed int,
30535 vector unsigned long long vec_mule (vector unsigned int,
30536 vector unsigned int)
30537 vector signed long long vec_mulo (vector signed int,
30539 vector unsigned long long vec_mulo (vector unsigned int,
30540 vector unsigned int)
30541 vector signed char vec_sldw (vector signed char,
30542 vector signed char,
30544 vector unsigned char vec_sldw (vector unsigned char,
30545 vector unsigned char,
30547 vector signed short vec_sldw (vector signed short,
30548 vector signed short,
30550 vector unsigned short vec_sldw (vector unsigned short,
30551 vector unsigned short,
30553 vector signed int vec_sldw (vector signed int,
30556 vector unsigned int vec_sldw (vector unsigned int,
30557 vector unsigned int,
30559 vector signed long long vec_sldw (vector signed long long,
30560 vector signed long long,
30562 vector unsigned long long vec_sldw (vector unsigned long long,
30563 vector unsigned long long,
30565 * config/rs6000/rs6000-c: Add support for built-in functions
30566 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30567 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30568 * doc/extend.texi: Update the built-in documentation for the
30569 new built-in functions.
30571 2017-05-16 Marek Polacek <polacek@redhat.com>
30575 * tree.c (save_expr): Don't fold the expression.
30577 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30579 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30580 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30581 and (?*y,m). Update insn attributes.
30583 2017-05-16 Martin Liska <mliska@suse.cz>
30585 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30586 flags argument of print_gimple_stmt, print_gimple_expr,
30587 print_generic_stmt and print_generic_expr.
30588 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30589 * coretypes.h: Likewise.
30590 * except.c (dump_eh_tree): Likewise.
30591 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30592 * gimple-pretty-print.h: Likewise.
30593 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30594 (backprop::push_to_worklist): Likewise.
30595 (backprop::pop_from_worklist): Likewise.
30596 (backprop::process_use): Likewise.
30597 (backprop::intersect_uses): Likewise.
30598 (note_replacement): Likewise.
30599 * gimple-ssa-store-merging.c
30600 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30601 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30602 (pass_store_merging::execute): Likewise.
30603 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30604 (ssa_base_cand_dump_callback): Likewise.
30605 (dump_incr_vec): Likewise.
30606 (replace_refs): Likewise.
30607 (replace_mult_candidate): Likewise.
30608 (create_add_on_incoming_edge): Likewise.
30609 (create_phi_basis): Likewise.
30610 (insert_initializers): Likewise.
30611 (all_phi_incrs_profitable): Likewise.
30612 (introduce_cast_before_cand): Likewise.
30613 (replace_one_candidate): Likewise.
30614 * gimplify.c (gimplify_expr): Likewise.
30615 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30616 (set_rename): Likewise.
30617 (rename_uses): Likewise.
30618 (copy_loop_phi_nodes): Likewise.
30619 (add_close_phis_to_merge_points): Likewise.
30620 (copy_loop_close_phi_args): Likewise.
30621 (copy_cond_phi_args): Likewise.
30622 (graphite_copy_stmts_from_block): Likewise.
30623 (translate_pending_phi_nodes): Likewise.
30624 * graphite-poly.c (print_pdr): Likewise.
30625 (dump_gbb_cases): Likewise.
30626 (dump_gbb_conditions): Likewise.
30627 (print_scop_params): Likewise.
30628 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30629 (build_cross_bb_scalars_use): Likewise.
30630 (gather_bbs::before_dom_children): Likewise.
30631 * hsa-dump.c (dump_hsa_immed): Likewise.
30632 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30633 (get_replacement_map): Likewise.
30634 * ipa-inline-analysis.c (dump_condition): Likewise.
30635 (estimate_function_body_sizes): Likewise.
30636 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30637 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30638 * ipa-prop.c (ipa_dump_param): Likewise.
30639 (ipa_print_node_jump_functions_for_edge): Likewise.
30640 (ipa_modify_call_arguments): Likewise.
30641 (ipa_modify_expr): Likewise.
30642 (ipa_dump_param_adjustments): Likewise.
30643 (ipa_dump_agg_replacement_values): Likewise.
30644 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30645 * ipa-pure-const.c (check_stmt): Likewise.
30646 (pass_nothrow::execute): Likewise.
30647 * ipa-split.c (execute_split_functions): Likewise.
30648 * omp-offload.c (dump_oacc_loop_part): Likewise.
30649 (dump_oacc_loop): Likewise.
30650 * trans-mem.c (tm_log_emit): Likewise.
30651 (tm_memopt_accumulate_memops): Likewise.
30652 (dump_tm_memopt_set): Likewise.
30653 (dump_tm_memopt_transform): Likewise.
30654 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30655 (print_loop): Likewise.
30656 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30657 (chkp_gather_checks_info): Likewise.
30658 (chkp_get_check_result): Likewise.
30659 (chkp_remove_check_if_pass): Likewise.
30660 (chkp_use_outer_bounds_if_possible): Likewise.
30661 (chkp_reduce_bounds_lifetime): Likewise.
30662 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30663 (chkp_mark_completed_bounds): Likewise.
30664 (chkp_register_incomplete_bounds): Likewise.
30665 (chkp_mark_invalid_bounds): Likewise.
30666 (chkp_maybe_copy_and_register_bounds): Likewise.
30667 (chkp_build_returned_bound): Likewise.
30668 (chkp_get_bound_for_parm): Likewise.
30669 (chkp_build_bndldx): Likewise.
30670 (chkp_get_bounds_by_definition): Likewise.
30671 (chkp_generate_extern_var_bounds): Likewise.
30672 (chkp_get_bounds_for_decl_addr): Likewise.
30673 * tree-chrec.c (chrec_apply): Likewise.
30674 * tree-data-ref.c (dump_data_reference): Likewise.
30675 (dump_subscript): Likewise.
30676 (dump_data_dependence_relation): Likewise.
30677 (analyze_overlapping_iterations): Likewise.
30678 * tree-inline.c (expand_call_inline): Likewise.
30679 (tree_function_versioning): Likewise.
30680 * tree-into-ssa.c (dump_defs_stack): Likewise.
30681 (dump_currdefs): Likewise.
30682 (dump_names_replaced_by): Likewise.
30683 (dump_update_ssa): Likewise.
30684 (update_ssa): Likewise.
30685 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30686 * tree-parloops.c (build_new_reduction): Likewise.
30687 (try_create_reduction_list): Likewise.
30688 (ref_conflicts_with_region): Likewise.
30689 (oacc_entry_exit_ok_1): Likewise.
30690 (oacc_entry_exit_single_gang): Likewise.
30691 * tree-pretty-print.h: Likewise.
30692 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30693 (get_scalar_evolution): Likewise.
30694 (add_to_evolution): Likewise.
30695 (get_loop_exit_condition): Likewise.
30696 (analyze_evolution_in_loop): Likewise.
30697 (analyze_initial_condition): Likewise.
30698 (analyze_scalar_evolution): Likewise.
30699 (instantiate_scev): Likewise.
30700 (number_of_latch_executions): Likewise.
30701 (gather_chrec_stats): Likewise.
30702 (final_value_replacement_loop): Likewise.
30703 (scev_const_prop): Likewise.
30704 * tree-sra.c (dump_access): Likewise.
30705 (disqualify_candidate): Likewise.
30706 (create_access): Likewise.
30707 (reject): Likewise.
30708 (maybe_add_sra_candidate): Likewise.
30709 (create_access_replacement): Likewise.
30710 (analyze_access_subtree): Likewise.
30711 (analyze_all_variable_accesses): Likewise.
30712 (sra_modify_assign): Likewise.
30713 (initialize_constant_pool_replacements): Likewise.
30714 (find_param_candidates): Likewise.
30715 (decide_one_param_reduction): Likewise.
30716 (replace_removed_params_ssa_names): Likewise.
30717 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30718 * tree-ssa-copy.c (dump_copy_of): Likewise.
30719 (copy_prop_visit_cond_stmt): Likewise.
30720 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30721 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30722 (record_equivalences_from_stmt): Likewise.
30723 * tree-ssa-dse.c (compute_trims): Likewise.
30724 (delete_dead_call): Likewise.
30725 (delete_dead_assignment): Likewise.
30726 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30727 (forward_propagate_into_cond): Likewise.
30728 (pass_forwprop::execute): Likewise.
30729 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30730 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30732 (move_computations_worker): Likewise.
30733 (execute_sm): Likewise.
30734 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30735 (remove_exits_and_undefined_stmts): Likewise.
30736 (remove_redundant_iv_tests): Likewise.
30737 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30738 (adjust_iv_update_pos): Likewise.
30739 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30740 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30741 (value_replacement): Likewise.
30742 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30743 * tree-ssa-pre.c (print_pre_expr): Likewise.
30744 (get_representative_for): Likewise.
30745 (create_expression_by_pieces): Likewise.
30746 (insert_into_preds_of_block): Likewise.
30747 (eliminate_insert): Likewise.
30748 (eliminate_dom_walker::before_dom_children): Likewise.
30749 (eliminate): Likewise.
30750 (remove_dead_inserted_code): Likewise.
30751 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30752 * tree-ssa-reassoc.c (get_rank): Likewise.
30753 (eliminate_duplicate_pair): Likewise.
30754 (eliminate_plus_minus_pair): Likewise.
30755 (eliminate_not_pairs): Likewise.
30756 (undistribute_ops_list): Likewise.
30757 (eliminate_redundant_comparison): Likewise.
30758 (update_range_test): Likewise.
30759 (optimize_range_tests_var_bound): Likewise.
30760 (optimize_vec_cond_expr): Likewise.
30761 (rewrite_expr_tree): Likewise.
30762 (rewrite_expr_tree_parallel): Likewise.
30763 (linearize_expr): Likewise.
30764 (break_up_subtract): Likewise.
30765 (linearize_expr_tree): Likewise.
30766 (attempt_builtin_powi): Likewise.
30767 (attempt_builtin_copysign): Likewise.
30768 (transform_stmt_to_copy): Likewise.
30769 (transform_stmt_to_multiply): Likewise.
30770 (dump_ops_vector): Likewise.
30771 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30772 (print_scc): Likewise.
30773 (set_ssa_val_to): Likewise.
30774 (visit_reference_op_store): Likewise.
30775 (visit_use): Likewise.
30776 (sccvn_dom_walker::before_dom_children): Likewise.
30777 (run_scc_vn): Likewise.
30778 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30780 (expr_hash_elt::print): Likewise.
30781 (const_and_copies::pop_to_marker): Likewise.
30782 (const_and_copies::record_const_or_copy_raw): Likewise.
30783 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30784 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30785 (dump_predicates): Likewise.
30786 (find_uninit_use): Likewise.
30787 (warn_uninitialized_phi): Likewise.
30788 (pass_late_warn_uninitialized::execute): Likewise.
30789 * tree-ssa.c (verify_vssa): Likewise.
30790 (verify_ssa): Likewise.
30791 (maybe_optimize_var): Likewise.
30792 * tree-vrp.c (dump_value_range): Likewise.
30793 (dump_all_value_ranges): Likewise.
30794 (dump_asserts_for): Likewise.
30795 (register_edge_assert_for_2): Likewise.
30796 (vrp_visit_cond_stmt): Likewise.
30797 (vrp_visit_switch_stmt): Likewise.
30798 (vrp_visit_stmt): Likewise.
30799 (vrp_visit_phi_node): Likewise.
30800 (simplify_cond_using_ranges_1): Likewise.
30801 (fold_predicate_in): Likewise.
30802 (evrp_dom_walker::before_dom_children): Likewise.
30803 (evrp_dom_walker::push_value_range): Likewise.
30804 (evrp_dom_walker::pop_value_range): Likewise.
30805 (execute_early_vrp): Likewise.
30807 2017-05-16 Richard Biener <rguenther@suse.de>
30809 * dwarf2out.c (loc_list_from_tree_1): Do not create
30810 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30812 2017-05-16 Richard Biener <rguenther@suse.de>
30814 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30816 (note_variable_value_in_expr): If we resolved the decl ref
30817 do not push to the stack.
30819 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30821 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30822 operations in fast-math mode.
30823 (vaddq_f16): Likewise.
30824 (vmul_f16): Likewise.
30825 (vmulq_f16): Likewise.
30826 (vsub_f16): Likewise.
30827 (vsubq_f16): Likewise.
30828 * config/arm/neon.md (add<mode>3): New.
30830 (fma:<VH:mode>3): New. Also remove outdated comment.
30833 2017-05-16 Martin Liska <mliska@suse.cz>
30837 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30838 (odr_types_equivalent_p): Likewise.
30840 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30842 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30844 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30847 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30848 non-interunit SSE move alternatives with '?'.
30849 (zero-extendsidi peephole2): New peephole to skip intermediate
30850 general register in SSE zero-extend sequence.
30852 2017-05-15 Jeff Law <law@redhat.com>
30854 * reorg.c (relax_delay_slots): Create a new variable to hold
30855 the temporary target rather than clobbering TARGET_LABEL.
30857 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30858 missing argument to extract_bit_field call.
30859 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30861 2017-05-15 Martin Liska <mliska@suse.cz>
30864 * gcc.c (process_command): Do not allow empty argument of -o option.
30866 2017-05-15 Renlin Li <renlin.li@arm.com>
30868 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30869 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30870 * config/aarch64/constraints.md (Usf): Add long call check.
30871 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30872 (call_value): Likewise.
30873 (sibcall): Likewise.
30874 (sibcall_value): Likewise.
30876 (call_value_insn): New.
30877 (sibcall_insn): Update rtx pattern.
30878 (sibcall_value_insn): Likewise.
30879 (call_internal): Remove.
30880 (call_value_internal): Likewise.
30881 (sibcall_internal): Likewise.
30882 (sibcall_value_internal): Likewise.
30883 (call_reg): Likewise.
30884 (call_symbol): Likewise.
30885 (call_value_reg): Likewise.
30886 (call_value_symbol): Likewise.
30888 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30891 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30893 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30895 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30896 compatible with CCGOCmode and with CCZmode.
30898 2017-05-14 Martin Sebor <msebor@redhat.com>
30900 PR middle-end/77671
30901 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30902 (gimple_fold_builtin_snprintf): Same.
30903 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30904 (gimple_fold_builtin_snprintf): Same.
30905 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30906 of character types.
30907 (is_call_safe): New function.
30908 (try_substitute_return_value): Call it.
30909 (try_simplify_call): New function.
30910 (pass_sprintf_length::handle_gimple_call): Call it.
30912 2017-05-14 Martin Sebor <msebor@redhat.com>
30914 PR middle-end/80669
30915 * builtins.c (expand_builtin_stpncpy): Simplify.
30917 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30919 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30920 * config/i386/i386.h
30921 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30922 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30923 (struct machine_function): Add new members call_ms2sysv,
30924 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30925 (struct machine_frame_state): New fields sp_realigned and
30926 sp_realigned_offset.
30927 * config/i386/i386.c
30928 (enum xlogue_stub): New enum.
30929 (enum xlogue_stub_sets): New enum.
30930 (class xlogue_layout): New class.
30931 (struct ix86_frame): New fields stack_realign_allocate_offset,
30932 stack_realign_offset and outlined_save_offset. Modify comments to
30933 detail stack layout when using out-of-line stubs.
30934 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30935 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30936 -mcall-ms2sysv-xlogues.
30937 (stub_managed_regs): New static variable.
30938 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30939 registers managed by out-of-line stub.
30940 (disable_call_ms2sysv_xlogues): New function.
30941 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30942 m->call_ms2sysv when appropriate and compute frame layout for
30944 (sp_valid_at, fp_valid_at): New inline functions.
30945 (choose_basereg): New function.
30946 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30948 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30949 Use align parameter of choose_baseaddr to generated aligned SSE movs
30951 (pro_epilogue_adjust_stack): Modify to track
30952 machine_frame_state::sp_realigned.
30953 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30954 (ix86_nsaved_sseregs): Likewise.
30955 (ix86_emit_save_regs): Likewise.
30956 (ix86_emit_save_regs_using_mov): Likewise.
30957 (ix86_emit_save_sse_regs_using_mov): Likewise.
30958 (get_scratch_register_on_entry): Likewise.
30959 (gen_frame_set): New function.
30960 (gen_frame_load): Likewise.
30961 (gen_frame_store): Likewise.
30962 (emit_outlined_ms2sysv_save): Likewise.
30963 (emit_outlined_ms2sysv_restore): Likewise.
30964 (ix86_expand_prologue): Modify stack re-alignment code and call
30965 emit_outlined_ms2sysv_save when appropriate.
30966 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30967 parameter rtx_insn *insn, which allows the function to be used to only
30968 generate the notes.
30969 (ix86_expand_epilogue): Modify validity checks of frame and stack
30970 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30971 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30972 * config/i386/predicates.md
30973 (save_multiple): New predicate.
30974 (restore_multiple): Likewise.
30975 * config/i386/sse.md
30976 (save_multiple<mode>): New pattern.
30977 (save_multiple_realign<mode>): Likewise.
30978 (restore_multiple<mode>): Likewise.
30979 (restore_multiple_and_return<mode>): Likewise.
30980 (restore_multiple_leave_return<mode>): Likewise.
30981 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30983 2017-05-14 Julia Koval <julia.koval@intel.com>
30985 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30986 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30987 (__builtin_ia32_xsetbv): New builtins.
30988 * config/i386/i386.c (ix86_expand_special_args_builtin):
30990 (ix86_expand_builtin): Special expand for new intrinsics.
30991 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30992 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30993 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30995 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30997 * cfganal.c (inverted_post_order_compute): Change argument type
30999 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31000 * df-core.c (rest_of_handle_df_initialize): Adjust.
31001 (rest_of_handle_df_finish): Likewise.
31002 (df_analyze_1): Likewise.
31003 (df_analyze): Likewise.
31004 (loop_inverted_post_order_compute): Change argument to be a vec *.
31005 (df_analyze_loop): Adjust.
31006 (df_get_n_blocks): Likewise.
31007 (df_get_postorder): Likewise.
31008 * df.h (struct df_d): Change field to be a vec.
31009 * lcm.c (compute_laterin): Adjust.
31010 (compute_available): Likewise.
31011 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31012 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31013 * tree-ssa-pre.c (compute_antic): Likewise.
31015 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31017 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31018 (depth_first_search::depth_first_search): Change structure init
31019 function to this constructor.
31020 (depth_first_search::add_bb): Rename function to this member.
31021 (depth_first_search::execute): Likewise.
31022 (flow_dfs_compute_reverse_finish): Adjust.
31024 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31026 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31027 (longest_simple_path): Likewise.
31028 * shrink-wrap.c (spread_components): Likewise.
31029 (disqualify_problematic_components): Likewise.
31030 (emit_common_heads_for_components): Likewise.
31031 (emit_common_tails_for_components): Likewise.
31032 (insert_prologue_epilogue_for_components): Likewise.
31034 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31036 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31039 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31041 * df-core.c (df_set_blocks): Start using auto_bitmap.
31042 (df_compact_blocks): Likewise.
31043 * df-problems.c (df_rd_confluence_n): Likewise.
31044 * df-scan.c (df_insn_rescan_all): Likewise.
31045 (df_process_deferred_rescans): Likewise.
31046 (df_update_entry_block_defs): Likewise.
31047 (df_update_exit_block_uses): Likewise.
31048 (df_entry_block_bitmap_verify): Likewise.
31049 (df_exit_block_bitmap_verify): Likewise.
31050 (df_scan_verify): Likewise.
31051 * lra-constraints.c (lra_constraints): Likewise.
31052 (undo_optional_reloads): Likewise.
31053 (lra_undo_inheritance): Likewise.
31054 * lra-remat.c (calculate_gen_cands): Likewise.
31055 (do_remat): Likewise.
31056 * lra-spills.c (assign_spill_hard_regs): Likewise.
31057 (spill_pseudos): Likewise.
31058 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31059 (bitmap_set_subtract_values): Likewise.
31061 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31063 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31064 management with auto_bitmap.
31065 (fix_inter_tick): Likewise.
31066 (fix_recovery_deps): Likewise.
31067 * ira.c (add_store_equivs): Likewise.
31068 (find_moveable_pseudos): Likewise.
31069 (split_live_ranges_for_shrink_wrap): Likewise.
31070 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31071 (rtx_reuse_manager::seen_def_p): Likewise.
31072 (rtx_reuse_manager::set_seen_def): Likewise.
31073 * print-rtl.h (class rtx_reuse_manager): Likewise.
31075 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31077 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31079 (migrate_btr_def): Likewise.
31080 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31081 * df-core.c (loop_post_order_compute): Likewise.
31082 (loop_inverted_post_order_compute): Likewise.
31083 * hsa-common.h: Likewise.
31084 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31085 * init-regs.c (initialize_uninitialized_regs): Likewise.
31086 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31087 (inline_small_functions): Likewise.
31088 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31089 * ira.c (combine_and_move_insns): Likewise.
31090 (build_insn_chain): Likewise.
31091 * loop-invariant.c (find_invariants): Likewise.
31092 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31093 * predict.c (tree_predict_by_opcode): Likewise.
31094 (predict_paths_leading_to): Likewise.
31095 (predict_paths_leading_to_edge): Likewise.
31096 (estimate_loops_at_level): Likewise.
31097 (estimate_loops): Likewise.
31098 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31099 (spread_components): Likewise.
31100 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31101 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31102 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31103 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31104 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31105 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31106 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31107 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31108 (mark_threaded_blocks): Likewise.
31109 (thread_through_all_blocks): Likewise.
31110 * tree-ssa.c (verify_ssa): Likewise.
31111 (execute_update_addresses_taken): Likewise.
31112 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31114 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31116 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31118 (post_order_compute): Likewise.
31119 (inverted_post_order_compute): Likewise.
31120 (pre_and_rev_post_order_compute_fn): Likewise.
31122 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31124 * genrecog.c (int_set::int_set): Explicitly construct our
31125 auto_vec base class.
31126 * vec.h (auto_vec::auto_vec): New constructor.
31128 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31130 * bitmap.h (class auto_bitmap): New constructor taking
31131 bitmap_obstack * argument.
31133 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31135 * bitmap.h (class auto_bitmap): Change type of m_bits to
31136 bitmap_head, and adjust ctor / dtor and member operators.
31138 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31140 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31141 when returned register mode doesn't match original mode.
31143 2017-05-12 Jeff Law <law@redhat.com>
31144 Jakub Jelinek <jakub@redhat.com>
31146 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31147 we look for cc setter after the compare-elim changes.
31148 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31149 within the vector to match what compare-elim now expects.
31150 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31151 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31153 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31154 after the compare-elim changes.
31155 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31156 the vector to match what compare-elim now expects.
31157 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31158 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31159 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31160 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31161 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31163 * config/visium/visium.c (single_set_and_flags): Fix where
31164 we look for cc setter after the compare-elim changes.
31165 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31166 with the vector to match what compare-elim now expects.
31167 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31168 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31169 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31170 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31171 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31172 (neg<mode>2_insn_set_overflow): Likewise.
31174 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31176 PR middle-end/79794
31177 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31178 maybe_expand_insn call, set ops[0].target. If still set after call,
31179 set alt_rtl. Add extra arg to recursive calls.
31180 (extract_bit_field): Add alt_rtl argument. Pass to
31182 * expmed.h (extract_bit_field): Fix prototype.
31183 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31184 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31185 to extract_bit_field_calls.
31186 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31187 Pass alt_rtl to extract_bit_field calls.
31188 * calls.c (store_unaligned_arguments_into_psuedos)
31189 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31190 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31192 * optabs.h (struct expand_operand): Add target bitfield.
31194 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31196 * compare-elim.c (try_eliminate_compare): Canonicalize
31197 operation with embedded compare to
31198 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31199 (set (reg) (operation)].
31201 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31203 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31206 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31207 cost of adding a carry flag for ADC instruction.
31208 [case MINUS]: Ignore the cost of subtracting a carry flag
31209 for SBB instruction.
31211 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31213 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31215 * config/rs6000/bmiintrin.h: New file.
31216 * config/rs6000/bmi2intrin.h: New file.
31217 * config/rs6000/x86intrin.h: New file.
31219 2017-05-12 Jeff Law <law@redhat.com>
31221 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31224 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31226 PR middle-end/80707
31227 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31229 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31231 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31232 early expansion of vector divide builtins.
31233 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31234 builtins identified as having unsigned arguments.
31236 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31238 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31239 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31240 expansion of vector logical operations (and, andc, or, xor,
31243 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31245 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31246 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31248 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31250 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31251 early GIMPLE expansion of vector multiplies.
31253 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31255 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31256 TARGET_HAVE_MOVT conditional.
31257 (movt splitter): Likewise.
31259 2017-05-12 Richard Biener <rguenther@suse.de>
31261 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31262 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31263 Fold all stmts not inplace.
31265 2017-05-12 Richard Biener <rguenther@suse.de>
31267 PR tree-optimization/80713
31268 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31269 inserted_exprs bit for not removed stmts.
31271 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31273 PR middle-end/69921
31274 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31275 parallelized" attribute for parallelized OpenACC kernels.
31276 * omp-offload.c (execute_oacc_device_lower): Use it.
31278 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31279 Set "oacc kernels" attribute.
31280 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31281 parameter. Adjust all users.
31282 (oacc_fn_attrib_kernels_p): Remove function.
31283 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31284 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31285 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31286 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31287 assert "oacc kernels" attribute is set.
31289 2017-05-11 Carl Love <cel@us.ibm.com>
31291 * config/rs6000/rs6000-c: Add support for built-in functions
31292 vector unsigned char vec_popcnt (vector signed char)
31293 vector unsigned char vec_popcnt (vector unsigned char)
31294 vector unsigned short vec_popcnt (vector signed short)
31295 vector unsigned short vec_popcnt (vector unsigned short)
31296 vector unsigned int vec_popcnt (vector signed int)
31297 vector unsigned int vec_popcnt (vector unsigned int)
31298 vector unsigned long long vec_popcnt (vector signed long long)
31299 vector unsigned long long vec_popcnt (vector unsigned long long)
31300 vector signed long long vec_slo (vector signed long long,
31301 vector signed char)
31302 vector signed long long vec_slo (vector signed long long,
31303 vector unsigned char)
31304 vector unsigned long long vec_slo (vector unsigned long long,
31305 vector signed char)
31306 vector unsigned long long vec_slo (vector unsigned long long,
31307 vector unsigned char)
31308 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31309 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31310 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31311 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31312 * doc/extend.texi: Update the built-in documentation file for the
31313 new built-in functions.
31315 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31317 * attribs.h (sorted_attr_string): Move machine independent
31318 functions for target clone support from the i386 port to common
31319 code. Rename ix86_function_versions to common_function_versions.
31320 Rename make_name to make_unique_name.
31321 (common_function_versions): Likewise.
31322 (make_unique_name): Likewise.
31323 (make_dispatcher_decl): Likewise.
31324 (is_function_default_version): Likewise.
31325 * attribs.c (attr_strcmp): Likewise.
31326 (sorted_attr_string): Likewise.
31327 (common_function_versions): Likewise.
31328 (make_unique_name): Likewise.
31329 (make_dispatcher_decl): Likewise.
31330 (is_function_default_version): Likewise.
31331 * config/i386/i386.c (attr_strcmp): Likewise.
31332 (sorted_attr_string): Likewise.
31333 (ix86_function_versions): Likewise.
31334 (make_name): Likewise.
31335 (make_dispatcher_decl): Likewise.
31336 (is_function_default_version): Likewise.
31337 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31339 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31342 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31343 Account for direct move costs for vec_construct of integer
31346 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31349 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31350 (UNSPEC_STX_ATOMIC): Ditto.
31351 (loaddi_via_sse): New insn.
31352 (storedi_via_sse): Ditto.
31353 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31354 Update corresponding peephole2 patterns.
31355 (atomic_storedi_fpu): Ditto.
31357 2017-05-11 Julia Koval <julia.koval@intel.com>
31359 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31360 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31362 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31363 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31364 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31366 2017-05-11 Nathan Sidwell <nathan@acm.org>
31368 * graphite-poly.c: Include dumpfile.h.
31370 * dumpfle.h (dump_function): Declare here ...
31371 * tree-dump.h (dump_function): ... not here.
31372 * dumpfile.c: #include tree-cfg.h.
31373 (dump_function): Move here from ...
31374 * tree-dump.c (dump_function): ... here.
31375 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31376 * graphite-poly.c: Don't include tree-dump.h.
31377 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31378 * print-tree.c: Likewise.
31379 * stor-layout.c: Likewise.
31380 * tree-nested.c: Likewise.
31382 * dumpfile.c (dump_start): Use TDF_FLAGS.
31383 (dump_enable_all): Fix TDF_KIND check thinko.
31385 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31387 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31388 array entries to represent two legal parameterizations of the
31389 overloaded __builtin_cmpb function, as represented by the
31390 P6_OV_BUILTIN_CMPB constant.
31391 (altivec_resolve_overloaded_builtin): Add special case handling
31392 for the __builtin_cmpb function, as represented by the
31393 P6_OV_BUILTIN_CMPB constant.
31394 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31395 (BU_P6_64BIT_2): New macro.
31396 (BU_P6_OVERLOAD_2): New macro
31397 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31398 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31399 (CMPB): Add overload support to represent both 32-bit and 64-bit
31400 compare-bytes function.
31401 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31402 support for TARGET_CMPB.
31403 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31404 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31405 documentation of the __builtin_cmpb overloaded built-in function.
31407 2017-05-11 Richard Biener <rguenther@suse.de>
31409 PR tree-optimization/80705
31410 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31411 bases are not vectorizable.
31413 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31415 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31416 when counting register pressure.
31418 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31420 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31421 (iv_ca_more_deps): Renamed to ...
31422 (iv_ca_compare_deps): ... this.
31423 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31425 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31427 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31429 (determine_group_iv_costs): ... here.
31430 (find_inv_vars_cb): Record inv var if it's not recorded before.
31432 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31434 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31435 (get_shiftadd_cost): Ditto.
31437 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31439 * tree-ssa-address.c: Include header file.
31440 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31442 (add_to_parts): Refactor.
31443 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31444 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31447 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31449 PR tree-optimization/53090
31450 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31452 (extract_cond_operands): Detect condition with IV on both sides
31453 and return COMP_IV_EXPR_2.
31454 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31455 (rewrite_use_compare): Simplify by removing call to function
31456 extract_cond_operands.
31458 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31460 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31461 (extract_cond_operands): Detect condition comparing against non-
31462 invariant bound and return appropriate enum value.
31463 (find_interesting_uses_cond): Update use of extract_cond_operands.
31464 Handle its return value accordingly.
31465 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31467 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31469 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31470 nonlinear iv_use computation in loop invariant sensitive way.
31472 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31474 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31475 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31477 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31479 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31480 (dump_cand): Support iv_cand.inv_exprs.
31481 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31483 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31486 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31488 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31490 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31491 as local function. Include necessary header files.
31492 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31494 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31496 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31498 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31500 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31501 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31502 RSHIFT_EXPR and BIT_NOT_EXPR.
31504 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31506 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31507 (adjust_setup_cost): New parameter supporting round up adjustment.
31508 (struct address_cost_data): Delete.
31509 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31510 (split_address_cost, ptr_difference_cost): Delete.
31511 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31512 (struct ainc_cost_data): New struct.
31513 (get_address_cost_ainc): New function.
31514 (get_address_cost, get_computation_cost): Reimplement.
31515 (determine_group_iv_cost_address): Record inv_expr for all uses of
31517 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31518 (iv_ca_has_deps): Reimplemented to ...
31519 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31521 (iv_ca_extend): Call iv_ca_more_deps.
31523 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31525 * tree-ssa-address.c (struct mem_address): Move to header file.
31526 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31527 * tree-ssa-address.h (struct mem_address): Move from C file.
31528 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31530 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31532 * tree-affine.h (aff_combination_type): New interface.
31533 (aff_combination_zero_p): Remove static.
31534 (aff_combination_const_p): New interface.
31535 (aff_combination_singleton_var_p): New interfaces.
31537 2017-05-11 Richard Biener <rguenther@suse.de>
31539 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31540 Skip unreachable blocks and destinations.
31541 (eliminate): Move stmt removal and fixup ...
31542 (fini_eliminate): ... here. Skip inserted exprs.
31543 (pass_pre::execute): Move fini_pre after fini_eliminate.
31544 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31545 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31546 PRE to get rid of dead code that has invalid SSA form and
31547 split critical edges again.
31549 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31551 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31553 2017-05-11 Richard Biener <rguenther@suse.de>
31555 * passes.c (execute_function_todo): Verify loops if they are
31556 said to be up-to-date.
31557 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31558 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31560 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31563 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31564 handle calling assemble_external ourself.
31567 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31568 modes with zero size. Enhance comment.
31570 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31572 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31573 built-ins for vec_xl and vec_xst with short and char pointer
31576 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31578 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31579 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31580 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31581 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31582 (_mm_maskz_min_round_ss): New intrinsics.
31583 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31584 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31585 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31586 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31587 (__builtin_ia32_minss_mask_round): New builtins.
31588 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31589 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31590 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31592 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31593 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31595 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31598 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31600 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31601 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31602 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31603 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31604 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31605 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31606 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31607 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31608 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31609 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31610 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31611 (__builtin_ia32_mulss_mask_round): New builtins.
31612 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31613 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31614 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31616 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31617 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31619 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31622 2017-05-10 Julia Koval <julia.koval@intel.com>
31624 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31625 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31626 (_mm256_setr_m128i): New intrinsics.
31628 2017-05-10 Julia Koval <julia.koval@intel.com>
31630 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31631 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31632 (_mm_maskz_rcp14_ss): New intrinsics.
31633 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31634 (__builtin_ia32_rcp14ss_mask): New builtins.
31635 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31637 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31639 PR tree-optimization/51513
31640 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31641 (assert_unreachable_fallthru_edge_p): Use it.
31642 (group_case_labels_stmt): Likewise.
31643 * tree-cfg.h: Prototype it.
31644 * stmt.c: Include cfghooks.h and tree-cfg.h.
31645 (emit_case_dispatch_table) <gap_label>: New local variable.
31646 Use it to fill dispatch table gaps.
31647 Test for default_label before updating probabilities.
31648 (expand_case) <default_label>: Remove unneeded initialization.
31649 Test for unreachable default case statement and remove its edge.
31650 Set default_label accordingly.
31651 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31653 2017-05-10 Carl Love <cel@us.ibm.com>
31655 * config/rs6000/rs6000-c: Add support for built-in functions
31656 vector signed char vec_neg (vector signed char)
31657 vector signed short int vec_neg (vector short int)
31658 vector signed int vec_neg (vector signed int)
31659 vector signed long long vec_neg (vector signed long long)
31660 vector float vec_neg (vector float)
31661 vector double vec_neg (vector double)
31662 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31664 * config/rs6000/altivec.h: Add define for vec_neg
31665 * doc/extend.texi: Update the built-in documentation for the
31666 new built-in functions.
31668 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31670 PR tree-optimization/77644
31671 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31673 2017-05-10 Nathan Sidwell <nathan@acm.org>
31675 * dumpfile.h (TDI_lang_all): New.
31676 (TDF_KIND): New. Renumber others
31677 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31679 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31681 (get_dump_file_name): Adjust suffix generation.
31682 (dump_enable_all): Use TDF_KIND.
31683 * doc/invoke.texi (-fdump-lang-all): Document.
31685 * dumpfile.h: Tabify.
31687 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31690 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31691 Move member access before delete.
31693 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31695 * tree-inline.c (expand_call_inline): Split block at stmt
31698 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31701 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31702 are now unused after splitting mov{sf,sd}_hardfloat.
31703 (f32_lr2): Likewise.
31704 (f32_lm): Likewise.
31705 (f32_lm2): Likewise.
31706 (f32_li): Likewise.
31707 (f32_li2): Likewise.
31708 (f32_lv): Likewise.
31709 (f32_sr): Likewise.
31710 (f32_sr2): Likewise.
31711 (f32_sm): Likewise.
31712 (f32_sm2): Likewise.
31713 (f32_si): Likewise.
31714 (f32_si2): Likewise.
31715 (f32_sv): Likewise.
31716 (f32_dm): Likewise.
31717 (f32_vsx): Likewise.
31718 (f32_av): Likewise.
31719 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31720 For movsf, order stores so the VSX stores occur before the GPR
31721 store which encourages the register allocator to use a traditional
31722 FPR instead of a GPR. For movsd, order the stores so that the GPR
31723 store comes before the VSX stores to allow the power6 to work.
31724 This is due to the power6 not having a 32-bit integer store
31725 instruction from a FPR.
31726 (movsf_hardfloat): Likewise.
31727 (movsd_hardfloat): Likewise.
31729 2017-05-09 Martin Sebor <msebor@redhat.com>
31731 PR translation/80280
31732 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31735 PR translation/80280
31736 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31737 data member added in r247778.
31738 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31740 2017-05-09 Nathan Sidwell <nathan@acm.org>
31742 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31744 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31747 2017-05-09 Marek Polacek <polacek@redhat.com>
31749 * doc/invoke.texi: Fix typo.
31751 2017-05-09 Richard Biener <rguenther@suse.de>
31753 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31754 (vrp_val_is_min): Likewise.
31755 (set_value_range_to_value): Likewise.
31756 (set_value_range_to_nonnegative): Likewise.
31757 (gimple_assign_nonzero_p): Likewise.
31758 (gimple_stmt_nonzero_p): Likewise.
31759 (vrp_int_const_binop): Likewise. Remove unreachable case.
31760 (adjust_range_with_scev): Adjust comments.
31761 (compare_range_with_value): Likewise.
31762 (extract_range_from_phi_node): Likewise.
31763 (test_for_singularity): Likewise.
31765 2017-05-09 Richard Biener <rguenther@suse.de>
31767 * tree-vrp.c (get_single_symbol): Add assert that we don't
31768 get overflowed constants as invariant part.
31769 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31770 checks. Use wi::cmp instead of recursing for integer constants.
31771 (compare_values): Just ignore whether we assumed undefined
31772 overflow instead of failing the compare.
31773 (extract_range_for_var_from_comparison_expr): Add comment before the
31774 TREE_NO_WARNING sets.
31775 (test_for_singularity): Likewise.
31776 (extract_range_from_comparison): Do not disable optimization
31777 when we assumed undefined overflow.
31778 (extract_range_basic): Remove init of unused var.
31780 2017-05-09 Richard Biener <rguenther@suse.de>
31782 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31783 (extract_range_from_multiplicative_op_1): Adjust.
31784 (extract_range_from_binary_expr_1): Use int_const_binop.
31786 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31789 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31790 rs6000_store_data_bypass_p in seven define_bypass directives and
31791 in several comments.
31792 * config/rs6000/rs6000-protos.h: Add prototype for
31793 rs6000_store_data_bypass_p function.
31794 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31795 function implements slightly different (rs6000-specific) semantics
31796 than store_data_bypass_p, returning false rather than aborting
31797 with assertion error when arguments do not satisfy the
31798 requirements of store data bypass.
31799 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31800 rs6000_store_data_bypass_p.
31802 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31804 * config/xtensa/xtensa-protos.h
31805 (xtensa_initial_elimination_offset): New declaration.
31806 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31807 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31808 macro definition, add case for FRAME_POINTER_REGNUM when
31809 FRAME_GROWS_DOWNWARD.
31810 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31811 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31812 xtensa_initial_elimination_offset.
31814 2017-05-08 Nathan Sidwell <nathan@acm.org>
31816 * doc/invoke.texi: Alphabetize -fdump options.
31818 2017-05-08 Martin Sebor <msebor@redhat.com>
31820 PR translation/80280
31821 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31823 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31825 * target.def (compute_frame_layout): New optional target hook.
31826 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31827 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31828 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31830 * reload1.c (verify_initial_elim_offsets): Likewise.
31831 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31832 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31833 (arm_get_frame_offsets): Split up into this ...
31834 (arm_compute_frame_layout): ... and this function.
31836 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31838 * config/aarch64/constraints.md (Usa): New constraint.
31839 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31841 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31843 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31844 with_multilib_list after it has been checked.
31846 2017-05-08 Richard Biener <rguenther@suse.de>
31848 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31849 (bitmap_set_subtract_values): Likewise.
31851 2017-05-08 Richard Biener <rguenther@suse.de>
31853 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31854 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31856 (gimple_stmt_nonzero_warnv_p): Rename to ...
31857 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31859 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31860 (extract_range_basic): Adjust, do not disable propagation on
31861 strict overflow sensitive simplification.
31862 (vrp_visit_cond_stmt): Likewise.
31864 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31866 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31867 body size unconditionally.
31869 2017-05-07 Jeff Law <law@redhat.com>
31872 2017-05-06 Jeff Law <law@redhat.com>
31873 PR tree-optimization/78496
31874 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31877 PR tree-optimization/78496
31878 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31879 (simplify_stmt_using_ranges): Call it.
31880 (vrp_dom_walker::before_dom_children): Extract equivalences
31881 from an ASSERT_EXPR with an equality comparison against a
31884 2017-05-06 Jeff Law <law@redhat.com>
31886 PR tree-optimization/78496
31887 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31890 PR tree-optimization/78496
31891 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31892 (simplify_stmt_using_ranges): Call it.
31893 (vrp_dom_walker::before_dom_children): Extract equivalences
31894 from an ASSERT_EXPR with an equality comparison against a
31897 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31899 * lra-constraints.c (lra_copy_reg_equiv): New function.
31900 (split_reg): Use it to copy equivalence information from the
31901 original register to the spill register.
31903 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31905 PR rtl-optimization/75964
31906 * simplify-rtx.c (simplify_const_relational_operation): Remove
31907 invalid handling of comparisons of integer ABS.
31909 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31911 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31912 initialize to zero.
31913 (init_regs): Remove declaration.
31914 (function_arg_advance_32): Initialize error_p as boolean variable.
31916 2017-05-05 Nathan Sidwell <nathan@acm.org>
31918 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31919 lines. Use for (;;).
31921 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31923 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31924 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31925 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31926 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31927 VF=2 that require versioning.
31929 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31931 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31934 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31936 * diagnostic.h (diagnostic_override_option_index): Convert from
31937 macro to inline function.
31939 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31941 * diagnostic.c (last_module_changed_p): New function.
31942 (set_last_module): New function.
31943 (diagnostic_report_current_module): Convert macro usage to
31944 the above functions.
31945 * diagnostic.h (diagnostic_context::last_module): Strengthen
31946 from const line_map * to const line_map_ordinary *.
31947 (diagnostic_last_module_changed): Delete macro.
31948 (diagnostic_set_last_module): Delete macro.
31950 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31952 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31953 with diagnostic_report_diagnostic.
31954 (diagnostic_n_impl_richloc): Likewise.
31955 * diagnostic.h (report_diagnostic): Delete macro.
31956 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31957 with diagnostic_report_diagnostic.
31958 * substring-locations.c (format_warning_va): Likewise.
31960 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31962 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31963 save/restor of format_spec. Move option-printing code to...
31964 (print_option_information): ...this new function, and
31965 reimplement by simply printing to the pretty_printer,
31966 rather than appending to the format string.
31968 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31970 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31971 handling logic into...
31972 (update_effective_level_from_pragmas): ...this new function.
31974 2017-05-04 Andrew Waterman <andrew@sifive.com>
31976 * config/riscv/riscv.opt (mstrict-align): New option.
31977 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31978 (SLOW_UNALIGNED_ACCESS): Define.
31979 (riscv_slow_unaligned_access): Declare.
31980 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31982 (riscv_slow_unaligned_access): New variable.
31983 (rocket_tune_info): Set slow_unaligned_access to true.
31984 (optimize_size_tune_info): Set slow_unaligned_access to false.
31985 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31986 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31987 (riscv_option_override): Set riscv_slow_unaligned_access.
31988 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31990 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31992 * config/riscv/riscv.md: Unify indentation.
31994 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31999 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32001 (extendsi<mode>2): Add support for doing sign extension via
32002 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32003 don't have ISA 3.0 instructions.
32004 (extendsi<mode>2 splitter): Likewise.
32005 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32006 generate the normal insns since SImode can now go in vector
32007 registers. Disallow the special UNSPECs needed for previous
32008 machines to hide SImode being used. Add new insns
32009 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32010 (fix_trunc<mode>si2_stfiwx): Likewise.
32011 (fix_trunc<mode>si2_internal): Likewise.
32012 (fixuns_trunc<mode>si2): Likewise.
32013 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32014 (fctiw<u>z_<mode>_smallint): Likewise.
32015 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32016 of floating point to 32-bit integer from doing a direct move to
32017 the GPR registers to do a store.
32018 (fctiwz_<mode>): Break long line.
32020 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32022 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32023 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32024 (addr_list, addr_offset_valid_p): New.
32025 (split_address_groups): Check offset validity with above function.
32026 (gt-tree-ssa-loop-ivopts.h): Include header file.
32028 2017-05-05 Nathan Sidwell <nathan@acm.org>
32030 * config.gcc (arm*-*-*): Add missing 'fi'.
32032 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32034 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32035 included in -fopt-info does not matter.
32036 * doc/optinfo.texi (-fopt-info): Fix description of default
32037 behavour. Explicitly say order of options included in -fopt-info
32040 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32042 * config.gcc: Allow combinations of aprofile and rmprofile values for
32043 --with-multilib-list.
32044 * config/arm/t-multilib: New file.
32045 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32046 variables. Remove setting of ISA and floating-point ABI in
32047 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32048 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32049 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32050 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32052 * config/arm/t-rmprofile: Likewise except for the matches changes.
32053 * doc/install.texi (--with-multilib-list): Document the combination of
32054 aprofile and rmprofile values and warn about pitfalls in doing that.
32056 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32058 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32059 (movdi_aarch64): Likewise.
32061 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32063 PR tree-optimization/80632
32064 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32066 (build_arrays): Initialize it for virtual phis.
32067 (fix_phi_nodes): Use it for virtual phis.
32069 PR tree-optimization/80558
32070 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32071 [x, y] op z into [x op, y op z] for op & or | if conditions
32074 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32075 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32078 * config/arm/arm.md (use_literal_pool): Remove.
32079 (64-bit immediate split): No longer takes cost into consideration
32080 if arm_disable_literal_pool is enabled.
32081 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32082 used when arm_disable_literal_pool is enabled.
32083 (arm_max_const_double_inline_cost): Remove use of
32084 arm_disable_literal_pool.
32085 (push_minipool_fix): Add assert.
32086 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32087 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32088 (no_literal_pool_sf_immediate): New.
32090 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32092 PR tree-optimization/80613
32093 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32094 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32096 2017-05-05 Richard Biener <rguenther@suse.de>
32098 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32100 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32102 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32103 of this flag from insn conditions due to removal from r247495.
32105 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32107 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32109 (arm_early_store_addr_dep_ptr): Likewise.
32110 * config/arm/aarch-common-protos.h
32111 (arm_early_load_addr_dep_ptr): Add prototype.
32112 (arm_early_store_addr_dep_ptr): Likewise.
32113 * config/arm/cortex-a53.md: Add new bypasses.
32115 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32117 * tree.c (next_type_uid): Change type to unsigned.
32118 (type_hash_canon): Decrement back next_type_uid if
32119 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32120 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32123 2017-05-04 Martin Sebor <msebor@redhat.com>
32125 * builtins.c: Fix a trivial typo in a comment.
32127 PR middle-end/79234
32128 * builtins.c (check_sizes): Adjust to handle reading past the end.
32129 Avoid printing excessive upper bound of ranges. Use %E to print
32130 tree nodes instead of converting them to %wu.
32131 (expand_builtin_memchr): New function.
32132 (compute_dest_size): Rename...
32133 (compute_objsize): ...to this.
32134 (expand_builtin_memcpy): Adjust.
32135 (expand_builtin_mempcpy): Adjust.
32136 (expand_builtin_strcat): Adjust.
32137 (expand_builtin_strcpy): Adjust.
32138 (check_strncat_sizes): Adjust.
32139 (expand_builtin_strncat): Adjust.
32140 (expand_builtin_strncpy): Adjust and simplify.
32141 (expand_builtin_memset): Adjust.
32142 (expand_builtin_bzero): Adjust.
32143 (expand_builtin_memcmp): Adjust.
32144 (expand_builtin): Handle memcmp.
32145 (maybe_emit_chk_warning): Check strncat just once.
32147 2017-05-04 Martin Sebor <msebor@redhat.com>
32149 PR preprocessor/79214
32150 PR middle-end/79222
32151 PR middle-end/79223
32152 * builtins.c (check_sizes): Add inlining context and issue
32153 warnings even when -Wno-system-headers is set.
32154 (check_strncat_sizes): Same.
32155 (expand_builtin_strncat): Same.
32156 (expand_builtin_memmove): New function.
32157 (expand_builtin_stpncpy): Same.
32158 (expand_builtin): Handle memmove and stpncpy.
32160 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32162 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32163 which is not used any more.
32165 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32167 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32169 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32171 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32172 (cortexa53_tunings): Likewise.
32173 (cortexa57_tunings): Likewise.
32174 (cortexa72_tunings): Likewise.
32175 (cortexa73_tunings): Likewise.
32177 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32179 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32180 Set loop alignment to 8.
32182 2017-05-04 Martin Sebor <msebor@redhat.com>
32184 PR translation/80280
32185 * builtins.c (expand_builtin_object_size): Add missing quoting to
32186 %D and like directives.
32187 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32188 (hsa_type_for_tree_type): Same.
32189 (verify_function_arguments): Same.
32190 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32191 * varasm.c (get_section): Same.
32194 2017-05-04 Martin Sebor <msebor@redhat.com>
32196 PR translation/80280
32197 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32199 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32201 * config/aarch64/aarch64.c (generic_addrcost_table):
32202 Change HI/TI mode setting.
32204 2017-05-04 Martin Jambor <mjambor@suse.cz>
32206 PR tree-optimization/80622
32207 * tree-sra.c (comes_initialized_p): New function.
32208 (build_accesses_from_assign): Only set write lazily when
32209 comes_initialized_p is false.
32210 (analyze_access_subtree): Use comes_initialized_p.
32211 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32212 instead of testing for PARM_DECL.
32214 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32216 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32217 constraint on operand 0 to allow more general addressing modes.
32218 Adjust output template.
32219 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32221 * config/aarch64/aarch64-protos.h
32222 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32223 * config/aarch64/constraints.md (Dp): New address constraint.
32224 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32227 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32229 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32230 update use of estimate_ipcp_clone_size_and_time.
32231 (estimate_local_effects): Update use of
32232 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32233 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32234 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32235 Return nonspecialized time.
32237 2017-05-04 Richard Biener <rguenther@suse.de>
32239 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32240 for the last VUSE which def dominates the PHI. Directly call
32242 (get_continuation_for_phi_1): Remove.
32244 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32246 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32247 to explain the use of truncating division. Cap the number of
32248 iterations to the maximum given by nb_iterations_upper_bound,
32251 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32253 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32254 * configure: Regenerate.
32255 * config.in: Regenerate.
32256 * config/i386/driver-mingw32.c: new file.
32257 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32258 * config.host: Link driver-mingw32.o on MinGW host.
32259 * doc/install.texi: Document new --enable-mingw-wildcard configure
32262 2017-05-04 Marek Polacek <polacek@redhat.com>
32264 PR tree-optimization/80612
32265 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32267 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32268 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32270 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32271 (movt splitter): Likewise.
32272 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32273 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32274 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32275 block for Thumb-1 with MOVT.
32276 (thumb2_legitimate_address_p): Move code block ...
32277 (can_avoid_literal_pool_for_label_p): ... into this new function.
32278 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32280 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32281 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32282 "M-profile targets with the MOVT instruction".
32284 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32286 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32287 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32288 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32290 2017-05-04 Martin Liska <mliska@suse.cz>
32292 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32293 variable cond_code.
32295 2017-05-04 Richard Biener <rguenther@suse.de>
32297 * tree.c (array_at_struct_end_p): Handle arrays at struct
32298 end with flexarrays more conservatively. Refactor and treat
32299 arrays of arrays or aggregates more strict. Fix
32300 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32301 * tree.c (array_at_struct_end_p): Adjust prototype.
32302 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32303 * gimple-fold.c (get_range_strlen): Likewise.
32304 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32306 2017-05-04 Richard Biener <rguenther@suse.de>
32308 PR tree-optimization/31130
32309 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32311 (supports_overflow_infinity): Likewise.
32312 (is_negative_overflow_infinity): Likewise.
32313 (is_positive_overflow_infinity): Likewise.
32314 (is_overflow_infinity): Likewise.
32315 (stmt_overflow_infinity): Likewise.
32316 (overflow_infinity_range_p): Likewise.
32317 (usable_range_p): Remove as always returning true.
32318 (make_overflow_infinity): Remove.
32319 (negative_overflow_infinity): Likewise.
32320 (positive_overflow_infinity): Likewise.
32321 (avoid_overflow_infinity): Likewise.
32322 (set_value_range): Adjust accordingly.
32323 (set_value_range_to_nonnegative): Likewise, remove now unused
32324 overflow_infinity arg.
32325 (vrp_operand_equal_p): Adjust.
32326 (update_value_range): Likewise.
32327 (range_int_cst_singleton_p): Likewise.
32328 (operand_less_p): Likewise.
32329 (compare_values_warnv): Likewise.
32330 (extract_range_for_var_from_comparison_expr): Likewise.
32331 (vrp_int_const_binop): Likewise.
32332 (zero_nonzero_bits_from_vr): Likewise.
32333 (extract_range_from_multiplicative_op_1): Likewise.
32334 (extract_range_from_binary_expr_1): Likewise.
32335 (extract_range_from_unary_expr): Likewise.
32336 (extract_range_from_comparison): Likewise.
32337 (extract_range_basic): Likewise.
32338 (adjust_range_with_scev): Likewise.
32339 (compare_ranges): Likewise.
32340 (compare_range_with_value): Likewise.
32341 (dump_value_range): Likewise.
32342 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32344 (simplify_cond_using_ranges): Adjust.
32346 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32348 * brig-builtins.def: Added a builtin for class_f64.
32349 * builtin-types.def: Added a builtin type needed by class_f64.
32351 2017-05-03 Jason Merrill <jason@redhat.com>
32353 * timevar.def: Add TV_CONSTEXPR.
32355 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32357 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32359 2017-05-03 Martin Jambor <mjambor@suse.cz>
32361 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32362 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32363 * ipa-cp.c (ipcp_propagate_stage): Do not call
32364 ipa_update_after_lto_read.
32365 * ipa-inline.c (ipa_inline): Likewise.
32367 2017-05-03 Martin Jambor <mjambor@suse.cz>
32369 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32370 tag. Added a default constructor and a destructor.
32371 (ipa_edge_args_sum_t): New class;
32372 (ipa_edge_args_sum): Declare.
32373 (ipa_edge_args_vector): Remove declaration.
32374 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32375 (ipa_free_edge_args_substructures): Remove declaration.
32376 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32377 (ipa_edge_args_info_available_for_edge_p): Likewise.
32378 * ipa-prop.c (ipa_edge_args_vector): Removed.
32379 (edge_removal_hook_holder): Likewise.
32380 (edge_duplication_hook_holder): Likewise.
32381 (ipa_edge_args_sum): New variable.
32382 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32383 ipa_edge_args_vector.
32384 (ipa_free_edge_args_substructures): Likewise.
32385 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32386 ipa_edge_args_vector.
32387 (ipa_edge_removal_hook): Turned into method
32388 ipa_edge_args_sum_t::remove.
32389 (ipa_edge_duplication_hook): Turned into method
32390 ipa_edge_args_sum_t::duplicate.
32391 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32392 registering edge hooks.
32393 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32394 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32395 ipa_edge_args_sum instead of ipa_edge_args_vector.
32396 * ipa-profile.c (ipa_profile): Likewise.
32398 2017-05-03 Martin Jambor <mjambor@suse.cz>
32400 * symbol-summary.h (function_summary): New method exists.
32401 (function_summary::symtab_removal): Deallocate through release.
32402 (call_summary): New class.
32403 (gt_ggc_mx): New overload.
32404 (gt_pch_nx): Likewise.
32405 (gt_pch_nx): Likewise.
32407 2017-05-03 Jeff Law <law@redhat.com>
32409 PR tree-optimization/78496
32410 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32411 from simplify_cond_using_ranges. Split off code to walk
32412 backwards through casts into ...
32413 (simplify_cond_using_ranges_2): New function.
32414 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32415 (execute_vrp): After identifying jump threads, call
32416 simplify_cond_using_ranges_2.
32418 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32421 * ipa-inline.h (inline_summary): Add ctor.
32422 (create_ggc): Do not use ggc_cleared_alloc.
32424 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32425 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32427 * gcc.c (handle_braces): Support escaping in switch matching
32429 * doc/invoke.texi (Spec Files): Document it.
32430 Remove superfluous @code markup in items.
32432 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32434 * diagnostic-show-locus.c (struct column_range): New struct.
32435 (get_affected_columns): New function.
32436 (get_printed_columns): New function.
32437 (struct correction): New struct.
32438 (correction::ensure_capacity): New function.
32439 (correction::ensure_terminated): New function.
32440 (struct line_corrections): New struct.
32441 (line_corrections::~line_corrections): New dtor.
32442 (line_corrections::add_hint): New function.
32443 (layout::print_trailing_fixits): Reimplement in terms of the new
32445 (selftest::test_overlapped_fixit_printing): New function.
32446 (selftest::diagnostic_show_locus_c_tests): Call it.
32448 2017-05-03 Nathan Sidwell <nathan@acm.org>
32450 Canonicalize canonical type hashing
32451 * tree.h (type_hash_canon_hash): Declare.
32452 * tree.c (type_hash_list, attribute_hash_list): Move into
32453 type_hash_canon_hash.
32454 (build_type_attribute_qual_variant): Break out hash code calc into
32455 type_hash_canon_hash.
32456 (type_hash_canon_hash): New. Generic type hash computation.
32457 (build_range_type_1, build_array_type_1, build_function_type,
32458 build_method_type_directly, build_offset_type, build_complex_type,
32459 make_vector_type): Call it.
32461 2017-05-03 Richard Biener <rguenther@suse.de>
32463 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32464 When all DRs have unknown misaligned do not always peel
32465 when there is a store but apply the same costing model as if
32466 there were only loads.
32468 2017-05-03 Richard Biener <rguenther@suse.de>
32471 PR tree-optimization/80492
32472 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32473 compare_base_decls returning dont-know properly.
32475 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32477 * config/arm/iterators.md (CCSI): New mode iterator.
32478 (arch): New mode attribute.
32479 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32480 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32481 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32482 code iterator for success result mode.
32483 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32484 the corresponding new insn generators.
32486 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32489 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32490 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32492 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32494 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32495 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32496 (DDR_A): Wrap DDR argument in brackets.
32497 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32498 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32499 (DDR_REVERSED_P): Likewise.
32501 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32503 PR tree-optimization/79472
32504 * tree-switch-conversion.c (struct switch_conv_info): Add
32505 contiguous_range and default_case_nonstandard fields.
32506 (collect_switch_conv_info): Compute contiguous_range and
32507 default_case_nonstandard fields, don't clear final_bb if
32508 contiguous_range and only the default case doesn't have the required
32510 (check_all_empty_except_final): Set default_case_nonstandard instead
32511 of failing if contiguous_range and the default case doesn't have empty
32513 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32514 and only the default case doesn't have the required constants. Skip
32516 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32517 if default_case_nonstandard.
32518 (build_constructors): Build constant 1 just once. Assert that default
32519 values aren't inserted in between cases if contiguous_range. Skip
32521 (build_arrays): Skip virtual phis.
32522 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32523 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32524 Handle virtual phis.
32525 (gen_inbound_check): Handle default_case_nonstandard case.
32526 (process_switch): Adjust check_final_bb caller. Call
32527 gather_default_values with the first non-default case instead of
32528 default case if default_case_nonstandard.
32530 2017-05-02 Nathan Sidwell <nathan@acm.org>
32532 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32533 check. Fix formatting.
32535 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32537 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32538 errors when comparing specialized and unspecialized times.
32540 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32542 * diagnostic-show-locus.c
32543 (layout::should_print_annotation_line_p): Make private.
32544 (layout::print_annotation_line): Make private.
32545 (layout::annotation_line_showed_range_p): Make private.
32546 (layout::show_ruler): Make private.
32547 (layout::print_source_line): Make private. Pass in line and
32548 line_width, rather than calling location_get_source_line. Drop
32550 (layout::print_leading_fixits): New method.
32551 (layout::print_any_fixits): Rename to...
32552 (layout::print_trailing_fixits): ...this, and make private.
32553 Don't print newline fixits.
32554 (diagnostic_show_locus): Move logic for printing one row into...
32555 (layout::print_line): ...this new function. Move the
32556 location_get_source_line call and error-handling from
32557 print_source_line to here. Call print_leading_fixits, and rename
32558 print_any_fixits to print_trailing_fixits.
32559 (selftest::test_fixit_insert_containing_newline): Update now that
32560 newlines are partially supported.
32561 (selftest::test_fixit_insert_containing_newline_2): New test.
32562 (selftest::test_fixit_replace_containing_newline): Update comments.
32563 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32564 * edit-context.c (class added_line): New class.
32565 (class edited_line): Describe newline handling in comment.
32566 (edited_line::actually_edited_p): New method.
32567 (edited_line::print_content): Delete redundant decl.
32568 (edited_line::m_predecessors): New field.
32569 (edited_file::print_content): Call edited_line::print_content.
32570 (edited_file::print_diff): Update to support newlines.
32571 (edited_file::print_diff_hunk): Likewise.
32572 (edited_file::print_run_of_changed_lines): New function.
32573 (edited_file::print_diff_line): Convert to...
32574 (print_diff_line): ...this.
32575 (edited_file::get_effective_line_count): New function.
32576 (edited_line::edited_line): Initialize new field m_predecessors.
32577 (edited_line::~edited_line): Clean up m_predecessors.
32578 (edited_line::apply_fixit): Handle newlines.
32579 (edited_line::get_effective_line_count): New function.
32580 (edited_line::print_content): New function.
32581 (edited_line::print_diff_lines): New function.
32582 (selftest::test_applying_fixits_insert_containing_newline): New
32584 (selftest::test_applying_fixits_replace_containing_newline): New
32586 (selftest::insert_line): New function.
32587 (selftest::test_applying_fixits_multiple_lines): Add example of
32589 (selftest::edit_context_c_tests): Call the new tests.
32591 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32593 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32594 parameter cand. Update dump information.
32595 (get_computation_cost): Update uses.
32597 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32599 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32600 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32601 (get_computation_at, rewrite_use_address): Update use of
32602 get_computation_aff.
32604 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32606 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32607 (get_computation): Delete.
32608 (get_computation_cost): Implement like get_computation_cost_at.
32609 Use get_computation_at.
32610 (get_computation_cost_at): Delete.
32611 (rewrite_use_nonlinear_expr): Use get_computation_at.
32612 (rewrite_use_compare, remove_unused_ivs): Ditto.
32614 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32616 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32618 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32620 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32621 (ivopts_global_cost_for_size): Rename parameter and update uses.
32622 (iv_ca_recount_cost): Update uses.
32623 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32624 candidates seperately in n_invs and n_cands.
32625 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32627 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32629 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32630 (find_inv_vars_cb): New.
32631 (find_depends): Renamed to ...
32632 (find_inv_vars): ... this.
32633 (add_candidate_1, force_var_cost): Call find_inv_vars.
32634 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32636 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32638 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32639 inv_vars. Add inv_exprs.
32640 (struct iv_cand): Rename depends_on to inv_vars.
32641 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32642 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32643 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32644 (dump_cand): Dump inv_vars.
32645 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32646 (record_invariant, find_depends, add_candidate_1): Ditto.
32647 (set_group_iv_cost, force_var_cost): Ditto.
32648 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32649 (get_computation_cost_at, get_computation_cost): Ditto.
32650 (determine_group_iv_cost_generic): Ditto.
32651 (determine_group_iv_cost_address): Ditto.
32652 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32653 (determine_group_iv_costs): Ditto.
32654 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32655 (iv_ca_set_remove_invariants): Renamed to ...
32656 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32657 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32658 (iv_ca_set_add_invariants): Renamed to ...
32659 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32660 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32661 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32662 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32663 (create_new_ivs): Remove useless dump.
32665 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32667 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32669 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32670 (iv_ca_set_no_cp, create_new_iv): Ditto.
32672 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32674 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32676 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32678 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32679 function tree_check2.
32681 2017-05-02 Martin Liska <mliska@suse.cz>
32683 * doc/gcov.texi: Add missing preposition.
32684 * gcov.c (function_info::function_info): Properly fill up
32685 all member variables.
32687 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32689 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32691 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32693 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32695 2017-05-02 Martin Liska <mliska@suse.cz>
32698 * lto-streamer-in.c (lto_read_tree_1): Remove
32699 LTO_STREAMER_DEBUG.
32700 * lto-streamer.c (struct tree_hash_entry): Likewise.
32701 (struct tree_entry_hasher): Likewise.
32702 (tree_entry_hasher::hash): Likewise.
32703 (tree_entry_hasher::equal): Likewise.
32704 (lto_streamer_init): Likewise.
32705 (lto_orig_address_map): Likewise.
32706 (lto_orig_address_get): Likewise.
32707 (lto_orig_address_remove): Likewise.
32708 * lto-streamer.h: Likewise.
32709 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32710 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32712 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32714 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32715 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32716 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32717 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32718 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32719 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32720 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32721 (mm_maskz_sub_ss): New intrinsics.
32722 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32723 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32724 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32725 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32726 (__builtin_ia32_subss_mask_round): New builtins.
32727 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32728 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32729 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32731 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32732 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32734 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32737 2017-05-02 Martin Jambor <mjambor@suse.cz>
32739 PR tree-optimization/78687
32740 * tree-sra.c (access): New field parent.
32741 (process_subtree_disqualification): New function.
32742 (disqualify_candidate): Call it.
32743 (build_accesses_from_assign): Reset write flag if creating an
32745 (build_access_subtree): Fill in parent field and also prpagate
32746 down grp_write flag.
32747 (create_artificial_child_access): New parameter set_grp_write, set
32748 grp_write to its value.
32749 (propagate_subaccesses_across_link): Also propagate grp_write flag
32751 (propagate_all_subaccesses): Push the closest parent back to work
32752 queue if add_access_to_work_queue returned true.
32754 2017-05-02 Richard Biener <rguenther@suse.de>
32756 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32757 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32758 -fstrict-overflow documentation.
32759 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32760 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32761 flag_strict_overflow.
32762 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32763 * lto-opts.c (lto_write_options): Do not stream it.
32764 * lto-wrapper.c (merge_and_complain): Do not handle it.
32765 * opts.c (default_options_table): Do not set -fstrict-overflow.
32766 (finish_options): Likewise do not clear it when sanitizing.
32767 * simplify-rtx.c (simplify_const_relational_operation): Do not
32768 test flag_strict_overflow.
32770 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32772 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32773 using enabled attribute.
32774 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32775 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32776 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32777 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32778 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32779 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32780 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32781 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32782 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32783 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32785 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32787 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32789 2017-05-02 Richard Biener <rguenther@suse.de>
32791 PR tree-optimization/80591
32793 2017-04-10 Richard Biener <rguenther@suse.de>
32795 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32798 2017-05-02 Richard Biener <rguenther@suse.de>
32800 PR tree-optimization/80549
32801 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32802 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32803 headers if they do not have a preheader.
32805 2017-05-02 Martin Liska <mliska@suse.cz>
32808 * common.opt: Fix typo.
32809 * doc/invoke.texi: Likewise.
32811 2017-05-01 Jan Beulich <jbeulich@suse.com>
32813 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32814 swapping, add (x,x,m,x,n) alternative.
32816 2017-05-01 Nathan Sidwell <nathan@acm.org>
32818 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32819 unnecessary unadjusted_alignment check.
32821 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32824 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32826 * gimplify.c (gimplify_cilk_detach): New function.
32827 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32828 * tree-core.h: Document EXPR_CILK_SPAWN.
32829 * tree.h (EXPR_CILK_SPAWN): Define.
32831 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32833 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32834 to use new fixit_hint representation, using the "replace" logic.
32835 (get_line_span_for_fixit_hint): Likewise.
32836 (layout::print_any_fixits): Likewise.
32837 (selftest::test_one_liner_many_fixits): Rename to...
32838 (selftest::test_one_liner_many_fixits_1): ...this, and update
32839 comment and expected output to reflect that the multiple fix-it
32840 hints are now consolidated into one insertion.
32841 (selftest::test_one_liner_many_fixits_2): New test.
32842 (selftest::test_diagnostic_show_locus_one_liner): Update for
32844 (selftest::test_fixit_consolidation): Update for fix-it API
32846 * diagnostic.c (print_parseable_fixits): Likewise.
32847 * edit-context.c (edited_line::m_line_events): Convert from
32848 auto_vec <line_event *> to auto_vec <line_event>.
32849 (class line_event): Convert from abstract base class to a concrete
32850 class, taking over the role of replace_event.
32851 (class insert_event): Delete.
32852 (class replace_event): Rename to class line_event. Convert to
32854 (edit_context::add_fixits): Reimplement.
32855 (edit_context::apply_insert): Delete.
32856 (edit_context::apply_replace): Rename to...
32857 (edit_context::apply_fixit): ...this. Convert to half-open range.
32858 (edited_file::apply_insert): Delete.
32859 (edited_file::apply_replace): Rename to...
32860 (edited_file::apply_fixit): ...this.
32861 (edited_line::~edited_line): Drop deletion of events.
32862 (edited_line::apply_insert): Delete.
32863 (edited_line::apply_replace): Rename to...
32864 (edited_line::apply_fixit): ...this. Convert to half-open range.
32865 Update for change to type of m_line_events.
32866 * edit-context.h (edit_context::apply_insert): Delete.
32867 (edit_context::apply_replace): Rename to...
32868 (edit_context::apply_fixit): ...this.
32870 2017-05-01 Martin Sebor <msebor@redhat.com>
32872 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32875 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32878 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32879 __get_cpuid_max returns 0.
32880 (__get_cpuid_count): Ditto.
32882 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32884 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32885 replacement expression is another instance of one of its arguments.
32887 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32890 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32891 check for stack push/pop autoinc.
32892 * config/i386/i386.c (ix86_agi_dependent): Return false
32893 if the only reason why modified_in_p returned true is that
32894 addr is SP based and set_insn is a push or pop.
32896 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32898 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32901 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32904 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32905 (account_size_time): Use two predicates - exec_pred and
32907 (evaluate_conditions_for_known_args): Compute both clause and
32909 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32910 (inline_summary_t::duplicate): Update.
32911 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32913 (compute_inline_parameters): Likewise.
32914 (estimate_edge_size_and_time): Update caluclation of time.
32915 (estimate_node_size_and_time): Compute both time and nonspecialized
32917 (estimate_ipcp_clone_size_and_time): Update.
32918 (inline_merge_summary): Update.
32919 (do_estimate_edge_time): Update.
32920 (do_estimate_edge_size): Update.
32921 (do_estimate_edge_hints): Update.
32922 (inline_read_section, inline_write_summary): Stream both new predicates.
32923 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32925 (compute_inlined_call_time): Cleanup.
32926 (big_speedup_p): Update.
32927 (edge_badness): Update.
32928 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32929 (size_time_entry): Replace predicate by exec_predicate and
32930 nonconst_predicate.
32931 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32932 (estimate_edge_time): Return also nonspec_time.
32933 (reset_edge_growth_cache): Update.
32935 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32937 PR rtl-optimization/80491
32938 * ifcvt.c (noce_process_if_block): When looking for x setter
32939 with missing else_bb, don't check only the insn right before
32940 cond_earliest, but look for the last insn that x is modified in
32941 within the same bb.
32943 PR rtl-optimization/80491
32944 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32946 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32948 PR tree-optimization/80487
32949 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32951 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32953 PR tree-optimization/79697
32954 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32955 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32956 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32958 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32959 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32961 2017-04-28 Martin Sebor <msebor@redhat.com>
32963 PR tree-optimization/80523
32964 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32965 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32967 (maybe_warn, format_directive, parse_directive): Use new functions.
32968 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32970 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32972 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32974 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32976 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32977 target_header_dir): Set correctly.
32978 * configure: Regenerated.
32979 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32980 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32981 instead of SYSTEM_HEADER_DIR.
32983 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32985 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32986 (estimate_local_effects): Likewise.
32987 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32988 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32989 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32990 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32991 do_estimate_edge_time, estimate_edge_time): Likewise.
32992 * ipa-inline-analysis.c (estimate_node_size_and_time,
32993 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32994 (estimate_time_after_inlining): Remove.
32996 2017-04-28 Martin Liska <mliska@suse.cz>
32998 * doc/gcov.texi: Enhance documentation of gcov.
33000 2017-04-28 Martin Liska <mliska@suse.cz>
33002 * doc/gcov.texi: Sort options in alphabetic order.
33003 * doc/gcov-dump.texi: Likewise.
33004 * doc/gcov-tool.texi: Likewise.
33005 * gcov.c (print_usage): Likewise.
33006 * gcov-dump.c (print_usage): Likewise.
33007 * gcov-tool.c (print_merge_usage_message): Likewise.
33008 (print_rewrite_usage_message): Likewise.
33009 (print_overlap_usage_message): Likewise.
33011 2017-04-28 Martin Liska <mliska@suse.cz>
33013 PR gcov-profile/53915
33014 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33016 2017-04-28 Martin Liska <mliska@suse.cz>
33018 PR gcov-profile/79891
33019 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33020 is marked by compiler as living on a line.
33021 (get_cycles_count): Remove usage of the union.
33022 (output_intermediate_file): Likewise.
33023 (find_source): Fix GNU coding style.
33024 (accumulate_line_counts): Remove old non-all block mode.
33025 (output_lines): Remove usage of the union.
33026 * profile.c (output_location): Include all BBs, even if
33027 belonging to a same line (and file) as a previous BB.
33029 2017-04-28 Martin Liska <mliska@suse.cz>
33031 * gcov.c (process_args): Handle new argument 'w'.
33032 (read_graph_file): Assign ID to BBs.
33033 (output_branch_count): Display BB # if verbose flag is set.
33034 (output_lines): Likewise for arcs.
33035 (print_usage): Add '--verbose' option help.
33036 * doc/gcov.texi: Document --verbose (-w) option.
33038 2017-04-28 Martin Liska <mliska@suse.cz>
33040 * gcov.c (struct block_location_info): New struct.
33041 (process_file): Fill up the new structure.
33042 (read_graph_file): Replace usage of encoding by the newly added
33044 (add_line_counts): Likewise.
33045 (accumulate_line_counts): Remove usage of the union.
33046 (function_info::function_info): New function.
33047 (function_info::~function_info): Likewise.
33048 (process_file): Call delete instead of release_function.
33049 (release_function): Release the function.
33050 (release_structures): Call delete instead of release_function.
33051 (solve_flow_graph): Replace usage of num_blocks.
33052 (find_exception_blocks): Likewise.
33053 (output_lines): Fix GNU coding style.
33055 2017-04-28 Martin Liska <mliska@suse.cz>
33058 * coverage.c (coverage_remove_note_file): New function.
33059 * coverage.h: Declare the function.
33060 * toplev.c (finalize): Clean if an error has been seen.
33062 2017-04-28 Martin Liska <mliska@suse.cz>
33064 PR gcov-profile/80031
33065 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33066 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33067 * gcov.c (read_graph_file): Read just number of blocks.
33068 * profile.c (branch_prob): Do not stream 0 flags per a basic
33071 2017-04-28 Martin Liska <mliska@suse.cz>
33073 * gcov-dump.c (tag_*): Add new argument to declarations.
33074 (dump_gcov_file): Likewise.
33075 (tag_blocks): Add and use new argument depth.
33076 (tag_arcs): Likewise.
33077 (tag_lines): Likewise.
33078 (tag_counters): Likewise.
33079 (tag_summary): Likewise.
33080 (dump_working_sets): Use depth to do a proper indentation.
33082 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33085 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33086 * symtab.c (symtab_node::debug_symtab): Move definition here.
33088 2017-04-28 Richard Biener <rguenther@suse.de>
33090 * lto-streamer.h (LTO_major_version): Bump to 7.
33092 2017-04-28 Richard Biener <rguenther@suse.de>
33094 * tree-vrp.c (assert_info): New struct.
33095 (add_assert_info): New helper.
33096 (register_edge_assert_for_2): Refactor to add asserts to a vector
33098 (register_edge_assert_for_1): Likewise.
33099 (register_edge_assert_for): Likewise.
33100 (finish_register_edge_assert_for): New helper actually registering
33101 asserts where live on edge.
33102 (find_conditional_asserts): Adjust.
33103 (find_switch_asserts): Likewise.
33104 (evrp_dom_walker::try_find_new_range): Generalize.
33105 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33107 2017-04-27 Marek Polacek <polacek@redhat.com>
33110 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33111 arg10 and arg11 to itype.
33113 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33115 * doc/extend.texi (Object Size Checking): Improve grammar.
33117 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33120 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33121 that the logic for permitting reciprocal estimates matches that
33124 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33127 * tree.c (type_cache_hasher::equal): Only compare
33128 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33129 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33130 non-aggregate element types.
33131 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33132 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33134 2017-04-27 Richard Biener <rguenther@suse.de>
33136 PR middle-end/80533
33137 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33138 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33139 keeping a reference to a trailing array.
33141 2017-04-27 Richard Biener <rguenther@suse.de>
33143 PR middle-end/80539
33144 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33145 being in loop-closed SSA form conservatively.
33146 (chrec_fold_multiply_poly_poly): Likewise.
33148 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33150 PR middle-end/79665
33151 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33152 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33154 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33157 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33158 (aarch64_function_arg_alignment): Return unsigned int again, but still
33159 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33160 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33161 Don't emit -Wpsabi note.
33162 (aarch64_function_arg_boundary): Likewise.
33163 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33166 2017-04-26 Nathan Sidwell <nathan@acm.org>
33168 * tree.h (crc32_unsigned_n): Declare.
33169 (crc32_unsigned, crc32_unsigned): Make inline.
33170 * tree.c (crc32_unsigned_bits): Replace with ...
33171 (crc32_unsigned_n): ... this.
33172 (crc32_unsigned, crc32_byte): Remove.
33173 (crc32_string): Remove unnecessary braces.
33175 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33177 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33178 * ipa-inline-analysis.c (MAX_TIME): Remove.
33179 (account_size_time): Use sreal for time.
33180 (dump_inline_summary): Update.
33181 (estimate_function_body_sizes): Update.
33182 (estimate_edge_size_and_time): Update.
33183 (estimate_calls_size_and_time): Update.
33184 (estimate_node_size_and_time): Update.
33185 (inline_merge_summary): Update.
33186 (inline_update_overall_summary): Update.
33187 (estimate_time_after_inlining): Update.
33188 (inline_read_section): Update.
33189 (inline_write_summary): Update.
33190 * ipa-inline.c (compute_uninlined_call_time): Update.
33191 (compute_inlined_call_time): Update.
33192 (recursive_inlining): Update.
33193 (inline_small_functions): Update.
33194 (dump_overall_stats): Update.
33195 * ipa-inline.h: Include sreal.h.
33196 (size_time_entry): Turn time to sreal.
33197 (inline_summary): Turn self_time nad time to sreal.
33199 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33201 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33203 (sreal::stream_out, sreal::stream_in): New.
33204 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33206 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33208 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33211 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33214 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33216 (dimode_scalar_chain::compute_convert_gain): Ditto.
33217 (dimode_scalar_chain::make_vector_copies): Ditto.
33218 (dimode_scalar_chain::convert_reg): Ditto.
33219 (dimode_scalar_chain::convert_insn): Ditto.
33220 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33221 (VI248_AVX512BW_1): New mode iterator.
33222 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33223 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33226 2017-04-25 Martin Sebor <msebor@redhat.com>
33228 PR tree-optimization/80497
33229 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33230 constants are representable in HOST_WIDE_INT.
33231 (parse_directive): Ditto.
33233 2017-04-25 Martin Sebor <msebor@redhat.com>
33236 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33237 (new_zero_array): Adjust signature.
33238 (dom_info::dom_init): Used unsigned rather that size_t.
33239 (dom_info::dom_info): Same.
33241 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33242 Jakub Jelinek <jakub@redhat.com>
33245 * config/arm/arm.c: Include gimple.h.
33246 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33247 returns negative, increment ncrn only if it returned positive.
33248 (arm_needs_doubleword_align): Return int instead of bool,
33249 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33250 members, but if there is any such non-FIELD_DECL
33251 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33252 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33253 returns negative, increment nregs only if it returned positive.
33254 (arm_setup_incoming_varargs): Likewise.
33255 (arm_function_arg_boundary): Emit -Wpsabi note if
33256 arm_needs_doubleword_align returns negative, return
33257 DOUBLEWORD_ALIGNMENT only if it returned positive.
33259 2017-04-25 Marek Polacek <polacek@redhat.com>
33262 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33263 first argument to type.
33265 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33268 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33269 type checks to test for compatibility instead of equality.
33271 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33272 Jakub Jelinek <jakub@redhat.com>
33275 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33277 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33278 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33279 the alignment computation, but return their maximum in warn_alignment.
33280 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33281 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33283 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33284 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33287 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33289 * config/arc/simdext.md (dmpyh): Fix typo.
33291 2017-04-25 Richard Biener <rguenther@suse.de>
33293 PR tree-optimization/80492
33294 * alias.c (compare_base_decls): Handle registers with asm
33295 specification conservatively.
33296 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33297 compare_base_decls returning dont-know properly.
33299 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33301 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33302 (legitimate_offset_address_p): New function.
33303 (arc_legitimate_address_p): Use above function.
33305 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33307 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33309 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33311 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33312 ACCH registers whenever they are available.
33314 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33316 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33317 double regs fix when not used.
33319 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33321 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33323 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33324 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33326 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33328 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33329 when emitting short ADD instructions.
33331 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33333 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33335 (cmpsi_cc_c_insn): Likewise.
33336 (cbranchsi4_scratch): Compute proper instruction length using
33337 compact_hreg_operand.
33338 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33340 2017-04-25 Richard Biener <rguenther@suse.de>
33342 PR middle-end/80509
33343 * passes.c (pass_manager::pass_manager): Initialize
33344 m_name_to_pass_map.
33346 2017-04-25 Richard Biener <rguenther@suse.de>
33348 PR tree-optimization/79201
33349 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33351 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33354 * config/s390/vector.md: Split MEM->GPR vector moves for
33355 non-s_operand addresses.
33357 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33360 * config/s390/predicates.md (reload_const_wide_int_operand): New
33362 * config/s390/s390.md ("movti"): Remove d/P alternative.
33363 ("movti_bigconst"): New pattern definition.
33365 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33368 * s390-protos.h (s390_expand_cs_hqi): Removed.
33369 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33370 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33371 modes as well as CCZ1mode and CCZmode.
33372 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33373 signature of s390_emit_compare_and_swap.
33374 (s390_expand_cs_hqi): Likewise, make static.
33375 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33376 compare-and-swap, in some cases.
33377 (s390_expand_cs): Wrapper function.
33378 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33380 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33381 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33382 patterns for small and large integers. Forbid symref memory operands.
33383 Move expander to s390.c. Require cc register.
33384 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33385 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33386 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33387 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33388 symref memory operands. Remove CC mode and call s390_match_ccmode
33390 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33392 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33394 * config/s390/s390.md (define_peephole2): New peephole to help
33395 combining the load-and-test pattern with volatile memory.
33397 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33399 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33400 with CCZmode for TARGET_Z196.
33402 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33404 PR rtl-optimization/80501
33405 * combine.c (make_compound_operation_int): Set subreg_code to SET
33406 even for AND with mask of the sign bit of mode.
33408 PR rtl-optimization/80500
33409 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33410 sum's initial value.
33412 2017-04-25 Julian Brown <julian@codesourcery.com>
33413 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33415 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33417 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33419 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33421 2017-04-25 Julian Brown <julian@codesourcery.com>
33422 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33424 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33425 (thunderx2t99_sha): New Reservation.
33427 2017-04-25 Julian Brown <julian@codesourcery.com>
33428 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33430 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33431 type for 1-element load.
33433 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33435 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33437 2017-04-24 Martin Jambor <mjambor@suse.cz>
33439 PR tree-optimization/80293
33440 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33441 char arrays not totally scalarizable if it is false.
33442 (analyze_all_variable_accesses): Pass correct value in the new
33443 parameter. Add a statistics counter.
33445 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33447 PR middle-end/79931
33448 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33450 2017-04-24 Richard Biener <rguenther@suse.de>
33452 PR tree-optimization/80494
33453 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33454 out for complex types.
33456 2017-04-24 Richard Biener <rguenther@suse.de>
33458 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33459 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33460 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33461 (DFS): Adjust and never fail.
33462 (sccvn_dom_walker::fail): Remove.
33463 (sccvn_dom_walker::before_dom_children): Adjust.
33464 (run_scc_vn): Likewise and never fail.
33465 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33466 (pass_fre::execute): Likewise.
33468 2017-04-24 Richard Biener <rguenther@suse.de>
33470 PR tree-optimization/79725
33471 * tree-ssa-sink.c (statement_sink_location): Return whether
33472 failure reason was zero uses. Move that check later.
33473 (sink_code_in_bb): Deal with zero uses by removing the stmt
33476 2017-04-24 Richard Biener <rguenther@suse.de>
33479 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33480 pointer-based references.
33482 2017-04-24 Richard Biener <rguenther@suse.de>
33485 * pass_manager.h (pass_manager::operator new): Remove.
33486 (pass_manager::operator delete): Likewise.
33487 * passes.c (pass_manager::operator new): Remove.
33488 (pass_manager::operator delete): Likewise.
33489 (pass_manager::pass_manager): Zero individual pass members.
33491 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33494 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33495 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33496 Check "XEXP (src, 1)" operand here.
33497 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33498 Check "XEXP (src, 1)" operand here.
33499 (dimode_scalar_chain::make_vector_copies): Detect count register
33500 of a shift instruction. Zero extend count register from QImode
33501 to DImode to satisfy vector shift pattern count operand predicate.
33502 Substitute vector shift count operand with a DImode copy.
33503 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33506 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33508 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33509 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33510 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33511 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33512 (UNSPEC_NOREX_MEM): Remove definition.
33514 2017-04-21 Richard Biener <rguenther@suse.de>
33516 PR tree-optimization/79547
33517 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33518 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33519 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33520 without any constraints.
33522 2017-04-21 Richard Biener <rguenther@suse.de>
33524 PR tree-optimization/78847
33525 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33527 2017-04-21 Richard Biener <rguenther@suse.de>
33529 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33530 (build_distinct_type_copy): Likewise.
33531 (build_variant_type_copy): Likewise.
33532 * tree.c (build_qualified_type): Pass down mem-stat info.
33533 (build_distinct_type_copy): Likewise.
33534 (build_variant_type_copy): Likewise.
33536 2017-04-21 Richard Biener <rguenther@suse.de>
33538 PR tree-optimization/80237
33539 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33541 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33542 for a simplified result.
33544 2016-04-21 Richard Biener <rguenther@suse.de>
33546 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33547 sth as strict as a simple_iv but a chrec without symbols and an
33548 operand defined in the loop we are peeling (and not some subloop).
33549 (propagate_constants_for_unrolling): Propagate all constants.
33551 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33554 * config/i386/i386.c (print_reg): Remove assert for disalowed
33555 regno values, call output_operand_lossage instead.
33557 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33560 * config/i386/constraints.md (Yc): New register constraint.
33561 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33562 Use Yc constraint for alternative 2 of operand 0. Remove
33563 preferred_for_speed attribute.
33565 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33567 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33568 lastprivate clauses in SIMT case.
33570 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33572 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33574 2017-04-20 Richard Biener <rguenther@suse.de>
33576 PR tree-optimization/57796
33577 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33578 as N scalar stores.
33579 (vect_model_load_cost): Cost gathers as N scalar loads.
33581 2017-04-20 Richard Biener <rguenther@suse.de>
33583 * ggc-page.c (ggc_allocated_p): Rename to ...
33584 (safe_lookup_page_table_entry): ... this and return the lookup
33586 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33588 2017-04-20 Richard Biener <rguenther@suse.de>
33590 PR tree-optimization/80453
33591 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33592 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33593 from the conditions.
33594 (vn_phi_eq): Pass them down.
33595 (vn_phi_lookup): Record them.
33596 (vn_phi_insert): Likewise.
33598 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33600 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33601 uninitialized variable warning to avoid buffer overrun.
33603 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33606 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33607 is suppressed for '{ 0 }' in C.
33609 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33611 * BASE-VER: Set to 8.0.0.
33613 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33615 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33616 priority .init_array and .fini_array section with SECTION_NOTYPE
33619 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33621 PR middle-end/80423
33622 * tree.h (build_array_type): Add typeless_storage default argument.
33623 * tree.c (type_cache_hasher::equal): Also compare
33624 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33625 (build_array_type): Add typeless_storage argument, set
33626 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33628 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33629 (build_array_type): Likewise. Add typeless_storage argument.
33631 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33632 Jakub Jelinek <jakub@redhat.com>
33634 PR tree-optimization/80426
33635 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33636 operation on symbolic operands, also compute the overflow for the
33637 invariant part when the operation degenerates into a negation.
33639 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33642 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33643 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33646 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33648 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33651 * config/avr/avr.c (tree.h): Include it.
33652 (cgraph.h): Include it.
33653 (avr_encode_section_info): Don't warn for uninitialized progmem
33654 variable if it's just an alias.
33656 2017-04-19 Richard Biener <rguenther@suse.de>
33659 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33660 when needed by AutoPGO.
33662 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33665 * doc/lto.texi: Remove an extra 'that'.
33667 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33669 PR rtl-optimization/80429
33670 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33671 are only used in debug insns.
33673 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33674 Vladimir Makarov <vmakarov@redhat.com>
33676 * config/sparc/predicates.md (input_operand): Add comment. Return
33677 true for any memory operand when LRA is in progress.
33678 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33680 2017-04-18 Jeff Law <law@redhat.com>
33683 * mips.md ({return,simple_return}_internal): Do not overwrite
33686 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33688 PR tree-optimization/80443
33689 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33690 instead of adding 1, subtract -1 and similarly instead of subtracting
33693 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33695 PR rtl-optimization/80357
33696 * haifa-sched.c (tmp_bitmap): New variable.
33697 (model_recompute): Handle duplicate use records.
33698 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33699 (free_global_sched_pressure_data): Free it.
33701 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33704 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33705 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33706 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33707 instead of SYSTEM_HEADER_DIR.
33709 2017-04-18 Jeff Law <law@redhat.com>
33711 PR middle-end/80422
33712 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33713 predecessors after walking up the insn chain.
33715 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33718 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33719 sizetype type into debug info.
33721 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33724 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33725 unneeded test for TARGET_UPPER_REGS_SF.
33726 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33728 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33731 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33732 instead of gsi_after_labels.
33734 2017-04-18 Jeff Law <law@redhat.com>
33736 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33740 2017-04-13 Jeff Law <law@redhat.com>
33741 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33742 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33744 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33747 * config/avr/avr.c (intl.h): Include it.
33748 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33750 2017-04-18 Martin Liska <mliska@suse.cz>
33752 PR gcov-profile/78783
33753 * gcov-tool.c (gcov_output_files): Validate that destination
33754 file is either removed by the tool or by a user.
33756 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33757 Guy Benyei <guybe@mellanox.com>
33759 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33760 block, and do not negate it, the stored id is already negative.
33762 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33764 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33766 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33769 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33770 masks of options that should be turned off if the VSX vector
33771 options are turned off.
33772 (OTHER_P8_VECTOR_MASKS): Likewise.
33773 (OTHER_VSX_VECTOR_MASKS): Likewise.
33774 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33775 rs6000_disable_incompatible_switches to validate no type switches
33777 (rs6000_incompatible_switch): New function to disallow turning on
33778 other vector options if -mno-vsx, -mno-power8-vector, or
33779 -mno-power9-vector are specified.
33781 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33783 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33785 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33787 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33788 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33789 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33790 (ARG_POINTER_CFA_OFFSET): Likewise.
33792 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33794 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33795 conditions to take advantage of various optimizations.
33797 2017-04-13 Jeff Law <law@redhat.com>
33799 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33800 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33801 (zero_extendsidi2_dext): Likewise.
33803 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33806 * fold-const.c (fold_ternary_loc): Revert
33807 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33810 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33812 PR rtl-optimization/80343
33813 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33814 new scratch pseudo.
33816 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33819 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33820 to ubsan_encode_value.
33822 2017-04-13 Jeff Law <law@redhat.com>
33824 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33825 appearing in DEBUG_INSNs.
33827 2017-04-13 Martin Liska <mliska@suse.cz>
33829 PR gcov-profile/80413
33830 * gcov-io.c (gcov_write_string): Copy to buffer just when
33831 allocated size is greater than zero.
33833 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33836 * dwarf2out.c (decls_for_scope): Ignore declarations of
33837 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33839 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33842 * ipa-visibility.c (non_local_p): Fix typos.
33843 (localize_node): When localizing symbol in same comdat group,
33844 dissolve the group only when we know external symbols are going
33846 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33848 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33850 PR tree-optimization/79390
33851 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33852 order does not result in usable sequence, retry with reversed operand
33858 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33859 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33860 op0 instead of fold_convert_loc (loc, type, arg0).
33862 2017-04-12 Jeff Law <law@redhat.com>
33864 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33865 has a delay slot in the generated code.
33867 * config/cris/cris.md (cris_preferred_reload_class): Return
33868 GENNONACR_REGS rather than GENERAL_REGS.
33870 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33873 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33874 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33875 signedness of the result type.
33877 2017-04-12 Richard Biener <rguenther@suse.de>
33878 Jeff Law <law@redhat.com>
33880 PR tree-optimization/80359
33881 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33882 trim stores to TARGET_MEM_REFs.
33884 2017-04-12 Richard Biener <rguenther@suse.de>
33886 PR tree-optimization/79390
33887 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33888 threading case even more.
33890 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33893 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33894 for quad_address_p for TImode, instead of just not indexed_address.
33896 2017-04-12 Richard Biener <rguenther@suse.de>
33897 Bernd Edlinger <bernd.edlinger@hotmail.de>
33899 PR middle-end/79671
33900 * alias.c (component_uses_parent_alias_set_from): Handle
33901 TYPE_TYPELESS_STORAGE.
33902 (get_alias_set): Likewise.
33903 * tree-core.h (tree_type_common): Add typeless_storage flag.
33904 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33905 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33906 for types containing members with TYPE_TYPELESS_STORAGE.
33907 (place_field): Likewise.
33908 (layout_type): Likewise for ARRAY_TYPE.
33909 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33910 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33911 TYPE_TYPELESS_STORAGE.
33912 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33914 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33917 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33918 first argument to type.
33920 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33924 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33925 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33926 (rs6000_expand_binop_builtin): Likewise.
33927 (rs6000_expand_ternop_builtin): Likewise; also add missing
33928 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33929 vshasigma built-ins.
33930 * doc/extend.texi: Document that vec_xxpermdi's third argument
33931 must be a constant.
33933 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33935 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33936 Use shift_const cost parameter when calculating gain of STV shifts.
33938 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33940 PR rtl-optimization/70478
33941 * lra-constraints.c (process_alt_operands): Check memory for
33942 disfavoring memory insn operand.
33944 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33946 PR middle-end/80100
33947 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33948 left shift in unsigned HOST_WIDE_INT type.
33950 PR rtl-optimization/80385
33951 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33952 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33955 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33956 if they have any depend clauses.
33958 2017-04-11 Martin Liska <mliska@suse.cz>
33961 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33962 * ipa-split.c (split_function): Create a local comdat symbol
33963 if caller is in a comdat group.
33965 2017-04-11 Martin Liska <mliska@suse.cz>
33968 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33971 2017-04-11 Martin Sebor <msebor@redhat.com>
33973 PR middle-end/80364
33974 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33975 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33977 (directive::set_width, directive::set_precision, format_character):
33979 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33982 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33985 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33986 conflict, set target->arch_name instead of target->cpu_name.
33988 2017-04-11 Richard Biener <rguenther@suse.de>
33990 PR tree-optimization/80374
33991 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33992 build_zero_cst, remove fold_convertible_p check again.
33994 2017-04-11 Martin Liska <mliska@suse.cz>
33997 * ubsan.c (instrument_object_size): Do not instrument register
34000 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34003 * config/i386/i386-builtin-types.def
34004 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34005 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34006 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34007 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34008 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34009 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34010 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34011 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34012 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34013 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34014 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34015 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34016 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34017 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34018 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34019 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34020 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34021 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34022 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34023 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34024 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34025 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34026 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34027 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34028 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34029 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34030 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34031 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34032 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34033 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34034 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34035 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34036 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34037 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34038 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34039 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34040 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34041 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34042 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34043 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34044 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34045 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34046 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34047 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34048 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34049 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34051 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34052 flag to second_arg_count, handle 4 argument function type _COUNT
34053 aliases, handle second_arg_count on second argument rather than last.
34055 2017-04-10 Jeff Law <law@redhat.com>
34057 PR tree-optimization/80374
34058 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34059 record anything if we can not convert integer_zero_node to the
34062 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34065 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34066 Enhance special handling given to the TARGET_P9_MINMAX option in
34067 relation to certain other options.
34069 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34071 PR tree-optimization/80153
34072 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34073 remove POINTER_PLUS_EXPR's base part directly, rather than through
34076 2017-04-10 Richard Biener <rguenther@suse.de>
34077 Bin Cheng <bin.cheng@arm.com>
34079 PR tree-optimization/80153
34080 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34081 the first element of pointer type aff_tree. Build result expr in
34083 (add_elt_to_tree): Convert to type unconditionally. Remove other
34084 fold_convert calls.
34085 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34086 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34088 2017-04-10 Richard Biener <rguenther@suse.de>
34090 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34093 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34095 PR rtl-optimization/70478
34096 * lra-constraints.c (curr_small_class_check): New.
34097 (update_and_check_small_class_inputs): New.
34098 (process_alt_operands): Update curr_small_class_check. Disfavor
34099 alternative insn memory operands. Check available regs for small
34102 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34105 * config/mips/mips.opt (-mvirt): Update description.
34106 * doc/invoke.texi (-mvirt): Likewise.
34108 2017-04-10 Richard Biener <rguenther@suse.de>
34110 PR middle-end/80362
34111 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34112 looking for NEGATE_EXPR in -A / -B to A / B folding.
34114 2017-04-10 Martin Liska <mliska@suse.cz>
34116 PR gcov-profile/80224
34117 * gcov.c (print_usage): Fix usage string.
34118 (get_gcov_intermediate_filename): Remove.
34119 (output_gcov_file): Use both for normal and intermediate format.
34120 (generate_results): Do not initialize special file for
34121 intermediate format.
34123 2017-04-10 Richard Biener <rguenther@suse.de>
34125 PR tree-optimization/80304
34126 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34129 2017-04-10 Nathan Sidwell <nathan@acm.org>
34132 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34133 (rs6000_init_builtins): Use it.
34135 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34137 * config/arm/arm.md (<mrc>): Add mode to SET source.
34138 (<mrrc>): Likewise.
34140 2017-04-10 Richard Biener <rguenther@suse.de>
34142 PR middle-end/80344
34143 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34145 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34148 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34149 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34150 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34151 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34152 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34153 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34154 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34155 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34156 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34157 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34158 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34159 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34160 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34161 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34162 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34163 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34164 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34165 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34166 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34167 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34168 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34169 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34170 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34172 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34174 PR rtl-optimization/70478
34175 * lra-constraints.c: Reverse the last patch.
34177 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34179 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34180 Add comment for WCHAR_T.
34182 2017-04-08 Martin Liska <mliska@suse.cz>
34185 2017-04-07 Martin Liska <mliska@suse.cz>
34188 * ipa-split.c (split_function): Add function part to a same comdat
34191 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34194 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34196 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34198 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34199 (vec_pairing): ...to this.
34200 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34201 (rs6000_sched_init): Adjust for name change.
34202 (struct rs6000_sched_context): Likewise.
34203 (rs6000_init_sched_context): Likewise.
34204 (rs6000_set_sched_context): Likewise.
34206 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34212 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34214 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34215 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34216 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34218 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34220 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34222 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34224 PR rtl-optimization/70703
34225 * ira-color.c (update_conflict_hard_regno_costs): Use
34226 int64_t instead of HOST_WIDE_INT.
34228 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34230 PR rtl-optimization/70478
34231 * lra-constraints.c (process_alt_operands): Disfavor alternative
34232 insn memory operands.
34234 2017-04-07 Jeff Law <law@redhat.com>
34236 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34237 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34239 2017-04-07 Martin Liska <mliska@suse.cz>
34242 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34243 Show error message instead of an ICE.
34245 2017-04-07 Martin Liska <mliska@suse.cz>
34248 * ipa-split.c (split_function): Add function part to a same comdat
34251 2017-04-07 Richard Biener <rguenther@suse.de>
34253 PR middle-end/80341
34254 * tree.c (get_unwidened): Also handle ! for_type case for
34256 * convert.c (do_narrow): Split out from ...
34257 (convert_to_integer_1): ... here. Do not pass final truncation
34258 type to get_unwidened for TRUNC_DIV_EXPR.
34260 2017-04-07 Richard Biener <rguenther@suse.de>
34262 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34264 (aff_combination_const): Adjust.
34265 (aff_combination_scale): Likewise.
34266 (aff_combination_add_elt): Likewise.
34267 (aff_combination_add_cst): Likewise.
34268 (aff_combination_convert): Likewise.
34269 (add_elt_to_tree): Likewise. Remove unused argument.
34270 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34272 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34274 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34276 * config/arm/arm.c (arm_default_short_enums): Use
34277 ARM_DEFAULT_SHORT_ENUMS.
34278 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34280 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34283 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34284 members with redundant out-of-class redeclaration.
34286 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34289 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34290 * config/i386/i386.md (*zero_extendsidi2):
34291 Add (?*x,*x) and (?*v,*v) alternatives.
34293 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34296 * config/i386/i386.c (ix86_expand_builtin)
34297 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34298 mode from insn data. Convert operands to insn operand mode.
34299 Copy operands that don't satisfy insn predicate to a register.
34301 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34303 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34306 2017-04-06 Richard Biener <rguenther@suse.de>
34308 PR tree-optimization/80334
34309 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34310 preserve alignment of accesses.
34312 2017-04-06 Richard Biener <rguenther@suse.de>
34314 PR tree-optimization/80262
34315 * tree-sra.c (build_ref_for_offset): Preserve address-space
34317 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34318 Drop useless address-space information on MEM_REF offsets.
34320 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34322 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34324 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34326 PR rtl-optimization/70703
34327 * ira-color.c (update_conflict_hard_regno_costs): Use
34328 HOST_WIDE_INT instead of long.
34330 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34333 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34334 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34336 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34337 for x86_64 target. Handle -m3dnowa option.
34339 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34341 PR rtl-optimization/70703
34342 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34343 (update_conflict_hard_regno_costs): Use long instead of unsigned
34344 arithmetic for cost calculation.
34346 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34347 Bernd Edlinger <bernd.edlinger@hotmail.de>
34350 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34353 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34356 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34357 ptr_mode with Pmode throughout.
34358 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34359 into probe_stack_range and use DImode.
34361 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34364 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34365 call_eh_return is true.
34367 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34369 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34370 Initialize last_match_fntype_index.
34372 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34375 * tree-nvr.c: Include internal-fn.h.
34376 (pass_return_slot::execute): Ignore internal calls without
34379 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34380 Richard Biener <rguenther@suse.de>
34383 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34384 captures used multiple times, except for the last use.
34385 * generic-match-head.c: Include gimplify.h.
34387 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34389 PR tree-optimization/79390
34390 * target.h (struct noce_if_info): Declare.
34391 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34392 * target.def (noce_conversion_profitable_p): New target hook.
34393 * ifcvt.h (struct noce_if_info): New type, moved from ...
34394 * ifcvt.c (struct noce_if_info): ... here.
34395 (noce_conversion_profitable_p): Renamed to ...
34396 (default_noce_conversion_profitable_p): ... this. No longer
34398 (noce_try_store_flag_constants, noce_try_addcc,
34399 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34400 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34401 instead of noce_conversion_profitable_p.
34402 * config/i386/i386.c: Include ifcvt.h.
34403 (ix86_option_override_internal): Don't override
34404 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34405 (ix86_noce_conversion_profitable_p): New function.
34406 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34407 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34408 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34409 * doc/tm.texi: Regenerated.
34411 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34413 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34416 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34419 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34420 instructions for small multiply cores.
34422 2017-04-04 Jeff Law <law@redhat.com>
34424 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34426 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34427 that are not set by the loop over the elements.
34429 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34432 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34433 int mode, convert_modes it to mode as unsigned, otherwise use
34434 lowpart_subreg to mode rather than SImode.
34435 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34436 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34437 Use DImode instead of SImode for the shift count operand.
34438 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34441 2017-04-04 Richard Biener <rguenther@suse.de>
34443 PR middle-end/80281
34444 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34445 arithmetic done for the negate or the plus. Simplify.
34446 (A - (-B) -> A + B): Likewise.
34447 * fold-const.c (split_tree): Make sure to not negate pointers.
34449 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34451 PR rtl-optimization/60818
34452 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34453 a compare of comparisons with the thing compared if this results
34454 in a different machine mode.
34456 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34458 * alias.c (base_alias_check): Fix typo in comment.
34459 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34460 * cgraphunit.c (symbol_table::compile): Likewise.
34461 * collect2.c (maybe_run_lto_and_relink): Likewise.
34462 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34463 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34464 * config/avr/avr.c (avr_map_op_t): Likewise.
34465 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34466 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34467 * config/epiphany/epiphany.md (movcc): Likewise.
34468 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34469 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34471 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34472 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34473 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34474 * config/sh/sh.c (sh_rtx_costs): Likewise.
34475 * fold-const.c (fold_truth_andor): Likewise.
34476 * genautomata.c (collapse_flag): Likewise.
34477 * gengtype.h (struct type::u::s): Likewise.
34478 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34479 * input.c (FORMAT_AMOUNT): Likewise.
34480 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34481 (known_aggs_to_agg_replacement_list): Likewise.
34482 * ipa-inline-analysis.c: Likewise.
34483 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34484 * ipa-polymorphic-call.c
34485 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34486 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34487 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34489 * modulo-sched.c (apply_reg_moves): Likewise.
34490 * omp-expand.c (build_omp_regions_1): Likewise.
34491 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34492 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34493 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34494 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34495 * value-prof.c: Likewise.
34496 * var-tracking.c (val_reset): Likewise.
34498 2017-04-03 Richard Biener <rguenther@suse.de>
34500 PR tree-optimization/80275
34501 * fold-const.c (split_address_to_core_and_offset): Handle
34504 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34506 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34507 descriptors is at least equal to that of functions.
34509 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34511 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34513 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34516 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34517 (mov<IMOD4:mode>): New expander.
34518 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34520 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34522 PR rtl-optimization/79405
34523 * fwprop.c (propagations_left): New variable.
34524 (forward_propagate_into): Decrement it.
34525 (fwprop_init): Initialize it.
34526 (fw_prop): If the variable has reached zero, stop propagating.
34527 (fwprop_addr): Ditto.
34529 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34532 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34533 a FUNCTION_DECL, pass it as decl instead of origin to
34536 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34538 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34541 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34544 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34545 TARGET_VSX_SMALL_INTEGER.
34547 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34549 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34550 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34552 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34554 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34555 extraction from odd-numbered MSA register.
34557 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34559 PR middle-end/80173
34560 * expmed.c (store_bit_field_1): Don't attempt to create
34561 a word subreg out of hard registers wider than word if they
34562 have HARD_REGNO_NREGS of 1 for their mode.
34564 PR middle-end/80163
34565 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34566 conversions to integer types wider than word and pointer.
34569 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34570 (rtx_equal_for_cselib_p): Pass 0 to it.
34571 * cselib.c (cselib_hasher::equal): Likewise.
34572 (rtx_equal_for_cselib_1): Add depth argument. If depth
34573 is 128, don't look up VALUE locs and punt. Increment
34574 depth in recursive calls when walking VALUE locs.
34576 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34578 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34579 (make_gcov_file_name): Use the canonical path name for generating
34581 (read_line): Fix handling of files with ascii null bytes.
34583 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34585 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34586 to initialise a vector register instead
34587 of using a const_int.
34589 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34591 PR translation/80189
34592 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34593 diagnostic messages.
34595 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34598 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34599 (dfp_diex_<mode>): Update mode of operand 1.
34600 * doc/extend.texi (dxex, dxexq): Document change to return type.
34601 (diex, diexq): Document change to argument type.
34603 2017-03-30 Martin Jambor <mjambor@suse.cz>
34606 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34607 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34608 it reflects the signature changes performed at the callee side.
34609 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34610 to cgraph_build_function_type_skip_args.
34611 (build_function_decl_skip_args): Adjust call to the above function.
34613 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34616 * config/i386/sse.md
34617 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34618 register as dest whenever it is a MEM not rtx_equal_p to the
34619 corresponding dup operand, and when forcing into reg move the
34620 reg into the memory afterwards.
34621 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34622 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34623 for the force_reg mode.
34624 (avx512vl_vextractf128<mode>): Use register as dest either
34625 always when a MEM, or when it is a MEM not rtx_equal_p to the
34626 corresponding dup operand, or even not when it is a CONST_VECTOR
34627 depending on the mode and lo vs. hi.
34628 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34630 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34631 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34632 Likewise. Require that operands[2] is even.
34633 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34634 Remove extraneous parens. Require that operands[2] is a multiple
34636 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34637 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34638 disallow memory then.
34640 2017-03-30 Richard Biener <rguenther@suse.de>
34642 PR tree-optimization/77498
34643 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34644 to non-constants over backedges.
34646 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34648 PR rtl-optimization/80233
34649 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34650 as last_combined_insn. Do not test for BARRIER_P separately.
34652 2017-03-29 Andreas Schwab <schwab@suse.de>
34655 * calls.c (prepare_call_address): Convert funexp to Pmode before
34656 copying to temp reg.
34658 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34660 PR tree-optimization/80158
34661 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34662 Handle possible future case of more than one alternate
34664 (replace_rhs_if_not_dup): Likewise.
34665 (replace_one_candidate): Likewise.
34667 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34669 PR rtl-optimization/80193
34670 * ira.c (ira): Do not check allocation for LRA.
34672 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34674 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34675 (nvptx_output_simt_exit): Declare.
34676 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34677 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34678 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34679 (nvptx_declare_function_name): ...here. Emit declaration of local
34680 memory space buffer for omp_simt_enter insn.
34681 (nvptx_output_unisimt_switch): New.
34682 (nvptx_output_softstack_switch): New.
34683 (nvptx_output_simt_enter): New.
34684 (nvptx_output_simt_exit): New.
34685 * config/nvptx/nvptx.h (struct machine_function): New fields
34686 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34687 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34688 (UNSPECV_SIMT_EXIT): Ditto.
34689 (omp_simt_enter_insn): New insn.
34690 (omp_simt_enter): New expansion.
34691 (omp_simt_exit): New insn.
34692 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34694 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34695 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34696 (expand_GOMP_SIMT_EXIT): New.
34697 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34698 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34699 (GOMP_SIMT_EXIT): Ditto.
34700 * target-insns.def (omp_simt_enter): New insn.
34701 (omp_simt_exit): Ditto.
34702 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34704 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34705 (lower_rec_input_clauses): Likewise.
34706 (lower_lastprivate_clauses): Handle SIMT privatization.
34708 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34709 (ompdevlow_adjust_simt_enter): New.
34710 (find_simtpriv_var_op): New.
34711 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34712 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34714 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34715 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34716 (copy_decl_for_dup_finish): Ditto.
34718 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34720 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34723 * config/i386/i386.c (ix86_option_override_internal): Always
34724 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34726 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34728 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34730 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34732 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34733 mark new edge's irreducible flag accordign to it.
34734 (vect_do_peeling): Check loop preheader edge's irreducible flag
34735 and pass it to function slpeel_add_loop_guard.
34737 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34739 PR tree-optimization/80218
34740 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34741 Update block frequencies and counts.
34743 2017-03-28 Richard Biener <rguenther@suse.de>
34745 PR tree-optimization/78644
34746 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34747 of a simplification result we may not use it at all.
34749 2017-03-28 Richard Biener <rguenther@suse.de>
34752 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34753 without arguments, generate default definition of a SSA name.
34755 2017-03-28 Richard Biener <rguenther@suse.de>
34757 PR middle-end/80222
34758 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34759 TYPE_REF_CAN_ALIAS_ALL references.
34760 * fold-const.c (fold_indirect_ref_1): Likewise.
34762 2017-03-28 Martin Liska <mliska@suse.cz>
34765 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34766 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34768 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34769 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34771 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34772 (EXTRA_SPECS): Define.
34773 (SUBTARGET_EXTRA_SPECS): Likewise.
34774 (SUBTARGET_CPP_SPEC): Likewise.
34775 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34776 SUBTARGET_EXTRA_SPECS.
34777 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34779 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34781 * config/arc/simdext.md (vst64_insn): Update pattern.
34782 (vld32wh_insn): Likewise.
34783 (vld32wl_insn): Likewise.
34784 (vld64_insn): Likewise.
34785 (vld32_insn): Likewise.
34787 2017-03-28 Marek Polacek <polacek@redhat.com>
34790 * fold-const.c (fold_comparison): Use protected_set_expr_location
34791 instead of SET_EXPR_LOCATION.
34793 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34795 * tree.c (add_expr): Avoid name lookup warning.
34797 2017-03-27 Jeff Law <law@redhat.com>
34799 PR tree-optimization/80216
34800 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34801 function name. Limit recursion depth.
34802 (record_temporary_equivalences): Corresponding changes.
34804 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34806 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34809 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34812 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34814 * cfgcleanup.c (reg_note_cfa_p): New array.
34815 (insns_have_identical_cfa_notes): New function.
34816 (old_insns_match_p): Don't cross-jump in between /f
34817 and non-/f instructions. If both i1 and i2 are frame related,
34818 verify all CFA notes, their order and content.
34820 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34823 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34824 HImode and SImode with zero extend to DImode to one insn.
34825 (bswap<mode>2_extenddi): Likewise.
34826 (bswapsi2_extenddi): Likewise.
34827 (bswaphi2_extendsi): Likewise.
34828 (bswaphi2): Combine bswap HImode and SImode into one insn.
34829 Separate memory insns from swapping register.
34830 (bswapsi2): Likewise.
34831 (bswap<mode>2): Likewise.
34832 (bswaphi2_internal): Delete, no longer used.
34833 (bswapsi2_internal): Likewise.
34834 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34835 store, and gpr<-gpr swap insns.
34836 (bswap<mode>2_store): Likewise.
34837 (bswaphi2_reg): Register only splitter, combine with the splitter.
34838 (bswaphi2 splitter): Likewise.
34839 (bswapsi2_reg): Likewise.
34840 (bswapsi2 splitter): Likewise.
34841 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34842 the insns into load, store, and register/register insns.
34843 (bswapdi2_ldbrx): Likewise.
34844 (bswapdi2_load): Likewise.
34845 (bswapdi2_store): Likewise.
34846 (bswapdi2_reg): Likewise.
34848 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34850 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34851 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34853 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34856 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34858 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34859 special handling for target option conflicts between dform
34860 options (-mpower9-dform, -mpower9-dform-vector,
34861 -mpower9-dform-scalar) and -mno-direct-move.
34863 2017-03-27 Richard Biener <rguenther@suse.de>
34865 PR tree-optimization/80181
34866 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34868 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34870 * config/arc/predicates.md (move_double_src_operand): Replace the
34871 call to move_double_src_operand with a call to address_operand.
34873 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34875 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34876 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34877 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34879 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34881 * config/arc/predicates.md (long_immediate_loadstore_operand):
34882 Consider scaled addresses cases.
34884 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34886 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34887 restored when in interrupt.
34888 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34889 doesn't have delay slot.
34891 2017-03-27 Richard Biener <rguenther@suse.de>
34894 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34895 inlined thunk clones.
34897 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34900 * asan.c (instrument_derefs): Copy over last operand from
34901 original COMPONENT_REF to the new COMPONENT_REF with
34902 DECL_BIT_FIELD_REPRESENTATIVE.
34903 * ubsan.c (instrument_object_size): Likewise.
34905 2017-03-27 Richard Biener <rguenther@suse.de>
34907 PR tree-optimization/80170
34908 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34909 sure DR/SCEV didnt fold in constants we do not see when looking
34910 at the reference base alignment.
34912 2017-03-27 Richard Biener <rguenther@suse.de>
34914 PR middle-end/80171
34915 * gimple-fold.c (fold_ctor_reference): Properly guard against
34916 NULL return value from canonicalize_constructor_val.
34918 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34921 * config/i386/i386.c (ix86_expand_builtin)
34922 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34923 flags reg setting and flags reg using instructions.
34924 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34925 clobbering instructions to zero extend op2.
34927 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34929 * doc/install.texi (Configuration) <--with-aix-soname>:
34930 Update link to AIX ld.
34932 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34934 PR rtl-optimization/80160
34935 PR rtl-optimization/80159
34936 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34937 reg_alternate_class into account.
34939 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34942 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34943 to consider in curr_insn_transform.
34945 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34947 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34948 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34949 and emit_mode_inner.
34951 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34953 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34954 argument to the overloaded builtin variants. Use the new flag to
34955 deprecate certain builtin variants.
34956 * config/s390/s390-builtin-types.def: Add new builtin types.
34957 * config/s390/s390-builtins.h: Support new flags field for
34958 overloaded builtins.
34959 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34960 (s390_macro_to_expand): Enable vector float data type.
34961 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34962 builtins by incrementing the __VEC__ version number.
34963 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34965 (s390_resolve_overloaded_builtin): Emit error messages depending
34966 on the builtin flags.
34967 * config/s390/s390.c (s390_expand_builtin): Support additional
34968 flags argument. Change error message to match the messages
34969 emitted in s390-c.c.
34970 * config/s390/s390.md: New UNSPEC_* constants.
34971 (op_type): Add new instruction types.
34972 * config/s390/vecintrin.h: Add new builtins and test data class
34974 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34975 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34976 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34977 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34978 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34979 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34981 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34982 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34983 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34984 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34986 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34987 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34988 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34989 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34991 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34992 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34993 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34994 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34995 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34996 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34997 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34999 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35000 ("vec_scatter_element<V_HW_4:mode>_DI")
35001 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35002 ("vec_fpint<mode>", "vflls")
35003 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35004 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35005 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35006 ("*vec_cmphe<mode>_cc"): ... these.
35008 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35009 mode constant instead of magic value.
35011 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35013 * config/s390/s390.c (s390_expand_vec_compare): Support other
35014 vector floating point modes than just V2DF.
35015 (s390_expand_vcond): Likewise.
35016 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35017 (s390_cannot_change_mode_class): Prevent mode changes between TF
35018 and V1TF in vector registers.
35019 * config/s390/s390.md (DF, SF): New mode attributes.
35020 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35021 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35022 SFmode support for VRs.
35023 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35025 (VFT, VF_HW): New mode iterators.
35026 (vw, sdx): New mode attributes.
35027 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35028 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35029 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35030 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35031 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35032 also the new vector floating point modes. Renaming to ...
35034 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35035 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35036 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35037 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35038 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35039 ("vec_unordered<mode>"): ... these.
35041 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35042 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35043 ("*vec_extendv2df"): New insn definitions.
35045 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35047 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35048 ("mulditi3_2", "*muldi3_sign"): New patterns.
35049 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35050 rename the pattern definition.
35052 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35054 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35056 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35058 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35060 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35061 instruction if possible.
35062 * config/s390/vector.md (vec_halfnumelts): New mode
35064 ("*vec_vllezlf<mode>"): New pattern.
35066 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35068 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35069 ("popcountv4si2", "popcountv2di2"): Rename to ...
35070 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35071 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35073 ("popcount<mode>2_vxe"): New pattern.
35075 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35077 * common/config/s390/s390-common.c (processor_flags_table): Add
35079 * config.gcc: Add arch12.
35080 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35081 Default to arch12 for unknown CPU model numbers.
35082 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35083 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35084 PROCESSOR_max sanity check.
35085 * config/s390/s390-opts.h (enum processor_type): Add
35087 * config/s390/s390.c (processor_table): Add arch12.
35088 (s390_expand_builtin): Add check for B_VXE flag.
35089 (s390_issue_rate): Add PROCESSOR_ARCH12.
35090 (s390_get_sched_attrmask): Likewise.
35091 (s390_get_unit_mask): Likewise.
35092 (s390_sched_score): Enable z13 scheduling for arch12.
35093 (s390_sched_reorder): Likewise.
35094 (s390_sched_variable_issue): Likewise.
35095 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35097 (s390_tune_attr): Use z13 scheduling also for arch12.
35098 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35099 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35100 (TARGET_VXE_P): New macros.
35101 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35102 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35103 * config/s390/s390.opt: Add arch12 as processor_type.
35105 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35107 * config/s390/s390.md
35108 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35109 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35110 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35112 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35113 Rename expanders to ...
35115 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35116 ("fixuns_truncdddi2_emu"): ... these.
35118 ("fixuns_trunc<mode>si2_emu"): New expander.
35120 ("*fixuns_truncdfdi2_z13"): Rename to ...
35121 ("*fixuns_truncdfdi2_vx"): ... this.
35123 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35125 * config/s390/2964.md: Remove the single element vector compare
35126 instructions which are no longer used.
35127 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35129 (s390_canonicalize_comparison): Remove handling of DFmode
35131 (s390_expand_vec_compare_scalar): Remove function.
35132 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35133 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35135 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35137 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35139 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35141 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35142 will anyway by matched by mov<mode>_64dfp.
35144 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35146 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35147 vlef/vstef. Add missing operand to vleif.
35149 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35151 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35152 pair for all vector types with 64 bit elements.
35153 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35154 * config/s390/vector.md (V_HW_64): ... here.
35155 (V_128_NOSINGLE): New mode iterator.
35156 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35157 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35158 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35159 ("*vec_load_pairv2di"): Change to ...
35160 ("*vec_load_pair<mode>"): ... this one.
35162 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35164 * config/s390/constraints.md: Add comments.
35165 (jKK): Reject element sizes > 8 bytes.
35166 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35168 * config/s390/s390.md: Add the s_operand checks formerly in
35169 s390_split_ok_p to various splitters where they are still
35171 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35172 for 128 bit vectors. Plus two splitters.
35174 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35176 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35179 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35182 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35183 error if the boundary argument is not constant.
35185 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35187 PR rtl-optimization/80112
35188 * loop-doloop.c (doloop_condition_get): Don't check condition
35189 if cmp isn't SET with IF_THEN_ELSE src.
35191 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35193 PR tree-optimization/80158
35194 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35195 replacing a candidate statement, also replace it for the
35196 candidate's alternate interpretation.
35197 (replace_rhs_if_not_dup): Likewise.
35198 (replace_one_candidate): Likewise.
35200 2017-03-24 Richard Biener <rguenther@suse.de>
35202 PR tree-optimization/80167
35203 * graphite-isl-ast-to-gimple.c
35204 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35206 (translate_isl_ast_to_gimple::get_rename): Likewise.
35208 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35210 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35211 handling of certain combinations of target options, including the
35212 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35213 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35215 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35218 * config/arm/arm.md (*load_multiple): Add reload_completed to
35219 matching condition.
35221 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35222 Richard Biener <rguenth@suse.de>
35224 PR tree-optimization/79908
35225 PR tree-optimization/80136
35226 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35227 been cast away, gimplify_and_add suffices.
35229 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35231 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35233 2017-03-23 Richard Biener <rguenther@suse.de>
35235 PR tree-optimization/80032
35236 * gimplify.c (gimple_push_cleanup): Forced unconditional
35237 cleanups still have to go to the conditional_cleanups
35240 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35242 PR tree-optimization/80072
35243 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35245 (next_operand_entry_id): Change type to unsigned int.
35246 (sort_by_operand_rank): Make sure to return the right return value
35247 even if unsigned fields are bigger than INT_MAX.
35248 (struct oecount): Change cnt and id type to unsigned int.
35249 (oecount_hasher::equal): Formatting fix.
35250 (oecount_cmp): Make sure to return the right return value
35251 even if unsigned fields are bigger than INT_MAX.
35252 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35255 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35256 TREE_READONLY on result if writing it more than once.
35259 * doc/invoke.texi (-fsanitize=thread): Document that with
35260 -fnon-call-exceptions atomics are not able to throw
35264 * tsan.c: Include tree-eh.h.
35265 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35266 maybe_clean_or_replace_eh_stmt where needed.
35267 (instrument_memory_accesses): Add cfg_changed argument.
35268 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35269 if it returned true.
35270 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35272 PR rtl-optimization/63191
35273 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35274 wrapper function, moved the whole old content into ...
35275 (ix86_delegitimize_address_1): ... this. New inline function.
35276 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35277 true as last argument instead of ix86_delegitimize_address.
35279 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35281 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35282 cortexa57_branch_cost.
35284 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35286 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35288 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35291 * doc/md.texi (Constraints): Document wA constraint.
35292 * config/rs6000/constraints.md (wA): New.
35293 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35294 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35295 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35296 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35298 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35301 * gimplify.c (is_oacc_declared): New function.
35302 (oacc_default_clause): Use it to set default flags for acc declared
35303 variables inside parallel regions.
35304 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35305 declared variables.
35306 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35307 declare attribute to any decl as necessary.
35309 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35312 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35313 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35314 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35315 (arm_arch_lpae): This.
35316 * config/arm/arm.c (arm_arch7ve): Rename into ...
35317 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35318 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35321 2017-03-22 Martin Liska <mliska@suse.cz>
35324 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35325 error message instead of an ICE.
35327 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35329 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35331 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35333 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35335 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35338 2017-03-21 Martin Sebor <msebor@redhat.com>
35340 * doc/extend.texi: Use "cannot" instead of "can't."
35341 * doc/hostconfig.texi: Same.
35342 * doc/install.texi: Same.
35343 * doc/invoke.texi: Same.
35344 * doc/loop.texi: Same.
35345 * doc/md.texi: Same.
35346 * doc/objc.texi: Same.
35347 * doc/rtl.texi: Same.
35348 * doc/tm.texi: Same.
35349 * doc/tm.texi.in: Same.
35350 * doc/trouble.texi: Same.
35352 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35355 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35356 (collect_checksum_attributes): Set it.
35357 (die_checksum_ordered): Use it.
35359 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35361 PR tree-optimization/79908
35362 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35363 change: For a VA_ARG whose LHS has been cast away, use
35364 force_gimple_operand to construct the side effects.
35366 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35368 PR translation/80001
35369 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35370 more amenable to translation.
35371 (oacc_loop_auto_partitions): Likewise.
35373 2017-03-21 Marek Polacek <polacek@redhat.com>
35374 Martin Sebor <msebor@redhat.com>
35376 PR tree-optimization/80109
35377 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35378 on INTEGRAL_TYPE_P.
35380 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35381 Segher Boessenkool <segher@kernel.crashing.org>
35384 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35385 check reg_used_between_p between insn and one of succ or succ2
35386 depending on if succ is artificial insn not inserted into insn
35389 2017-03-21 Martin Liska <mliska@suse.cz>
35391 PR gcov-profile/80081
35392 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35393 * doc/gcc.texi: Include gcov-dump stuff.
35394 * doc/gcov-dump.texi: New file.
35396 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35398 PR rtl-optimization/79150
35399 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35400 conditional jump, if the jump is the last insn of the loop.
35402 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35403 Richard Biener <rguenth@suse.de>
35405 PR tree-optimization/79908
35406 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35407 been cast away, use force_gimple_operand to construct the side
35410 2017-03-21 Martin Liska <mliska@suse.cz>
35412 PR libfortran/79956
35413 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35416 2017-03-21 Brad Spengler <spender@grsecurity.net>
35419 * plugin.c (htab_hash_plugin): New function.
35420 (add_new_plugin): Use it and adjust.
35421 (parse_plugin_arg_opt): Adjust.
35422 (init_one_plugin): Likewise.
35424 2017-03-21 Richard Biener <rguenther@suse.de>
35426 PR tree-optimization/80032
35427 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35428 if set force the cleanup to happen unconditionally.
35429 (gimplify_target_expr): Push inserted clobbers with force_uncond
35430 to avoid them being removed by control-dependent DCE.
35432 2017-03-21 Richard Biener <rguenther@suse.de>
35434 PR tree-optimization/80122
35435 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35436 va_arg_pack_len when the inlined call stmt requires pack
35438 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35440 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35443 * tsan.c (instrument_builtin_call): If the memory model argument
35444 is not a constant, assume it is valid.
35447 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35450 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35452 PR rtl-optimization/79910
35453 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35454 if its dest is used by an insn before I2 (other than the combined
35455 insns themselves, which are properly handled already).
35457 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35460 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35462 * combine.c (record_used_regs): New static function.
35463 (try_combine): Handle situations where there is an additional
35464 instruction between I2 and I3 which needs to have a LOG_LINK
35468 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35470 * combine.c (try_combine): Delete redundant i1 test. Call
35471 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35473 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35476 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35477 alternatives 13/14.
35479 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35481 PR tree-optimization/80054
35482 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35483 the optimization if a PHI or any of its arguments is not dominated
35484 by the candidate's basis. Use gphi* rather than gimple* as
35486 (replace_profitable_candidates): Clean up a gimple* variable that
35487 should be a gphi* variable.
35489 2017-03-20 Martin Sebor <msebor@redhat.com>
35492 * doc/extend.texi (attribute constructor): Document present limitation.
35494 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35497 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35498 __POWER9_VECTOR__ #ifdef control, change template definition to
35499 use Power9-specific built-in function.
35500 (vec_any_eq): Likewise.
35501 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35502 to control outcomes from this test.
35503 (vector_ae_<mode>p): For VEC_F modes, likewise.
35505 2017-03-20 Ian Lance Taylor <iant@google.com>
35507 * config/i386/i386.c (ix86_function_regparm): Save an extra
35508 register for -fsplit-stack with DECL_STATIC_CHAIN.
35510 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35513 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35514 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35516 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35518 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35520 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35523 2017-03-20 Marek Polacek <polacek@redhat.com>
35526 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35528 2017-03-20 Richard Biener <rguenther@suse.de>
35530 PR tree-optimization/80113
35531 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35532 allocate extra SSA name for PHI def.
35533 (add_close_phis_to_outer_loops): Likewise.
35534 (add_close_phis_to_merge_points): Likewise.
35535 (copy_loop_close_phi_args): Likewise.
35536 (copy_cond_phi_nodes): Likewise.
35538 2017-03-20 Martin Liska <mliska@suse.cz>
35540 PR middle-end/79753
35541 * tree-chkp.c (chkp_build_returned_bound): Do not build
35542 returned bounds for a LHS that's not a BOUNDED_P type.
35544 2017-03-20 Martin Liska <mliska@suse.cz>
35548 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35549 COMPLEX_CST and VECTOR_CST.
35551 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35554 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35555 target operand. A new splitter adds the clobber statement in case
35556 the target operand is dead anyway.
35558 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35560 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35561 to age-old versions of binutils and glibc.
35563 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35565 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35567 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35569 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35571 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35573 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35574 requirement for binutils 2.13.
35576 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35578 * combine.c (try_combine): Delete redundant i1 test. Call
35579 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35581 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35583 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35584 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35586 <riscv64-*-elf>: Re-arrange section
35587 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35588 <riscv32-*-linux>: Likewise.
35589 <riscv64-*-elf>: Likewise
35590 <riscv64-*-linux>: Likewise.
35592 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35595 * aarch64.opt(verbose-cost-dump): Fix typo.
35597 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35600 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35601 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35603 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35605 * reload.c (find_reloads): When reloading a nonoffsettable address,
35606 use RELOAD_OTHER for it and its address reloads.
35608 PR rtl-optimization/79910
35609 * combine.c (record_used_regs): New static function.
35610 (try_combine): Handle situations where there is an additional
35611 instruction between I2 and I3 which needs to have a LOG_LINK
35614 2017-03-17 Jeff Law <law@redhat.com>
35616 PR tree-optimization/71437
35617 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35618 conditional in the hash table first.
35619 (vrp_dom_walker::before_dom_children): Extract condition from
35620 ASSERT_EXPR. Record condition, its inverion and any implied
35621 conditions as well.
35623 2017-03-17 Marek Polacek <polacek@redhat.com>
35624 Markus Trippelsdorf <markus@trippelsdorf.de>
35626 PR tree-optimization/80079
35627 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35630 2017-03-17 Richard Biener <rguenther@suse.de>
35632 PR middle-end/80075
35633 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35634 Properly verify the LHS before the RHS possibly claims to be
35636 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35639 2017-03-17 Martin Jambor <mjambor@suse.cz>
35641 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35642 (List of -O2 options): Likewise.
35643 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35646 2017-03-17 Tom de Vries <tom@codesourcery.com>
35648 * gcov-dump.c (print_usage): Print bug_report_url.
35650 2017-03-17 Richard Biener <rguenther@suse.de>
35652 PR middle-end/80050
35653 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35654 (parser::peek): Likewise.
35656 2017-03-17 Richard Biener <rguenther@suse.de>
35658 PR tree-optimization/80048
35659 * sese.c (free_sese_info): Properly release rename_map and
35660 copied_bb_map elements.
35662 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35664 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35665 Add linked-list forward and backlinks. Insert on
35666 construction, remove on destruction.
35667 (class pass_store_merging): Add m_stores_head field.
35668 (pass_store_merging::terminate_and_process_all_chains):
35669 Iterate over m_stores_head list.
35670 (pass_store_merging::terminate_all_aliasing_chains):
35672 (pass_store_merging::execute): Check for debug stmts first.
35673 Push new chains onto the m_stores_head stack.
35675 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35678 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35679 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35680 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35682 2017-03-16 Jeff Law <law@redhat.com>
35684 PR tree-optimization/71437
35685 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35686 member function. Implementation moved into after_dom_children
35687 member function and into the threader's thread_outgoing_edges
35689 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35690 some code into new thread_outgoing_edges.
35691 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35692 definition. Simplify marker handling (do it here). Assume we always
35693 have the available expression and the const/copies tables.
35694 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35696 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35697 * tree-vrp.c (equiv_stack): No longer file scoped.
35698 (vrp_dom_walker): New class.
35699 (vrp_dom_walker::before_dom_children): New member function.
35700 (vrp_dom_walker::after_dom_children): Likewise.
35701 (identify_jump_threads): Setup domwalker. Use it rather than
35702 walking edges in a random order by hand. Simplify setup/finalization.
35703 (finalize_jump_threads): Remove.
35704 (vrp_finalize): Do not call identify_jump_threads here.
35705 (execute_vrp): Do it here instead and call thread_through_all_blocks
35708 PR tree-optimization/71437
35709 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35711 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35713 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35714 (dom_opt_dom_walker::thread_across_edge): Remove
35715 handle_dominating_asserts argument. All callers changed.
35716 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35717 changes. Remove calls to lhs_of_dominating_assert. Other
35718 uses of handle_dominating_asserts turn into unconditional code
35719 (simplify_control_stmt_condition_1): Likewise.
35720 (simplify_control_stmt_condition): Likewise.
35721 (thread_through_normal_block, thread_across_edge): Likewise.
35722 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35723 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35724 object if it is not an SSA_NAME.
35725 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35726 before calling into the VRP specific simplifiers.
35727 (identify_jump_threads): Remove handle_dominating_asserts
35730 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35733 * tree-diagnostic.c (default_tree_printer): No longer static.
35734 * tree-diagnostic.h (default_tree_printer): New prototype.
35736 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35738 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35739 Change ins into fmov.
35741 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35743 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35744 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35745 Use h_con constraint for operand 1.
35746 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35747 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35749 2017-03-15 Jeff Law <law@redhat.com>
35751 PR tree-optimization/71437
35752 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35753 (record_temporary_equivalences): Use it.
35755 PR tree-optimization/71437
35756 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35757 tree-ssa-scopedtables.
35758 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35759 (record_conditions, record_cond, vuse_eq): Likewise.
35760 (record_edge_info): Adjust to API tweak of record_conditions.
35761 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35762 (record_temporary_equivalences, optimize_stmt): Likewise.
35763 (eliminate_redundant_computations): Likewise.
35764 (record_equivalences_from_stmt): Likewise.
35765 * tree-ssa-scopedtables.c: Include options.h and params.h.
35766 (vuse_eq): New function, moved from tree-ssa-dom.c
35767 (build_and_record_new_cond): Likewise.
35768 (record_conditions): Likewise. Accept vector of conditions rather
35769 than edge_equivalence structure for first argument.
35770 for the first argument.
35771 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35772 from tree-ssa-dom.c.
35773 (avail_exprs_stack::record_cond): Likewise.
35774 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35775 from tree-ssa-dom.c.
35776 (avail_exprs_stack): Add new member functions lookup_avail_expr
35778 (record_conditions): Declare.
35780 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35783 * lra-constraints.c (process_alt_operands): Increase reject for
35784 reloading an input/output operand.
35786 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35789 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35790 insns to convert from signed/unsigned char/short to IEEE 128-bit
35792 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35794 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35797 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35798 subreg of inner mode for values already in registers.
35800 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35802 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35803 iteration reg is used after the loop.
35805 2017-03-14 Martin Sebor <msebor@redhat.com>
35807 PR tree-optimization/79800
35808 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35809 precision in negative-positive range.
35810 (format_floating): Call non-const overload with adjusted precision.
35812 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35815 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35818 2017-03-14 Martin Sebor <msebor@redhat.com>
35820 PR middle-end/80020
35821 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35822 * builtins.def (aligned_alloc): Use it.
35825 * Makefile.in (GTFILES): Add calls.c.
35826 * calls.c: Include "gt-calls.h".
35828 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35830 PR rtl-optimization/79728
35831 * regs.h (struct target_regs): New field
35832 x_contains_allocatable_regs_of_mode.
35833 (contains_allocatable_regs_of_mode): New macro.
35834 * reginfo.c (init_reg_sets_1): Initialize it, and change
35835 contains_reg_of_mode so it includes global regs as well.
35836 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35837 rather than contains_regs_of_mode.
35839 2017-03-14 Martin Liska <mliska@suse.cz>
35841 * doc/invoke.texi: Document options that can't be combined with
35842 -fcheck-pointer-bounds.
35844 2017-03-14 Martin Liska <mliska@suse.cz>
35846 PR middle-end/79831
35847 * doc/invoke.texi (-Wchkp): Document the option.
35849 2017-03-14 Martin Liska <mliska@suse.cz>
35851 * Makefile.in: Install gcov-dump.
35853 2017-03-14 Martin Liska <mliska@suse.cz>
35855 * multiple_target.c (expand_target_clones): Bail out for
35856 an invalid attribute.
35858 2017-03-14 Richard Biener <rguenther@suse.de>
35860 * alias.c (struct alias_set_entry): Pack properly.
35861 * cfgloop.h (struct loop): Likewise.
35862 * cse.c (struct set): Likewise.
35863 * ipa-utils.c (struct searchc_env): Likewise.
35864 * loop-invariant.c (struct invariant): Likewise.
35865 * lra-remat.c (struct cand): Likewise.
35866 * recog.c (struct change_t): Likewise.
35867 * rtl.h (struct address_info): Likewise.
35868 * symbol-summary.h (function_summary): Likewise.
35869 * tree-loop-distribution.c (struct partition): Likewise.
35870 * tree-object-size.c (struct object_size_info): Likewise.
35871 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35872 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35873 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35874 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35875 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35876 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35877 (struct _stmt_vec_info): Likewise.
35879 2017-03-14 Martin Liska <mliska@suse.cz>
35882 * multiple_target.c (create_dispatcher_calls): Check that
35883 a target can create a function dispatcher.
35885 2017-03-14 Martin Liska <mliska@suse.cz>
35888 * multiple_target.c (expand_target_clones): Drop local.local
35889 flag for default implementation.
35891 2017-03-14 Richard Biener <rguenther@suse.de>
35893 PR tree-optimization/80030
35894 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35896 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35898 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35899 gcc_fallthrough() instead of __attribute__((fallthrough));
35901 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35903 * doc/gcc.texi: Remove "up" link to (DIR).
35904 * doc/gccint.texi: Ditto.
35906 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35908 * doc/install.texi (Specific) <avr>: Remove reference to
35911 2017-03-13 Jeff Law <law@redhat.com>
35913 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35914 attribute rather than comments.
35916 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35917 match_scratch operand is highest.
35919 2017-03-13 Martin Liska <mliska@suse.cz>
35921 PR middle-end/78339
35922 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35923 is a CHKP clone, use original declaration.
35925 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35927 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35928 (arc_conditional_register_usage): Use a different allocation order
35929 when optimizing for size.
35930 * common/config/arc/arc-common.c (arc_option_optimization_table):
35931 Section anchors default on when optimizing for size.
35933 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35935 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35937 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35939 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35940 * config/arc/arc.md (cpu_facility): Add cd variant.
35941 (*movqi_insn): Add code density variant.
35942 (*movhi_insn): Likewise.
35943 (*movqi_insn): Likewise.
35944 (*addsi3_mixed): Likewise.
35945 (subsi3_insn): Likewise.
35947 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35949 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35951 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35953 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35954 expressions with MINUS and UNARY ops.
35956 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35959 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35961 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35962 between vec_select and vector argument.
35963 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35964 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35965 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35966 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35967 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35968 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35970 2017-03-13 Richard Biener <rguenther@suse.de>
35973 * params.def (vect-max-peeling-for-alignment): Fix typo.
35975 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35977 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35978 issue that only occurred with binutils below 2.18.
35980 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35982 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35983 refer to binutils 2.11/2.12 minimum.
35985 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35987 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35988 ftp.kernel.org and simplify binutils requirement.
35990 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35992 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35994 (Optimize Options): Ditto. Also remove redundancy.
35996 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35998 PR translation/79848
35999 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36001 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36002 to G_ to avoid double translation.
36004 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36006 PR translation/79923
36007 * auto-profile.c (get_combined_location): Convert leading
36008 character of diagnostics to lower case and remove trailing period.
36009 (read_profile): Likewise for various diagnostics.
36010 * config/arm/arm.c (arm_option_override): Remove trailing period
36011 from various diagnostics.
36012 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36013 (msp430_expand_delay_cycles): Likewise.
36015 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36018 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36019 full command-line argument, rather than just "str".
36020 (aarch64_validate_march): Likewise.
36021 (aarch64_validate_mtune): Likewise.
36023 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36025 PR rtl-optimization/78911
36026 * lra-assigns.c (must_not_spill_p): New function.
36027 (spill_for): Use it.
36029 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36031 PR tree-optimization/79981
36032 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36033 ATOMIC_COMPARE_EXCHANGE ifn result.
36034 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36035 IFN_ATOMIC_COMPARE_EXCHANGE.
36037 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36040 * opts.c (parse_sanitizer_options): Add missing question mark to
36041 "did you mean" message.
36043 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36045 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36047 (VMULEUH_UNS): Likewise.
36048 (VMULOUB_UNS): Likewise.
36049 (VMULOUH_UNS): Likewise.
36050 * config/rs6000/rs6000.c (builtin_function_type): Remove
36051 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36053 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36056 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36057 x with result of extra_parsing_for_operand_code_0.
36058 (function_reader::extra_parsing_for_operand_code_0): Convert
36059 return type from void to rtx, returning x. When reading
36060 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36061 larger size containing struct block_symbol.
36063 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36065 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36066 -mfloat128-hardware without -m64.
36068 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36071 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36072 entries to the case statement that marks unsigned arguments to
36073 overloaded functions.
36075 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36077 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36078 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36080 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36083 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36084 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36086 2017-03-10 Martin Liska <mliska@suse.cz>
36090 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36091 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36094 2017-03-10 Olivier Hainque <hainque@adacore.com>
36096 * tree-switch-conversion (array_value_type): Start by resetting
36097 candidate type to it's main variant.
36099 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36101 PR rtl-optimization/79909
36102 * combine.c (try_combine): Use simplify_replace_rtx on individual
36103 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36104 of the whole CALL_INSN_FUNCTION_USAGE.
36106 PR tree-optimization/79972
36107 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36108 get_range_info on SSA_NAMEs. Formatting fixes.
36110 2017-03-10 Richard Biener <rguenther@suse.de>
36111 Jakub Jelinek <jakub@redhat.com>
36113 PR tree-optimization/77975
36114 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36115 edge to be constant.
36116 (get_val_for): For constant x return it. Formatting fix.
36117 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36118 would use the same bases as the current one.
36120 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36122 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36123 instead of vec_select for V1TImode.
36124 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36126 (VSX_LE_128): Add V1TI to this mode iterator.
36127 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36128 (*vsx_le_perm_store_<mode>): Likewise.
36129 (pre-reload splitter for VSX stores): Likewise.
36130 (post-reload splitter for VSX stores): Likewise.
36131 (*vsx_xxpermdi2_le_<mode>): Likewise.
36132 (*vsx_lxvd2x2_le_<mode>): Likewise.
36133 (*vsx_stxvd2x2_le_<mode>): Likewise.
36135 2017-03-09 Michael Eager <eager@eagercon.com>
36137 Correct failures with --enable-checking=yes,rtl.
36139 * config/microblaze/microblaze.c (microblaze_expand_shift):
36140 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36141 test for const0_rtx.
36142 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36143 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36145 2017-03-09 Richard Biener <rguenther@suse.de>
36147 PR tree-optimization/79977
36148 * graphite-scop-detection.c (scop_detection::merge_sese):
36149 Handle the case of extra exits to blocks dominating the entry.
36151 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36153 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36156 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36158 PR rtl-optimization/79949
36159 * lra-constraints.c (process_alt_operands): Check memory when
36160 trying to predict a cycle. Print about the overall increase.
36162 2017-03-09 Richard Biener <rguenther@suse.de>
36164 PR middle-end/79971
36165 * gimple-expr.c (useless_type_conversion_p): Preserve
36166 TYPE_SATURATING for fixed-point types.
36168 2017-03-09 Richard Biener <rguenther@suse.de>
36171 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36172 alignment of BLKmode params.
36174 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36177 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36178 (VALL_NO_V2Q): Likewise.
36180 * config/aarch64/aarch64-simd.md
36181 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36183 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36184 VALL_NO_V2Q mode iterator.
36185 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36187 2017-03-09 Martin Liska <mliska@suse.cz>
36189 PR tree-optimization/79631
36190 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36191 tree_int_cst_sign_bit just for INTEGER constants.
36193 2017-03-09 Martin Liska <mliska@suse.cz>
36197 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36200 2017-03-09 Marek Polacek <polacek@redhat.com>
36203 * tree.c (inchash::add_expr): Handle TREE_VEC.
36205 2017-03-09 Martin Liska <mliska@suse.cz>
36208 (chkp_narrow_size_and_offset): New function.
36209 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36210 (void chkp_parse_bit_field_ref): New function.
36211 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36212 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36214 2017-03-09 Martin Liska <mliska@suse.cz>
36217 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36218 (chkp_find_bounds_1): Remove gcc_unreachable.
36220 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36223 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36224 BUILT_IN_SYNC*, determine the access type from the size suffix and
36225 always build a MEM_REF with that type. Handle forgotten
36226 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36229 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36230 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36231 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36232 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36233 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36234 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36235 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36236 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36237 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36238 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36239 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36240 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36241 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36242 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36243 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36244 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36245 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36246 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36247 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36248 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36249 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36250 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36251 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36252 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36253 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36254 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36255 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36256 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36257 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36258 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36259 definitions outside of __OPTIMIZE__ guarded section.
36262 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36263 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36264 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36265 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36268 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36270 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36271 ("vfenez<mode>"): Add missing constraints.
36273 2017-03-08 Martin Sebor <msebor@redhat.com>
36276 * config/nds32/nds32.c (nds32_option_override):
36277 Fix misspelled diagnostic.
36279 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36282 * gimplify.c (gimplify_omp_for): Replace index var in outer
36283 taskloop statement with an artificial variable and add
36284 OMP_CLAUSE_PRIVATE clause for it.
36286 2017-03-08 Richard Biener <rguenther@suse.de>
36288 PR tree-optimization/79955
36289 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36290 for accesses that are completely outside of the variable.
36292 2017-03-08 Andrew Haley <aph@redhat.com>
36294 PR tree-optimization/79943
36295 * tree-ssa-loop-split.c (compute_new_first_bound): When
36296 calculating the new upper bound, (END-BEG) should be added, not
36299 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36301 * config/avr/avr.md (setmemhi): Make sure match_dup
36302 operand number comes before match_scratch.
36304 2017-03-08 Richard Biener <rguenther@suse.de>
36306 PR tree-optimization/79920
36307 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36308 with ncopies == 1 to ...
36309 (vect_transform_slp_perm_load): ... here. Properly compute
36310 all element loads by iterating VF times over the group. Do
36311 not handle ncopies (computed in a broken way) in
36312 vect_create_mask_and_perm.
36314 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36317 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36318 is a uniform vector, use uniform_vector_p return value instead of
36319 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36321 2017-03-07 Marek Polacek <polacek@redhat.com>
36323 PR middle-end/79809
36324 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36325 (alloca_call_type): Likewise.
36327 2017-03-07 Martin Liska <mliska@suse.cz>
36329 * gcov.c (process_args): Put comment to correct location.
36331 2017-03-07 Martin Liska <mliska@suse.cz>
36333 PR middle-end/68270
36334 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36335 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36336 (chkp_narrow_bounds_for_field): Likewise.
36337 (chkp_parse_array_and_component_ref): Pass one more argument to
36340 2017-03-07 Richard Biener <rguenther@suse.de>
36342 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36345 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36347 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36348 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36350 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36353 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36354 to end of description.
36355 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36357 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36359 PR rtl-optimization/79901
36360 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36362 (*avx512f_<code><mode>3<mask_name>): ... this.
36363 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36364 iterator instead of VI8_AVX2_AVX512BW.
36366 PR rtl-optimization/79901
36367 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36368 min/max expander, expand it using expand_vec_cond_expr.
36371 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36374 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36377 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36378 to void * for PCH reasons.
36379 * dwarf2out.c (output_loc_operands, output_die): Cast
36380 v.val_vec.array to unsigned char *.
36382 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36385 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36388 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36390 PR rtl-optimization/79571
36391 * lra-constraints.c (process_alt_operands): Calculate static
36392 reject and subtract it from overall when only addresses will be
36395 2017-03-06 Julia Koval <julia.koval@intel.com>
36398 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36399 incoming stack boundary to 128 for 64-bit targets.
36401 2017-03-06 Richard Biener <rguenther@suse.de>
36403 PR tree-optimization/79894
36404 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36405 to NULL after folding it.
36407 2017-03-06 Richard Biener <rguenther@suse.de>
36409 PR tree-optimization/79824
36410 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36411 check disabling peeling for gaps.
36413 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36415 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36416 attributes): Document gettimeofday.
36418 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36420 * config/s390/s390.c (s390_option_override_internal): Set
36421 PARAM_MIN_VECT_LOOP_BOUND
36423 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36425 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36426 * config/s390/s390.md: Likewise.
36428 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36431 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36432 (<avx2_avx512>_perm<mode>): Rename to ...
36433 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36435 (<avx512>_perm<mode>_mask): Rename to ...
36436 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36438 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36439 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36440 instead of VI8F_256_512.
36441 (avx512f_perm<mode>): New define_expand.
36442 (avx512f_perm<mode>_mask): Likewise.
36443 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36444 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36446 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36448 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36449 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36451 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36453 2017-03-06 Martin Liska <mliska@suse.cz>
36456 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36457 when having a SSA NAME w/o VAR_DECL assigned to it.
36459 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36461 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36462 msa_dpsub_<su>_d): Fix MODE for vec_select.
36464 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36466 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36468 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36470 2017-03-06 Richard Biener <rguenther@suse.de>
36472 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36473 * plugin.c (register_plugin_info): Likewise.
36474 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36476 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36478 * config/i386/sse.md (sse_storehps, sse_storelps,
36479 avx_<castmode><avxsizesuffix>_<castmode>,
36480 avx512f_<castmode><avxsizesuffix>_<castmode>,
36481 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36482 in condition that at least one operand is not a MEM.
36484 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36486 PR middle-end/79805
36487 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36488 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36490 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36491 gimple_call_nothrow_p flag based on whether original builtin can throw.
36492 If it can, emit following stmts on the fallthrough edge.
36493 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36494 don't create new bb if inserting just debug stmts on the edge, try to
36495 insert them on the fallthru bb or just reset debug stmts.
36497 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36500 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36501 restore recog_data (including the operand rtxes inside it) around
36502 the call to get_insn_template.
36504 2017-03-03 Martin Sebor <msebor@redhat.com>
36506 PR tree-optimization/79699
36507 * context.c (context::~context): Free MPFR caches to avoid
36508 a memory leak on program exit.
36510 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36512 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36513 Use wide_int::ulow () instead of .elt (0).
36515 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36517 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36518 (*pushxf): Limit oF constraint to 32bit targets and add oC
36519 constraint for 64bit targets.
36520 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36521 (*pushdf): Change rmF constraint to rmC.
36523 2017-03-03 Martin Liska <mliska@suse.cz>
36525 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36526 Remove unused variable.
36528 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36531 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36532 is a memory operand, increase num_memory.
36533 (ix86_expand_args_builtin): Likewise.
36535 2017-03-03 Jan Hubicka <jh@suse.cz>
36538 * ipa-devirt.c (maybe_record_node): Properly handle
36539 __cxa_pure_virtual visibility.
36541 2017-03-03 Martin Liska <mliska@suse.cz>
36543 PR tree-optimization/79803
36544 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36546 (pass_loop_prefetch::execute): Disabled optimization if an
36547 assumption about L1 cache size is not met.
36549 2017-03-03 Martin Liska <mliska@suse.cz>
36551 PR rtl-optimization/79574
36552 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36553 (hash_scan_set): Likewise.
36554 (dump_hash_table): Likewise.
36555 (hoist_code): Likewise.
36557 2017-03-03 Richard Biener <rguenther@suse.de>
36559 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36560 in place of uhigh ().
36561 (fixed_convert_from_real): Likewise.
36563 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36566 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36568 2017-03-03 Richard Biener <rguenther@suse.de>
36570 PR middle-end/79818
36571 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36572 TYPE_OVERFLOW_UNDEFINED check.
36574 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36576 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36578 (vector_ae_<mode>_p): Likewise.
36579 (vector_nez_<mode>_p): Likewise.
36580 (vector_ne_v2di_p): Likewise.
36581 (vector_ae_v2di_p): Likewise.
36582 (vector_ne_<mode>_p): Likewise.
36583 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36585 (vsx_tsqrt<mode>2_fe): Likewise.
36587 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36590 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36592 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36594 PR rtl-optimization/79780
36595 * cprop.c (one_cprop_pass): When second and further conditional trap
36596 in a single basic block is turned into an unconditional trap, turn it
36597 into a deleted note to avoid RTL verification failures.
36599 2017-03-02 Richard Biener <rguenther@suse.de>
36601 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36603 2017-03-02 Richard Biener <rguenther@suse.de>
36605 PR tree-optimization/79345
36607 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36608 param and abort the walk, returning -1 if it is hit.
36609 (walk_aliased_vdefs): Take a limit param and pass it on.
36610 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36611 defaulting to 0 and return a signed int.
36612 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36613 (check_defs): New helper.
36614 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36615 about uninitialized memory.
36616 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36617 bogus uninitialized warning.
36618 (fixed_convert_from_real): Likewise.
36620 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36622 PR tree-optimization/66768
36623 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36624 iv_use if base object can't be determined.
36626 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36628 PR tree-optimization/79345
36629 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36630 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36631 (get_pattern_stats): Initialize it.
36632 * genemit.c (gen_expand): Verify match_scratch numbers come after
36633 match_operand/match_dup numbers.
36634 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36635 match_scratch numbers.
36636 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36638 * config/s390/s390.md (trunctdsd2): Likewise.
36640 2017-03-02 Richard Biener <rguenther@suse.de>
36642 * wide-int.h (wide_int_storage::operator=): Implement in terms
36645 2017-03-02 Richard Biener <rguenther@suse.de>
36647 PR tree-optimization/79777
36648 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36649 the to insert expression to sth existing.
36651 2017-03-01 Martin Sebor <msebor@redhat.com>
36653 PR middle-end/79692
36654 * gimple-ssa-sprintf.c
36655 (directive::known_width_and_precision): New function.
36656 (format_integer): Use it.
36657 (get_mpfr_format_length): Consider the full range of precision
36658 when computing %g output with the # flag. Set the likely byte
36659 count to 3 rather than 1 when precision is indeterminate.
36660 (format_floating): Correct the lower bound of precision.
36662 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36664 * doc/invoke.texi: Document default code model for 64-bit Linux.
36666 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36669 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36670 udiv rather than div since input pattern is unsigned.
36672 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36674 * config/i386/i386.c (print_reg): Warn for values of
36675 unsupported size in integer register.
36677 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36680 * config/rs6000/predicates.md (current_file_function_operand): Do
36681 not allow self calls to be local if the function is replaceable.
36683 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36686 * config/rs6000/altivec.h (vec_ctz and others): Change the
36687 preprocessor macro that controls conditional compilation from
36688 _ARCH_PWR9 to __POWER9_VECTOR__.
36689 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36690 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36691 control (instead of _ARCH_PWR9 control) so that template
36692 definition uses power9-specific function.
36693 (vec_any_eq): Likewise.
36694 (vec_all_ne): Change macro definition to use a power9-specific
36695 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36696 _ARCH_PWR9 control).
36697 (vec_any_eq) Likewise.
36698 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36699 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36700 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36701 support for xvcmpnedp instruction.
36702 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36703 macro expansion so that Power9 implementation of vec_all_ne does
36704 not use the AltiVec predicate framework.
36705 (VCMPNEH_P): Likewise.
36706 (VCMPNEW_P): Likewise.
36707 (VCMPNED_P): Likewise.
36708 (VCMPNEFP_P): Likewise.
36709 (VCMPNEDP_P): Likewise.
36710 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36711 implementation of vec_any_eq to not use AltiVec predicate
36713 (VCMPAEH_P): Likewise.
36714 (VCMPAEW_P): Likewise.
36715 (VCMPAED_P): Likewise.
36716 (VCMPAEFP_P): Likewise.
36717 (VCMPAEDP_P): Likewise.
36718 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36719 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36720 not use the AltiVec predicate framework.
36721 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36722 of vec_any_eq to not use AltiVec predicate framework.
36723 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36724 support for predefined __POWER9_VECTOR__ macro to indicate that
36725 Power9 instruction selection is enabled.
36726 (altivec_overloaded_builtins): Remove extraneous
36727 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36728 function argument types RS6000_BTI_bool_V16QI and
36729 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36730 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36731 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36732 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36733 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36734 Power9 for implementations of vec_cmpne. Change the signature for
36735 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36736 (representing vec_all_ne) to remove the previously described first
36737 argument of type RS6000_BTI_INTSI, as this was an artifact of
36738 reliance on the AltiVec predicate framework, which is no longer
36739 used in the implementation of these functions. Add
36740 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36741 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36742 since, unlike the AltiVec predicate framework implementation, we
36743 do not share function descriptors between vec_alle and vec_anyeq.
36744 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36745 set of modes that receive special treatment even when
36746 TARGET_P9_VECTOR is true. The special treatment emits code that
36747 does not depend on Power9 instructions.
36748 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36749 define_expand to not rely on AltiVec predicate framework.
36750 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36752 (vector_ne_v2di_p): Change this define_expand to not rely on
36753 AltiVec predicate framework.
36754 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36756 (vector_ne_<mode>_p): Change this define_expand to not rely on
36757 AltiVec predicate framework.
36758 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36760 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36761 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36762 define_insn pattern.
36763 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36764 define_insn pattern because the xvcmpne<VSs>. instruction is not
36766 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36767 instruction is not supported.
36769 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36771 * config/nvptx/nvptx.c: Include intl.h.
36773 2017-03-01 Martin Jambor <mjambor@suse.cz>
36776 * ipa-prop.h (ipa_bits): Removed field known.
36777 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36778 to pointers. Adjusted their comments to warn about their sharing.
36779 (ipcp_transformation_summary): Change bits to a vector of pointers.
36780 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36781 (ipa_get_ipa_bits_for_value): Declare.
36782 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36783 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36784 (ipa_bits_hash_table): Likewise.
36785 (ipa_vr_ggc_hash_traits): Likewise.
36786 (ipa_vr_hash_table): Likewise.
36787 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36788 being pointers and vr_known being removed.
36789 (ipa_set_jf_unknown): Likewise.
36790 (ipa_get_ipa_bits_for_value): New function.
36791 (ipa_set_jfunc_bits): Likewise.
36792 (ipa_get_value_range): New overloaded functions.
36793 (ipa_set_jfunc_vr): Likewise.
36794 (ipa_compute_jump_functions_for_edge): Use the above functions to
36795 construct bits and vr parts of jump functions.
36796 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36797 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36799 (ipcp_grow_transformations_if_necessary): Also allocate
36800 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36802 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36803 them. Fix too long lines.
36804 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36805 vr_known being removed.
36806 (ipa_read_jump_function): Use new setter functions to construct bits
36807 and vr parts of jump functions or set them to NULL.
36808 (write_ipcp_transformation_info): Adjust for bits being pointers.
36809 (read_ipcp_transformation_info): Likewise.
36810 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36812 Include gt-ipa-prop.h.
36813 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36815 (ipcp_store_bits_results): Likewise.
36816 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36817 Do not write to existing jump functions but use a temporary instead.
36819 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36822 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36823 attempt to use its first operand as BIT_FIELD_REF base.
36825 2017-03-01 Richard Biener <rguenther@suse.de>
36827 PR middle-end/79721
36828 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36829 interpolating formula in wrapping arithmetic.
36830 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36832 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36834 PR tree-optimization/79734
36835 * tree-vect-generic.c (expand_vector_condition): Optimize
36836 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36837 Handle VEC_COND_EXPR where comparison has different inner width from
36838 type's inner width.
36840 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36842 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36843 markup, and similar issues. Remove @opindex entries for things
36844 that aren't options. Add missing -mmpy-option entries.
36846 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36848 PR tree-optimization/79737
36849 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36850 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36851 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36852 instead of byte_size. Formatting fix.
36853 (shift_bytes_in_array_right): Formatting fix.
36855 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36858 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36859 condition on optimize for the leaf function test.
36861 2017-02-28 Martin Liska <mliska@suse.cz>
36864 * read-rtl-function.c (function_reader::handle_unknown_directive):
36865 Bail out when one uses -flto.
36867 2017-02-28 Martin Liska <mliska@suse.cz>
36869 * common.opt: Replace space with tabular for options of <number>
36871 * config/i386/i386.opt: Show <number> value for
36872 -mlarge-data-threshold.
36873 * opts.c (print_filtered_help): Do not display number in hexadecimal
36876 2017-02-28 Martin Liska <mliska@suse.cz>
36878 * common.opt: Fix --help=option -Q for options which are of
36881 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36883 * config/i386/i386.c (print_reg): Error out for values
36884 of 8-bit size in invalid integer register.
36886 2017-02-28 Martin Sebor <msebor@redhat.com>
36888 PR tree-optimization/79691
36889 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36891 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36894 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36895 gcc_unreachable with output_operand_lossage.
36897 2017-02-28 Richard Biener <rguenther@suse.de>
36899 PR tree-optimization/79740
36900 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36902 (visit_nary_op): Insert the nary into the hashtable if we
36903 pattern-matched sth.
36904 * tree-ssa-pre.c (eliminate_insert): Robustify.
36906 2017-02-28 Richard Biener <rguenther@suse.de>
36908 PR middle-end/79731
36909 * fold-const.c (decode_field_reference): Reject out-of-bound
36912 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36914 * config/i386/i386.c: Include intl.h.
36915 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36916 instead of just cond ? "..." : "...".
36917 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36918 * coverage.c (read_counts_file): Likewise.
36919 * omp-offload.c: Include intl.h.
36920 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36921 of just cond ? "..." : "...".
36922 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36923 of just cond ? "..." : "...".
36925 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36928 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36930 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36931 'tune for' CPU name.
36932 * config/arm/arm-cpu-data.h: Regenerated.
36934 2017-02-28 Richard Biener <rguenther@suse.de>
36936 PR tree-optimization/79732
36937 * tree-inline.c (expand_call_inline): Do not shadow var.
36939 2017-02-28 Richard Biener <rguenther@suse.de>
36941 PR tree-optimization/79723
36942 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36943 address-space properly.
36945 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36947 * doc/optinfo.texi (Optimization groups): Fix option used for
36949 * doc/invoke.texi (-fopt-info): Document "omp".
36950 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36951 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36952 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36953 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36954 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36956 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36958 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36959 OPTGROUP_OMP with "omp".
36961 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36964 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36965 for arithmetic shift of unsigned V2DI.
36967 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36969 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36970 arc/linux.h headers.
36971 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36972 (LINK_SPEC): Likewise.
36973 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36974 (EXTRA_SPECS): Likewise.
36975 (STARTFILE_SPEC): Likewise.
36976 (ENDFILE_SPEC): Likewise.
36977 (LIB_SPEC): Likewise.
36978 (TARGET_SDATA_DEFAULT): Likewise.
36979 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36980 (MULTILIB_DEFAULTS): Likewise.
36981 (DWARF2_UNWIND_INFO): Likewise.
36982 * config/arc/big.h: New file.
36983 * config/arc/elf.h: Likewise.
36984 * config/arc/linux.h: Likewise.
36985 * config/arc/t-uClibc: Remove.
36987 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36989 PR tree-optimization/77536
36990 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36991 (tree_transform_and_unroll_loop): Use above function to compute the
36992 estimated niter of unrolled loop and use it when scaling profile.
36993 Also use count info rather than frequency if it's non-zero.
36994 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36995 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36996 (vect_transform_loop): Call above function.
36998 2017-02-27 Richard Biener <rguenther@suse.de>
37000 PR tree-optimization/45397
37001 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37002 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37003 (visit_nary_op): Add pattern matching for CSEing sign-changed
37004 or truncated operations with wider ones.
37006 2017-02-27 Richard Biener <rguenther@suse.de>
37008 PR tree-optimization/79690
37009 * tree-vect-stmts.c (vectorizable_store): Use vector type
37010 built from the DR with address-space.
37012 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37014 * doc/invoke.texi (Optimize Options): Refine the description
37015 of asan-use-after-return.
37017 2017-02-25 Alan Modra <amodra@gmail.com>
37019 PR rtl-optimization/79584
37020 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37021 base, not ad->base_term, the reg within base. Remove assertion
37022 that ad->base == ad->base_term. Replace gen_int_mode using
37023 bogus mode with const0_rtx.
37025 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37027 PR middle-end/79396
37028 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37029 FMA_EXPR like tcc_binary or tcc_unary.
37031 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37034 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37036 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37037 (output_loc_operands): Handle DW_OP_call_ref and
37038 DW_OP_GNU_variable_value.
37039 (struct variable_value_struct): New type.
37040 (struct variable_value_hasher): Likewise.
37041 (variable_value_hash): New variable.
37042 (string_types): Remove.
37043 (copy_loc_descr): New function.
37044 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37045 (prepend_loc_descr_to_each): New function.
37046 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37047 instead of add_loc_descr_to_each if the first argument is single
37048 location list and the second has multiple.
37049 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37050 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37051 when looking for variable value which doesn't have other location info.
37052 (loc_list_from_tree): Formatting fix.
37053 (gen_array_type_die): Simplify DW_AT_string_length handling.
37054 (adjust_string_types): Remove.
37055 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37056 string_types. Call resolve_variable_values.
37057 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37058 (resolve_addr_in_expr): Likewise. Add A argument.
37059 (copy_deref_exprloc): Remove deref argument. Adjust for the
37060 original expression being DW_OP_GNU_variable_value with optionally
37061 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37062 optionally after it.
37063 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37064 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37065 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37066 (variable_value_hasher::hash, variable_value_hasher::equal): New
37068 (resolve_variable_value_in_expr, resolve_variable_value,
37069 resolve_variable_values, note_variable_value_in_expr,
37070 note_variable_value): New functions.
37071 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37074 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37077 * opts.h (handle_generated_option): Add GENERATED_P argument.
37078 * opts-common.c (handle_option): Adjust function comment.
37079 (handle_generated_option): Add GENERATED_P argument, pass it to
37081 (control_warning_option): Pass false to handle_generated_option
37083 * opts.c (maybe_default_option): Pass true to handle_generated_option
37085 * optc-gen.awk: Likewise.
37087 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37089 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37090 a REG, look at the REG it is a SUBREG of.
37091 (splitter for cmpeqsi_t): Ditto.
37093 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37095 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37096 the special USEs with the pattern of the insn, not the insn itself.
37098 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37101 * doc/invoke.texi: Document -mload-store-pairs.
37103 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37104 Sandra Loosemore <sandra@codesourcery.com>
37106 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37107 argument isn't a CONST_INT.
37108 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37109 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37110 (nios2_validate_compare): Bypass alternate compare logic if *op2
37111 is not a CONST_INT.
37112 (ldstwm_operation_p): Return false if first_base is not a REG or
37113 if first_offset is not a CONST_INT.
37115 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37117 * config/cris/cris.md: Use correct operand in a define_peephole2.
37119 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37121 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37123 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37125 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37126 this_insn if it is an INSN or JUMP_INSN.
37127 (force_offsettable): Look at base, not at addr.
37128 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37129 on things that aren't necessarily CONST_INTs.
37131 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37133 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37134 -mfpmath=sse is the default also for x86-32 targets with SSE2
37135 instruction set when @option{-ffast-math} is enabled
37137 2017-02-24 Jeff Law <law@redhat.com>
37139 PR rtl-optimizatoin/79286
37140 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37143 2017-02-24 Richard Biener <rguenther@suse.de>
37145 PR tree-optimization/79389
37146 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37149 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37151 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37152 function comment to reflect reality.
37153 (loop_exits_before_overflow): Fix typo in function description.
37155 2017-02-24 Richard Biener <rguenther@suse.de>
37157 PR tree-optimization/79389
37158 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37159 properly that a threading opportunity exists. Detect conditional
37160 copy/constant propagation opportunities.
37162 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37164 * config/visium/visium.md (type): Add trap.
37165 (b): New mode attribute.
37166 (*btst): Rename into...
37167 (*btst<mode>): ...this and adjust.
37168 (*cbranchsi4_btst_insn): Rename into...
37169 (*cbranch<mode>4_btst_insn): ...this and adjust.
37170 (trap): New define_insn.
37172 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37174 PR tree-optimization/79389
37175 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37176 (noce_reversed_cond_code): New function.
37177 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37178 reversed_comparison_code. Formatting fix.
37179 (noce_try_store_flag): Test rev_cond != NULL in addition to
37180 reversed_comparison_code.
37181 (noce_try_store_flag_constants): Likewise.
37182 (noce_try_store_flag_mask): Likewise.
37183 (noce_try_addcc): Use rev_cond if non-NULL instead of
37184 reversed_comparison_code.
37185 (noce_try_cmove_arith): Likewise. Formatting fixes.
37186 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37187 (noce_find_if_block): Initialize rev_cond.
37188 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37189 instead of false as last argument never attempt to reverse it
37192 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37194 PR tree-optimization/79663
37195 * tree-predcom.c (combine_chains): Process refs in reverse order
37196 only for ZERO length chains, and add explaining comment.
37198 2017-02-23 Jeff Law <law@redhat.com>
37200 PR tree-optimization/79578
37201 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37202 in call to operand_equal_p.
37204 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37207 * config/i386/cpuid.h: Fix another undefined behavior.
37209 2017-02-23 Richard Biener <rguenther@suse.de>
37211 PR tree-optimization/79683
37212 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37213 vector types for data-refs.
37215 2017-02-23 Martin Liska <mliska@suse.cz>
37217 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37219 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37221 PR middle-end/79665
37222 * internal-fn.c (get_range_pos_neg): Moved to ...
37223 * tree.c (get_range_pos_neg): ... here. No longer static.
37224 * tree.h (get_range_pos_neg): New prototype.
37225 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37226 are known to be in between 0 and signed maximum inclusive, try to
37227 expand both unsigned and signed divmod and use the cheaper one from
37230 2017-02-22 Jeff Law <law@redhat.com>
37232 PR tree-optimization/79578
37233 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37234 to compare base operands.
37236 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37239 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37240 gpc_reg_operand instead of fpr_reg_operand.
37242 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37244 * config/mips/mips.c (mips_return_in_memory): Force FP
37245 vector types to be returned in memory for o32 ABI.
37247 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37249 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37250 instead of DW_TAG_member for static data member declarations and don't
37251 set no_linkage_name for static inline data members.
37252 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37255 2017-02-22 Martin Liska <mliska@suse.cz>
37257 * doc/invoke.texi: Replace inequality signs with square brackets
37260 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37263 * lra-constraints.c (simplify_operand_subreg): Handle
37264 WORD_REGISTER_OPERATIONS targets.
37266 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37269 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37270 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37271 elimination by swapping fld*.
37273 2017-02-22 Richard Biener <rguenther@suse.de>
37275 PR tree-optimization/79673
37276 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37277 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37278 irrelevant address-space qualifiers and avoiding a
37279 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37281 2017-02-22 Richard Biener <rguenther@suse.de>
37283 PR tree-optimization/79666
37284 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37285 to not symbolically negate if that may introduce undefined
37288 2017-02-22 Martin Liska <mliska@suse.cz>
37291 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37292 * data-streamer-out.c (streamer_write_gcov_count_stream):
37294 * value-prof.c (stream_out_histogram_value): Make assert more
37295 precise based on type of counter.
37297 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37300 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37301 Use nonimmediate_operand instead of memory_operand for operand 1.
37302 (float-extend standard_x87sse_constant_load splitter): Ditto.
37304 2017-02-21 Jeff Law <law@redhat.com>
37306 PR tree-optimization/79621
37307 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37308 blocks with edges to themselves.
37310 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37313 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37314 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37315 Use gimple_call_builtin_p.
37318 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37319 on temporarily removed DEBUG_INSNs.
37321 PR tree-optimization/79649
37322 * tree-loop-distribution.c (classify_partition): Give up on
37323 non-generic address space loads/stores.
37325 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37327 * doc/loop.texi (Loop manipulation): Remove nonexistent
37328 tree_ssa_loop_version from the documentation.
37329 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37331 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37334 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37335 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37336 * config/rs6000/rs6000.c: Include except.h.
37337 (rs6000_expand_split_stack_prologue): Call
37338 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37340 2017-02-21 Martin Jambor <mjambor@suse.cz>
37343 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37344 have been analyzed.
37346 2017-02-21 Martin Jambor <mjambor@suse.cz>
37348 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37349 for backward compatibility only.
37350 * doc/invoke.texi (Option Summary): Remove all references to
37351 -fipa-cp-alignment.
37353 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37357 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37359 * lra-constraints.c (curr_insn_transform): Handle
37360 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37362 2017-02-21 Martin Liska <mliska@suse.cz>
37364 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37366 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37368 PR translation/79638
37369 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37371 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37374 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37375 (arm_function_ok_for_sibcall): Return false for an indirect call by
37376 descriptor if all the argument registers are used.
37377 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37378 alignment of the function.
37380 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37382 PR tree-optimization/61441
37383 * simplify-rtx.c (simplify_const_unary_operation): For
37384 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37385 the sNaN unmodified.
37387 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37389 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37390 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37391 instead of SYSTEM_HEADER_DIR.
37393 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37394 Martin Liška <mliska@suse.cz>
37396 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37397 Fix typos and grammar, use active voice, and clarify.
37399 2017-02-20 Marek Polacek <polacek@redhat.com>
37401 PR middle-end/79537
37402 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37405 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37407 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37410 * config/i386/i386.c (ix86_expand_builtin): Handle
37411 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37412 ix86_builtins_isa[fcode].isa as a requirement of those
37413 flags and any other flag in the bitmask.
37414 (ix86_init_mmx_sse_builtins): Use 0 instead of
37415 ~OPTION_MASK_ISA_64BIT as mask.
37416 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37417 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37418 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37419 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37421 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37424 * lra-constraints.c (split_reg): Check requested split mode
37425 is supported by the register.
37427 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37429 * lra-constraints.c (simplify_operand_subreg): Remove early
37432 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37435 * lra-constraints.c (curr_insn_transform): Tighten condition
37436 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37438 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37441 * lra-constraints.c (curr_insn_transform): Handle
37442 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37444 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37447 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37449 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37451 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37454 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37457 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37459 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37462 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37464 * final.c (last_columnnum, override_columnnum): New variables.
37465 (final_start_function): Set last_columnnum, pass it to begin_prologue
37466 hook and pass 0 to dwarf2out_begin_prologue.
37467 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37468 to source_line debug hook.
37469 (notice_source_line): Compute last_columnnum and for debug_column_info
37470 return true on column changes.
37471 * debug.h (struct gcc_debug_hooks): Add column argument to
37472 source_line and begin_prologue hooks.
37473 (debug_nothing_int_charstar_int_bool): Remove prototype.
37474 (debug_nothing_int_int_charstar,
37475 debug_nothing_int_int_charstar_int_bool): New prototypes.
37476 (dwarf2out_begin_prologue): Add column argument.
37477 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37478 begin_prologue hooks.
37479 (debug_nothing_int_charstar_int_bool): Remove.
37480 (debug_nothing_int_int_charstar,
37481 debug_nothing_int_int_charstar_int_bool): New functions.
37482 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37483 through to dwarf2out_source_line.
37484 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37485 (dwarf2out_source_line): Add column argument, emit it if requested.
37486 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37488 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37489 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37490 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37491 through to dwarf2out_begin_prologue.
37492 (vmsdbgout_source_line): Add column argument, pass it through to
37493 dwarf2out_source_line.
37494 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37495 dbxout_source_line caller.
37496 (dbxout_source_line): Add column argument.
37498 * common.opt (gno-column-info, gcolumn-info): New options.
37499 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37500 (check_die): Also test for multiple DW_AT_decl_column attributes.
37501 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37502 DW_AT_decl_column if requested.
37503 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37505 (gen_variable_die): Likewise.
37506 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37507 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37510 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37511 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37512 (ix86_handle_option): Handle OPT_m3dnowa.
37513 * doc/invoke.texi (-m3dnowa): Document.
37514 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37515 -m3dnowa instead of -m3dnow -march=athlon.
37518 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37519 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37521 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37524 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37525 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37526 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37527 generator for vsx_xxpermdi_<mode>_be.
37528 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37529 force big-endian semantics.
37530 (vsx_xxpermdi_<mode>_be): New define_expand with same
37531 implementation as previous version of vsx_xxpermdi_<mode>.
37533 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37535 PR tree-optimization/79327
37536 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37537 variable, its initialization and use.
37539 2017-02-17 Julia Koval <julia.koval@intel.com>
37541 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37542 (OPTION_MASK_ISA_PKU_UNSET): New.
37543 (ix86_handle_option): Handle -mrdpid.
37544 * config/i386/cpuid.h (bit_RDPID): New.
37545 * config/i386/driver-i386.c (host_detect_local_cpu):
37546 Detect RDPID feature.
37547 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37548 * config/i386/i386-c.c (ix86_target_macros_internal):
37550 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37551 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37552 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37553 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37554 * config/i386/i386.md (define_insn "rdpid"): New.
37555 * config/i386/i386.opt Add -mrdpid.
37556 * config/i386/immintrin.h (_rdpid_u32): New.
37558 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37560 PR rtl-optimization/79541
37561 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37562 instead of transforming it into USE.
37564 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37566 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37567 If HONOR_SNANS (SFmode) force the input to a register.
37568 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37569 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37570 an frsp or similar insn.
37572 2017-02-17 Martin Liska <mliska@suse.cz>
37574 PR rtl-optimization/79577
37575 * params.def (selsched-max-sched-times): Increase minimum to 1.
37577 2017-02-17 Martin Liska <mliska@suse.cz>
37579 PR rtl-optimization/79574
37580 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37582 2017-02-17 Martin Liska <mliska@suse.cz>
37584 PR tree-optimization/79529
37585 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37586 ssa_defined_default_def_p to handle cases which are implicitly
37588 * tree-ssa.c (ssa_defined_default_def_p): New function.
37589 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37590 which are implicitly defined.
37591 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37593 2017-02-17 Richard Biener <rguenther@suse.de>
37595 PR middle-end/79576
37596 * params.def (max-ssa-name-query-depth): Limit to 10.
37598 2017-02-17 Richard Biener <rguenther@suse.de>
37600 PR tree-optimization/79552
37601 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37604 2017-02-17 Richard Biener <rguenther@suse.de>
37607 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37609 2017-02-17 Marek Polacek <polacek@redhat.com>
37611 PR middle-end/79536
37612 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37613 (fold_negate_expr): New wrapper.
37615 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37617 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37618 Correct terminology and de-emphasize pre-standard behavior.
37620 2017-02-16 Alan Modra <amodra@gmail.com>
37622 PR rtl-optimization/79286
37623 * ira.c (def_dominates_uses): New function.
37624 (update_equiv_regs): Don't create an equivalence for insns that
37625 may trap where the register def does not dominate the use.
37627 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37629 PR rtl-optimization/78127
37630 * lra.c (lra): Call lra_eliminate before finish the loop after
37633 2017-02-16 Richard Biener <rguenther@suse.de>
37635 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37637 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37638 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37639 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37640 (isl_val_int_from_wi): New function.
37641 (extract_affine_gmp): Rename to ...
37642 (extract_affine_wi): ... this, take a widest_int.
37643 (extract_affine_int): Just wrap extract_affine_wi.
37644 (add_param_constraints): Use isl_val_int_from_wi.
37645 (add_loop_constraints): Likewise, and extract_affine_wi.
37647 2017-02-15 Jeff Law <law@redhat.com>
37649 PR middle-end/79521
37650 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37651 ira_init_register_move_cost_if_necessary.
37653 2017-02-15 Martin Sebor <msebor@redhat.com>
37655 PR middle-end/32003
37656 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37657 removed in a prior commit.
37659 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37661 PR tree-optimization/79347
37662 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37663 counters during peeling.
37665 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37667 * Makefile.in (site.exp): Remove "set ISLVER".
37669 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37672 * real.c (real_from_integer): Call real_convert even for decimal.
37674 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37677 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37679 2017-02-14 Andrew Pinski <apinski@cavium.com>
37681 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37682 cores and change the partno/implementer to be correct.
37683 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37684 the 'B" as the implementer.
37685 * config/aarch64/aarch64-tune.md: Regenerate.
37687 2017-02-14 Carl Love <cel@us.ibm.com>
37689 * config/rs6000/rs6000.c: Add case statement entry to make the
37690 xvcvuxdsp built-in argument unsigned.
37691 * config/rs6000/vsx.md: Fix the source and return operand types so they
37692 match the instruction definitions from the ISA document. Fix typo
37693 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37696 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37699 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37700 member early_clobber_alts.
37701 * lra-lives.c (reg_early_clobber_p): New.
37702 (process_bb_lives): Use it.
37703 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37704 (debug_operand_data): Initialize early_clobber_alts.
37705 (setup_operand_alternative): Set up early_clobber_alts.
37706 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37707 alternatives to new_insn_reg.
37708 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37710 (lra_update_insn_regno_info): Pass the new arg.
37712 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37714 PR middle-end/79505
37715 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37716 (new_oacc_loop_raw): Don't clear already cleared fields.
37719 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37720 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37721 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37723 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37726 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37728 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37731 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37732 the extra instruction to the right place to store 128-bit constant
37735 2017-02-14 Martin Sebor <msebor@redhat.com>
37737 PR middle-end/79448
37738 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37739 warning for strings of unknown length.
37741 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37743 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37745 2017-02-14 Jeff Law <law@redhat.com>
37748 * ira-costs.c (scan_one_insn): Initialize register move costs
37749 for pseudos seen in USE/CLOBBER insns.
37751 PR tree-optimization/79095
37752 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37753 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37754 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37755 if the operands are known to be not equal, then the resulting range
37757 (intersect_ranges): If the new range is ~[0,0] and the old range is
37758 wide, then prefer ~[0,0].
37759 * tree-vrp.c (overflow_comparison_p_1): New function.
37760 (overflow_comparison_p): New function.
37761 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37762 if NAME is used in an overflow test.
37763 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37764 overflow check that can be expressed as an equality test, then adjust
37765 ops to be that equality test.
37767 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37769 * config/s390/s390-builtin-types.def: Remove flags argument.
37770 * config/s390/s390.c (s390_init_builtins): Likewise.
37772 2017-02-14 Martin Liska <mliska@suse.cz>
37774 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37775 vector. Fix trailing white spaces.
37777 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37779 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37782 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37784 PR rtl-optimization/68664
37785 * config/arm/arm.c (arm_sched_can_speculate_insn):
37786 New function. Declare prototype.
37787 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37789 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37791 PR rtl-optimization/68664
37792 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37794 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37796 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37798 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37799 max skip bytes for function, loop and jump.
37801 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37803 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37804 ABS_EXPR for gimple dump.
37806 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37809 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37811 PR tree-optimization/79408
37812 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37813 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37814 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37815 also if rhs1 is INTEGER_CST.
37817 2017-02-14 Richard Biener <rguenther@suse.de>
37819 PR middle-end/79432
37820 * tree-into-ssa.c (insert_phi_nodes): When the function can
37821 have abnormal edges rewrite SSA names with broken use-def
37822 dominance out of SSA and register them for PHI insertion.
37824 2017-02-13 Martin Sebor <msebor@redhat.com>
37826 PR middle-end/79496
37827 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37828 clearing info.nowrite flag when snprintf size argument is a range.
37830 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37832 * cprop.c (cprop_jump): Add missing space in string literal.
37833 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37834 (get_constraint_for_component_ref): Likewise.
37835 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37836 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37837 * lra-constraints.c (process_alt_operands): Likewise.
37838 * ipa-inline.c (inline_small_functions): Likewise.
37839 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37840 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37841 * trans-mem.c (diagnose_tm_1_op): Likewise.
37842 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37843 (grid_parallel_clauses_gridifiable): Likewise.
37845 * config/nvptx/mkoffload.c (process): Add space in between
37848 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37849 "MOD4_SSE_REGS" and "ALL_REGS".
37851 * spellcheck.c (test_data): Add , in between "foo" and "food".
37853 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37856 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37857 boundary crossing check and subsequent code generation agree.
37859 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37861 * config/aarch64/aarch64.c (has_memory_op): Delete.
37862 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37865 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37867 PR rtl-optimization/79388
37868 PR rtl-optimization/79450
37869 * combine.c (distribute_notes): When removing TEM_INSN for which
37870 corresponding dest has last value recorded, invalidate that last
37873 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37875 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37876 of explicit '@'. Add missing assembly comment marker on branch costs
37879 2017-02-13 Nathan Sidwell <nathan@acm.org>
37881 * gengtype-lex.l (<in_struct>): Add '/'.
37883 2017-02-13 Martin Liska <mliska@suse.cz>
37886 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37888 2017-02-13 Richard Biener <rguenther@suse.de>
37890 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37892 * configure: Re-generate.
37893 * config.in: Likewise.
37894 * graphite-dependences.c: Simplify as if
37895 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37896 * graphite-isl-ast-to-gimple.c: Likewise.
37897 * graphite-optimize-isl.c: Likewise.
37898 * graphite-poly.c: Likewise.
37899 * graphite-sese-to-poly.c: Likewise.
37900 * graphite.h: Likewise.
37901 * toplev.c: Include isl/version.h and use isl_version () for
37902 printing the ISL version.
37903 * doc/install.texi: Update ISL requirement.
37905 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37907 * doc/standards.texi (Standards): Update reference to
37910 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37912 * doc/extend.texi (Named Address Spaces): sourceware.org now
37914 * doc/install.texi (Binaries): Ditto.
37917 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37919 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37920 standard terminology "stringize"/"stringizing" throughout.
37921 * doc/cppinternals.texi: Likewise.
37923 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37925 * doc/extend.texi: Fix some spelling mistakes and typos.
37926 * doc/invoke.texi: Likewise.
37928 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37931 * params.def (inline-min-speedup) Change from 10 to 8.
37933 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37935 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37938 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37941 * ipa-inline-analysis.c (get_minimal_bb): New function.
37942 (record_modified): Use it.
37943 (remap_edge_change_prob): Handle also ancestor functions.
37945 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37947 * doc/contrib.texi (Contributors): Remove broken link into
37948 the Mauve CVS repository.
37950 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37952 PR middle-end/79454
37953 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37954 result computation whenever lhs doesn't have vector mode, not
37955 just when it has BLKmode.
37957 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37959 * doc/makefile.texi (profiledbootstrap): Refer to the
37960 installation instructions only in textual form.
37962 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37965 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37967 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37969 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37970 (Specific): Update mingw-w64 reference.
37972 (Specific): Remove broken link to Renesas RX processor.
37974 2017-02-10 Richard Biener <rguenther@suse.de>
37976 * toplev.c (process_options): Do not mention obsolete graphite
37977 options when printing sorry message about missing graphite support.
37978 Mention -floop-nest-optimize.
37980 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37982 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37983 (vtst_p16): Likewise.
37984 (vtstq_p8): Likewise.
37985 (vtstq_p16): Likewise.
37987 (vtstq_p64): Likewise.
37988 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37989 (vset_lane_p64): New.
37990 (vsetq_lane_p64): New.
37992 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37994 PR tree-optimization/79411
37995 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37996 stmt operands are SSA_NAMEs used in abnormal phis.
37997 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38000 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38003 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38006 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38008 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38010 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38012 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38016 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38017 not arbitrary TREE_CONSTANT.
38020 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38021 "omp declare target link" attribute unless is_global_var.
38022 * omp-offload.c (find_link_var_op): Likewise.
38024 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38025 Chung-Lin Tang <cltang@codesourcery.com>
38027 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38029 (gimplify_adjust_omp_clauses): Don't delete TILE.
38030 (gimplify_omp_for): Deal with TILE.
38031 * internal-fn.c (expand_GOACC_TILE): New function.
38032 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38034 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38035 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38037 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38038 avoid DIV for outermost collapse var.
38039 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38040 Remove out of date comments, fix whitespace.
38041 * omp-general.c (omp_extract_for_data): Deal with tiling.
38042 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38043 adjust OLF_DIM_BASE value.
38044 (struct omp_for_data): Add tiling field.
38045 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38046 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38047 for auto loops. Remove default auto determining, moved to
38048 oacc_loop_fixed_partitions.
38049 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38050 stmts, add e_mask field.
38051 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38052 (oacc_thread_numbers): Use oacc_dim_call.
38053 (oacc_xform_tile): New.
38054 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38055 (finish_oacc_loop): Adjust for ifns vector.
38056 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38057 add case for GOACC_TILE fns.
38058 (oacc_loop_xform_loop): Delete.
38059 (oacc_loop_process): Iterate over call list directly, and add
38060 handling for GOACC_TILE fns.
38061 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38063 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38064 vector partitioning to outer loops. Assign 2 partitions to loops
38065 when available. Add TILE handling.
38066 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38067 (execite_oacc_device_lower): Process GOACC_TILE fns,
38068 ignore unknown specs.
38069 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38070 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38071 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38073 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38075 * configure.ac (ACX_BUGURL): Update.
38076 * configure: Regenerate.
38078 2017-02-09 Richard Biener <rguenther@suse.de>
38080 PR tree-optimization/69823
38081 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38082 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38084 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38086 * config/arc/arc-c.def: Add __NPS400__ definition.
38087 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38088 (TARGET_NPS400): Define.
38090 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38092 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38094 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38095 pointer, arch_info.
38096 (arc_cpu_types): Fill the arch_info field with a pointer into the
38097 arc_arch_types table.
38098 (arc_selected_cpu): Declare.
38099 * config/arc/arc.c (arc_selected_cpu): Make global.
38100 (arc_selected_arch): Delete.
38101 (arc_base_cpu): Delete.
38102 (arc_override_options): Remove references to deleted variables,
38103 update access to arch information.
38104 (ARC_OPT): Update access to arch information.
38105 (ARC_OPTX): Likewise.
38106 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38107 (TARGET_ARC600): Update access to arch information.
38108 (TARGET_ARC601): Likewise.
38109 (TARGET_ARC700): Likewise.
38110 (TARGET_EM): Likewise.
38111 (TARGET_HS): Likewise.
38112 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38115 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38118 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38119 condition/operands for integer GE/LE/GEU/LEU operations.
38121 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38123 PR translation/79397
38124 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38127 2017-02-08 Martin Jambor <mjambor@suse.cz>
38130 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38131 whether allocation happened.
38132 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38133 nothing was allocated.
38135 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38137 PR tree-optimization/79408
38138 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38139 constant, but SSA_NAME with a known integer range, use the minimum
38140 of that range instead of op1 to determine if modulo can be replaced
38141 with its first operand.
38143 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38145 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38147 2017-02-08 Richard Biener <rguenther@suse.de>
38149 PR tree-optimization/71824
38150 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38151 Check all loops contained in the merged region.
38153 2017-02-07 Andrew Pinski <apinski@cavium.com>
38155 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38157 2017-02-07 Andrew Pinski <apinski@cavium.com>
38159 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38160 (thunderxt88): Likewise.
38161 (thunderxt81): Disable LSE and change v8.1 to v8.
38162 (thunderxt83): Likewise.
38164 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38165 Richard Biener <rguenther@suse.de>
38167 PR middle-end/79399
38168 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38169 type from int to size_t.
38170 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38172 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38174 PR rtl-optimization/79386
38175 * cprop.c (bypass_conditional_jumps): Initialize
38176 bypass_last_basic_block already before splitting bbs after
38177 unconditional traps...
38178 (bypass_conditional_jumps): ... rather than here.
38181 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38182 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38183 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38184 fix -masm=intel patterns.
38186 2017-02-07 Richard Biener <rguenther@suse.de>
38188 PR tree-optimization/79256
38189 PR middle-end/79278
38190 * builtins.c (get_object_alignment_2): Use min_align_of_type
38191 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38192 and ADJUST_FIELD_ALIGN.
38194 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38196 * doc/tm.texi: Regenerate.
38197 * stor-layout.c (layout_decl): Adjust.
38198 (update_alignment_for_field): Likewise.
38199 (place_field): Likewise.
38200 (min_align_of_type): Likewise.
38201 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38202 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38203 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38204 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38205 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38206 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38207 * config/i386/i386.c (x86_field_alignment): Likewise.
38208 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38209 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38210 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38211 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38212 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38213 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38217 2017-01-30 Richard Biener <rguenther@suse.de>
38219 PR tree-optimization/79256
38220 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38221 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38224 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38226 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38227 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38228 builtins to SImode and emit a zero-extend, if necessary.
38230 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38232 * docs/invoke.texi (RISC-V Options): Alphabetize.
38234 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38236 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38239 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38241 * config/riscv/riscv.c: New file.
38242 * common/config/riscv/riscv-common.c: Likewise.
38243 * config.gcc: Likewise.
38244 * config/riscv/constraints.md: Likewise.
38245 * config/riscv/elf.h: Likewise.
38246 * config/riscv/generic.md: Likewise.
38247 * config/riscv/linux.h: Likewise.
38248 * config/riscv/multilib-generator: Likewise.
38249 * config/riscv/peephole.md: Likewise.
38250 * config/riscv/pic.md: Likewise.
38251 * config/riscv/predicates.md: Likewise.
38252 * config/riscv/riscv-builtins.c: Likewise.
38253 * config/riscv/riscv-c.c: Likewise.
38254 * config/riscv/riscv-ftypes.def: Likewise.
38255 * config/riscv/riscv-modes.def: Likewise.
38256 * config/riscv/riscv-opts.h: Likewise.
38257 * config/riscv/riscv-protos.h: Likewise.
38258 * config/riscv/riscv.h: Likewise.
38259 * config/riscv/riscv.md: Likewise.
38260 * config/riscv/riscv.opt: Likewise.
38261 * config/riscv/sync.md: Likewise.
38262 * config/riscv/t-elf-multilib: Likewise.
38263 * config/riscv/t-linux: Likewise.
38264 * config/riscv/t-linux-multilib: Likewise.
38265 * config/riscv/t-riscv: Likewise.
38266 * configure.ac: Likewise.
38267 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38268 Waterman as RISC-V maintainers.
38269 * doc/install.texi: Add RISC-V entries.
38270 * doc/invoke.texi: Add RISC-V options section.
38271 * doc/md.texi: Add RISC-V constraints section.
38272 * configure: Regenerated.
38274 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38277 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38278 false values to be constant vectors with all 0 or all 1 bits set.
38279 (vcondu<mode><mode>): Likewise.
38280 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38282 (fpmask_comparison_operator): Update comment.
38283 (vecint_comparison_operator): New predicate.
38284 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38285 vector conditionals when the true and false values are constant
38286 vectors with all 0 bits or all 1 bits set.
38288 2017-02-06 Martin Sebor <msebor@redhat.com>
38290 PR tree-optimization/79376
38291 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38293 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38295 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38296 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38297 to simplify split condition.
38299 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38301 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38302 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38305 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38307 PR rtl-optimization/68664
38308 * target.def (can_speculate_insn): New hook.
38309 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38310 * doc/tm.texi: Regenerate.
38311 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38312 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38313 (rs6000_sched_can_speculate_insn): New function.
38315 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38317 PR tree-optimization/79284
38318 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38319 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38320 vectorizable_mask_load_store, vectorizable_operation,
38321 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38322 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38323 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38324 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38325 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38326 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38327 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38328 is_gimple_assign (stmt). Replace another such test with
38329 is_gimple_assign (stmt).
38331 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38334 * config/avr/avr.c (rtl-iter.h): Include it.
38335 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38336 (avr_legitimate_combined_insn): ...and implementation.
38338 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38340 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38341 * config/s390/s390.c (s390_const_operand_ok)
38342 (s390_canonicalize_comparison, s390_extract_part)
38343 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38344 (s390_contiguous_bitmask_p, s390_rtx_costs)
38345 (legitimize_pic_address): Likewise.
38346 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38347 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38348 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38349 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38350 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38352 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38354 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38355 REGNO($0) == REGNO($1).
38357 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38359 * config/s390/linux.h(SIZE_TYPE): Add comment.
38361 2017-02-06 Julian Brown <julian@codesourcery.com>
38362 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38363 Virendra Pathak <virendra.pathak@broadcom.com>
38365 * config/aarch64/aarch64-cores.def: Change the scheduler
38367 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38368 * config/aarch64/thunderx2t99.md: New file.
38370 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38372 * doc/standards.texi (Go Language): Update link to language
38375 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38377 * tree-eh.c (lower_resx): Sanitize profile.
38378 (cleanup_empty_eh_move_lp): Likewise.
38380 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38383 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38385 * cfgloopmanip.h (loop_version): Update prototype.
38386 * modulo-sched.c (sms_schedule): Update call of loop_version.
38387 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38388 * tree-parloops.c (gen_parallel_loop): Likewise.
38389 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38390 * tree-ssa-loop-split.c (split_loop): Likewise.
38391 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38392 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38394 2017-02-05 Martin Liska <mliska@suse.cz>
38397 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38399 (print_operand_address): Initialize a struct to zero.
38401 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38403 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38404 garbage collector only in textual form.
38406 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38408 * doc/extend.texi (x86 specific memory model extensions for
38409 transactional memory): Simplify a phrase.
38411 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38414 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38415 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38416 (atomic_storedi_1): Likewise.
38418 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38420 PR tree-optimization/79338
38421 * tree-parloops.c (gather_scalar_reductions): Don't call
38422 vect_analyze_loop_form for loop->inner before destroying loop's
38425 2017-02-03 Martin Sebor <msebor@redhat.com>
38427 PR tree-optimization/79327
38428 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38429 when precision has resulted in leading zeros.
38430 (format_integer): Adjust the likely counter to assume an unknown
38431 argument that may be zero is non-zero.
38433 2017-02-03 Jason Merrill <jason@redhat.com>
38436 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38437 avoid copying non-taken branch.
38439 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38441 PR tree-optimization/79340
38442 * tree-vect-loop.c (vectorizable_reduction): Release
38443 vec_defs elements after safe_splicing them into other vectors.
38446 PR tree-optimization/79327
38447 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38448 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38450 (format_integer): Use wide_int_to_tree instead of build_int_cst
38451 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38452 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38453 of shortest and longest sequence.
38455 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38457 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38458 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38460 2017-02-03 Walter Lee <walt@tilera.com>
38463 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38464 after initial stackframe link reg save.
38465 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38467 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38470 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38471 wu for stxssp alternative.
38473 2017-02-03 Martin Sebor <msebor@redhat.com>
38475 PR tree-optimization/79352
38476 * gimple-fold.c (get_range_strlen): Add argument.
38477 (get_range_strlen): Change return type to bool.
38478 (get_maxval_strlen): Pass in a dummy argument.
38479 * gimple-fold.h (get_range_strlen): Change return type to bool.
38480 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38481 * tree.h (array_at_struct_end_p): Add argument.
38482 * tree.c (array_at_struct_end_p): Handle it.
38484 2017-02-03 Martin Liska <mliska@suse.cz>
38487 * multiple_target.c (create_dispatcher_calls): Redirect edge
38488 from a caller of a dispatcher.
38489 (expand_target_clones): Make the clones local.
38490 (ipa_target_clone): Do both target clones and resolvers.
38491 (ipa_dispatcher_calls): Remove the pass.
38492 (pass_dispatcher_calls::gate): Likewise.
38493 (make_pass_dispatcher_calls): Likewise.
38494 * passes.def (pass_target_clone): Put as very first IPA early
38497 2017-02-03 Martin Liska <mliska@suse.cz>
38499 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38500 in case of a function with ifunc attribute.
38502 2017-02-03 Martin Liska <mliska@suse.cz>
38504 * cgraph.c (cgraph_node::dump): Dump function version info.
38505 * symtab.c (symtab_node::dump_base): Add missing new line.
38507 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38509 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38510 (ifcombine_ifandif): Use it.
38512 2017-02-03 Martin Liska <mliska@suse.cz>
38514 * doc/invoke.texi: Document default value for
38515 use-after-scope-direct-emission-threshold.
38517 2017-02-03 Martin Liska <mliska@suse.cz>
38519 PR tree-optimization/79339
38520 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38521 (format_floating): Likewise.
38523 2017-02-03 Martin Liska <mliska@suse.cz>
38526 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38528 (ipa_node_params_t::remove): Likewise.
38529 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38530 initialization from removed ipa_node_params_t::insert.
38531 (ipa_node_params::~ipa_node_params): Move from removed
38532 ipa_node_params_t::release.
38533 * symbol-summary.h (symbol_summary::m_released): New member.
38534 Do not release a summary twice. Do not allow to call finalizer
38535 for types of a summary that live in GGC memory.
38537 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38539 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38542 2017-02-02 Martin Sebor <msebor@redhat.com>
38544 PR middle-end/79275
38545 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38546 (format_string): Tighten up the range of output for non-constant
38547 strings and correct the expected range for wide non-constant strings.
38549 2017-02-02 Martin Sebor <msebor@redhat.com>
38551 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38553 PR middle-end/32003
38554 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38556 (-fdump-tree-@var): Add to index and document how to come up
38557 with pass-specific option and dump file names.
38558 (-fdump-passes): Clarify where to look for output.
38560 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38562 PR middle-end/77445
38563 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38564 statistics of the analyzed path; allow threading for speed when
38565 any of BBs along the path are optimized for speed.
38567 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38569 PR middle-end/78468
38570 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38571 settings of the virtual registers.
38574 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38576 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38577 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38580 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38582 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38583 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38585 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38587 * config/s390/s390.md: Add missing comments with the expanded
38589 * config/s390/vector.md: Likewise.
38590 * config/s390/vx-builtins.md: Likewise.
38592 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38595 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38596 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38597 conditions on a single line.
38599 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38601 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38602 __S390_VX__ to __VX__.
38604 2017-02-01 Andrew Pinski <apinski@cavium.com>
38606 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38607 stmt_info to record_stmt_cost.
38608 (vect_get_known_peeling_cost): Pass stmt_info if known to
38610 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38611 cpu_vector_cost field into
38612 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38613 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38614 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38615 splitting of scalar_stmt_cost and vec_stmt_cost.
38616 (thunderx_vector_cost): Likewise.
38617 (cortexa57_vector_cost): LIkewise.
38618 (exynosm1_vector_cost): Likewise.
38619 (xgene1_vector_cost): Likewise.
38620 (thunderx2t99_vector_cost): Improve after the splitting of the two
38622 (aarch64_builtin_vectorization_cost): Update for the splitting of
38623 scalar_stmt_cost and vec_stmt_cost.
38625 2017-02-01 Torvald Riegel <triegel@redhat.com>
38626 Richard Henderson <rth@redhat.com>
38628 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38629 conditional on existance of a fast atomic load.
38630 * optabs-query.c (can_atomic_load_p): New function.
38631 * optabs-query.h (can_atomic_load_p): Declare it.
38632 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38633 no fast atomic load is available for the particular size of access.
38634 (expand_atomic_compare_and_swap): Likewise.
38635 (expand_atomic_load): Likewise.
38636 (expand_atomic_store): Likewise.
38637 (expand_atomic_fetch_op): Likewise.
38638 * testsuite/lib/target-supports.exp
38639 (check_effective_target_sync_int_128): Remove x86 because it provides
38640 no fast atomic load.
38641 (check_effective_target_sync_int_128_runtime): Likewise.
38643 2017-02-01 Richard Biener <rguenther@suse.de>
38645 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38646 (graphite_transform_loops): Provide opt-info for optimized nests.
38647 * tree-parloop.c (parallelize_loops): Provide opt-info for
38648 parallelized loops.
38650 2017-02-01 Richard Biener <rguenther@suse.de>
38652 PR middle-end/79315
38653 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38654 was not set before.
38656 2017-02-01 Richard Biener <rguenther@suse.de>
38658 PR tree-optimization/71824
38659 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38660 Verify the loops are valid in the merged SESE region.
38661 (scop_detection::can_represent_loop_1): Check analyzing the
38662 evolution of the number of iterations in the region succeeds.
38664 2017-01-31 Ian Lance Taylor <iant@golang.org>
38666 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38667 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38669 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38671 PR preprocessor/79210
38672 * input.c (get_substring_ranges_for_loc): Replace line_width
38673 assertion with error-handling.
38675 2017-01-31 Richard Biener <rguenther@suse.de>
38677 PR tree-optimization/77318
38678 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38679 (create_pw_aff_from_tree): Take loop parameter.
38680 (add_condition_to_pbb): Pass loop of the condition to
38681 create_pw_aff_from_tree.
38683 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38685 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38686 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38688 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38692 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38694 (convert_int_to_float128): Likewise.
38695 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38696 (convert_int_to_float128): Likewise.
38697 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38698 (UNSPEC_IEEE128_CONVERT): Likewise.
38699 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38700 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38701 Use local variables for IBM extended format.
38702 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38703 (fix_trunc<mode>si2_fprs): Likewise.
38704 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38705 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38706 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38707 to know that we can now have integers of all sizes in vector
38709 (fix<uns>_<mode>di2_hw): Likewise.
38710 (float<uns>_<mode>si2_hw): Likewise.
38711 (fix_<mode>si2_hw): Likewise.
38712 (fixuns_<mode>si2_hw): Likewise.
38713 (float<uns>_<mode>di2_hw): Likewise.
38714 (float_<mode>di2_hw): Likewise.
38715 (float_<mode>si2_hw): Likewise.
38716 (floatuns_<mode>di2_hw): Likewise.
38717 (floatuns_<mode>si2_hw): Likewise.
38718 (xscvqp<su>wz_<mode>): Delete, no longer used.
38719 (xscvqp<su>dz_<mode>): Likewise.
38720 (xscv<su>dqp_<mode>): Likewise.
38721 (ieee128_mfvsrd_64bit): Likewise.
38722 (ieee128_mfvsrd_32bit): Likewise.
38723 (ieee128_mfvsrwz): Likewise.
38724 (ieee128_mtvsrw): Likewise.
38725 (ieee128_mtvsrd_64bit): Likewise.
38726 (ieee128_mtvsrd_32bit): Likewise.
38728 2017-01-31 Martin Liska <mliska@suse.cz>
38731 * ipa-prop.c (ipa_free_all_node_params): Call release method
38732 instead of ~sumbol_summary to not to trigger double times
38735 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38737 PR tree-optimization/71691
38738 * bitmap.h (class auto_bitmap): New.
38739 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38740 is_maybe_undefined instead of ssa_undefined_value_p.
38742 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38744 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38745 __S390_ARCH_LEVEL__ to __ARCH__.
38747 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38749 PR tree-optimization/79267
38750 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38751 if should_remove_lhs_p is true.
38753 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38756 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38757 (add_alignment_attribute): New.
38758 (base_type_die): Add alignment attribute.
38759 (subrange_type_die): Likewise.
38760 (modified_type_die): Likewise.
38761 (gen_array_type_die): Likewise.
38762 (gen_descr_array_type_die: Likewise.
38763 (gen_enumeration_type_die): Likewise.
38764 (gen_subprogram_die): Likewise.
38765 (gen_variable_die): Likewise.
38766 (gen_field_die): Likewise.
38767 (gen_ptr_to_mbr_type_die): Likewise.
38768 (gen_struct_or_union_type_die): Likewise.
38769 (gen_subroutine_type_die): Likewise.
38770 (gen_typedef_die): Likewise.
38771 (base_type_cmp): Compare alignment attribute.
38773 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38776 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38777 (setb_unsigned) New pattern for setb with CCUNS.
38778 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38779 subfc./subfe sequence to avoid overflow problems. Generate a
38780 shorter sequence with cmpld/setb for power9.
38781 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38782 for generating subfc. instruction.
38783 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38784 now uses this instruction.
38786 2017-01-30 Ian Lance Taylor <iant@google.com>
38789 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38790 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38792 2017-01-30 Martin Sebor <msebor@redhat.com>
38794 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38795 Move constant to the right of a relational operator.
38796 (get_mpfr_format_length, format_character, format_string): Ditto.
38797 (should_warn_p, maybe_warn): Same.
38799 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38801 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38804 * asan.c (get_translation_unit_decl): Remove function.
38805 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38807 2017-01-30 Martin Liska <mliska@suse.cz>
38809 PR gcov-profile/79259
38810 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38811 -fprofile-generate.
38813 2017-01-30 Martin Liska <mliska@suse.cz>
38816 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38817 Initialize variables with NULL value.
38819 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38822 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38824 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38826 2017-01-30 Richard Biener <rguenther@suse.de>
38828 PR tree-optimization/79276
38829 * tree-vrp.c (process_assert_insertions): Properly adjust common
38830 when removing a duplicate.
38832 2017-01-30 Richard Biener <rguenther@suse.de>
38834 PR tree-optimization/79256
38835 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38836 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38838 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38840 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38843 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38844 ("*r<noxa>sbg_<mode>_sll_bitmask")
38845 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38846 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38847 Use contiguous_bitmask_nowrap_operand.
38849 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38852 * config/rs6000/altivec.h (vec_xl): Revise #define.
38853 (vec_xst): Likewise.
38855 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38857 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38859 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38861 PR rtl-optimization/79194
38862 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38863 traps before call to bypass_conditional_jumps.
38865 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38867 PR tree-optimization/71374
38868 * lra-constraints.c (check_conflict_input_operands): New.
38869 (match_reload): Use it.
38871 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38874 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38875 account to calculate conflict_set.
38877 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38879 PR rtl-optimization/78559
38880 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38881 other_insn in combine.
38883 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38885 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38886 uint16_type_node for BT_UINT16.
38888 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38890 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38891 "RTL Tests" to menu.
38892 (GIMPLE Tests): New node.
38893 (RTL Tests): New node.
38895 2017-01-27 Richard Biener <rguenther@suse.de>
38897 PR tree-optimization/79245
38898 * tree-loop-distribution.c (distribute_loop): Apply cost
38899 modeling also to detected patterns.
38901 2017-01-27 Richard Biener <rguenther@suse.de>
38903 PR tree-optimization/71433
38904 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38905 (compare_assert_loc): New function.
38906 (process_assert_insertions): Sort and optimize assert locations
38907 to remove duplicates and push down identical assertions on
38908 edges to their destination block.
38910 2017-01-27 Richard Biener <rguenther@suse.de>
38912 PR tree-optimization/79244
38913 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38914 out SSA names even if abnormal.
38916 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38918 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38919 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38920 instead of MPFR_RNDN.
38922 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38925 * arm.c (arm_option_override): Don't call build_target_option_node
38926 until after doing all option overrides.
38927 (arm_valid_target_attribute_tree): Likewise.
38929 2017-01-27 Martin Liska <mliska@suse.cz>
38931 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38932 for {cd}tors and C++ {cd}tors.
38934 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38936 * config/s390/s390.md ("*setmem_long_and")
38937 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38939 2017-01-26 Martin Sebor <msebor@redhat.com>
38941 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38944 2017-01-26 Martin Sebor <msebor@redhat.com>
38946 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38947 HAVE_DFmode before using XFmode or DFmode.
38948 (parse_directive): Avoid using the z length modifier to avoid
38949 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38951 PR middle-end/78703
38952 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38953 to accept adjustment as an array.
38954 (get_int_range): New function.
38955 (struct directive): Make width and prec arrays.
38956 (directive::set_width, directive::set_precision): Call get_int_range.
38957 (format_integer, format_floating): Handle width and precision ranges.
38958 (format_string, parse_directive): Same.
38960 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38963 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38964 comdat_type_p set, just clone them, but keep the children in the
38968 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38969 which have direct callers with -fvar-tracking-assignments enabled
38971 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38972 inside of type units.
38974 2017-01-26 Martin Sebor <msebor@redhat.com>
38976 PR middle-end/78703
38977 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38979 (struct format_result): Replace number_chars, number_chars_min,
38980 and number_chars_max with a single member of struct result_range.
38982 (format_result::operator+=): Adjust.
38983 (struct fmtresult): Remove bounded. Handle likely and unlikely
38985 (fmtresult::adjust_for_width_or_precision): New function.
38986 (fmtresult:type_max_digits): New function.
38987 (bytes_remaining): Handle likely and unlikely counters.
38988 (min_bytes_remaining): Remove.
38989 (format_percent): Simplify.
38990 (format_integer, format_floating): Set likely and unlikely counters.
38991 (get_string_length, format_character, format_string): Same.
38992 (format_plain, should_warn_p): New function.
38993 (maybe_warn): Call should_warn_p. Update diagnostic messages
38994 and handle those for all directives, including plain strings.
38995 (format_directive): Handle likely and unlikely counters.
38996 Remove unnecessary quoting from diagnostics. Add an informational
38998 (add_bytes): Remove.
38999 (pass_sprintf_length::compute_format_length): Simplify.
39000 (try_substitute_return_value): Handle likely and unlikely counters.
39002 2017-01-26 Carl Love <cel@us.ibm.com>
39004 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39005 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39007 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39010 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39011 endianess for subregs into account.
39012 * lra-constraints.c (lra_constraints): Do risky transformations
39013 always on the first iteration.
39014 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39015 last_call_used_reg_set.
39016 (process_bb_lives): Define and use last_call_used_reg_set.
39017 * lra.c (lra): Always continue after lra_constraints on the first
39020 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39022 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39024 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39026 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39028 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39029 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39030 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39031 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39032 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39033 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39034 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39035 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39036 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39038 2017-01-26 Marek Polacek <polacek@redhat.com>
39041 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39042 for the third operand.
39044 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39046 PR middle-end/79236
39047 * omp-low.c (struct omp_context): Add simt_stmt field.
39048 (scan_omp_for): Return omp_context *.
39049 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39050 context to the _simt_ SIMD stmt.
39051 (lower_omp_for): For combined SIMD with sibling _simt_
39052 SIMD, make sure to use the same decls in _looptemp_
39053 clauses as in the sibling.
39055 2017-01-26 David Sherwood <david.sherwood@arm.com>
39057 PR middle-end/79212
39058 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39061 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39064 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39065 emit fld b; fld a; if possible.
39067 * brig-builtins.def: Update copyright years.
39068 * config/arm/arm_acle_builtins.def: Update copyright years.
39070 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39073 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39074 constraint instead of o for the stxsd instruction.
39076 2017-01-25 Carl Love <cel@us.ibm.com>
39078 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39079 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39081 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39083 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39085 2017-01-25 Richard Biener <rguenther@suse.de>
39087 PR tree-optimization/69264
39088 * target.def (vector_alignment_reachable): Improve documentation.
39089 * doc/tm.texi: Regenerate.
39090 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39092 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39093 earlier changes with respect to TYPE_USER_ALIGN.
39094 (vector_alignment_reachable_p): Likewise. Improve dumping.
39096 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39099 * config/arm/arm.md (xordi3): Force constant operand into a register
39102 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39104 * doc/invoke.texi (-fstore-merging): Correct default optimization
39105 levels at which it is enabled.
39106 (-O): Move -fstore-merging from list to...
39109 2017-01-25 Richard Biener <rguenther@suse.de>
39112 * omp-expand.c: Include debug.h.
39113 (expand_omp_taskreg): Make sure to generate early debug before
39114 outlining anything from a function.
39115 (expand_omp_target): Likewise.
39116 (grid_expand_target_grid_body): Likewise.
39118 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39121 * asan.c (get_translation_unit_decl): New function.
39122 (asan_add_global): Extract modules file name from globals
39123 TRANSLATION_UNIT_DECL name.
39125 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39128 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39129 for long calls with APCS frame and VFP.
39131 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39133 * cfg.c (original_copy_tables_initialized_p): New function.
39134 * cfg.h (original_copy_tables_initialized_p): New decl.
39135 * cfgrtl.c (relink_block_chain): Guard the call to
39136 free_original_copy_tables with a call to
39137 original_copy_tables_initialized_p.
39138 * cgraph.h (symtab_node::native_rtl_p): New decl.
39139 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39140 (symtab_node::needed_p): Don't assert for early assembly output
39141 for __RTL functions.
39142 (cgraph_node::finalize_function): Set "force_output" for __RTL
39144 (cgraph_node::analyze): Bail out early for __RTL functions.
39145 (analyze_functions): Update assertion to support __RTL functions.
39146 (cgraph_node::expand): Bail out early for __RTL functions.
39147 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39149 * function.h (struct function): Update comment for field
39151 * gimple-expr.c: Include "tree-pass.h".
39152 (gimple_has_body_p): Return false for __RTL functions.
39153 * Makefile.in (OBJS): Add run-rtl-passes.o.
39154 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39156 (gcc::pass_manager::get_clean_slate): New accessor.
39157 * passes.c: Include "insn-addr.h".
39158 (should_skip_pass_p): Add logging. Update logic for running
39159 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39160 property-provider override so it is only done for gimple passes.
39162 (skip_pass): New function.
39163 (execute_one_pass): Call skip_pass when skipping passes.
39164 * read-md.c (md_reader::read_char): Support filtering
39165 the input to a subset of line numbers.
39166 (md_reader::md_reader): Initialize fields
39167 m_first_line and m_last_line.
39168 (md_reader::read_file_fragment): New function.
39169 * read-md.h (md_reader::read_file_fragment): New decl.
39170 (md_reader::m_first_line): New field.
39171 (md_reader::m_last_line): New field.
39172 * read-rtl-function.c (function_reader::create_function): Only
39173 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39174 curr_properties. Set DECL_INITIAL to a dummy block.
39175 (read_rtl_function_body_from_file_range): New function.
39176 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39178 * run-rtl-passes.c: New file.
39179 * run-rtl-passes.h: New file.
39181 2017-01-24 Jeff Law <law@redhat.com>
39183 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39186 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39188 PR tree-optimization/79159
39189 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39190 (record_nonwrapping_iv): Improve boundary using above function if no
39191 value range information.
39193 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39194 Martin Jambor <mjambor@suse.cz>
39196 * brig-builtins.def: New file.
39197 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39198 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39199 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39200 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39201 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39202 * builtin-types.def (BT_INT8): New.
39203 (BT_INT16): Likewise.
39204 (BT_UINT8): Likewise.
39205 (BT_UINT16): Likewise.
39206 (BT_FN_ULONG): Likewise.
39207 (BT_FN_UINT_INT): Likewise.
39208 (BT_FN_UINT_ULONG): Likewise.
39209 (BT_FN_UINT_LONG): Likewise.
39210 (BT_FN_UINT_PTR): Likewise.
39211 (BT_FN_ULONG_PTR): Likewise.
39212 (BT_FN_INT8_FLOAT): Likewise.
39213 (BT_FN_INT16_FLOAT): Likewise.
39214 (BT_FN_UINT32_FLOAT): Likewise.
39215 (BT_FN_UINT16_FLOAT): Likewise.
39216 (BT_FN_UINT8_FLOAT): Likewise.
39217 (BT_FN_UINT64_FLOAT): Likewise.
39218 (BT_FN_UINT16_UINT32): Likewise.
39219 (BT_FN_UINT32_UINT16): Likewise.
39220 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39221 (BT_FN_INT_PTR_INT): Likewise.
39222 (BT_FN_UINT_PTR_UINT): Likewise.
39223 (BT_FN_LONG_PTR_LONG): Likewise.
39224 (BT_FN_ULONG_PTR_ULONG): Likewise.
39225 (BT_FN_VOID_UINT64_UINT64): Likewise.
39226 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39227 (BT_FN_INT8_INT8_INT8): Likewise.
39228 (BT_FN_INT16_INT16_INT16): Likewise.
39229 (BT_FN_INT_INT_INT): Likewise.
39230 (BT_FN_UINT_FLOAT_UINT): Likewise.
39231 (BT_FN_FLOAT_UINT_UINT): Likewise.
39232 (BT_FN_ULONG_UINT_UINT): Likewise.
39233 (BT_FN_ULONG_UINT_PTR): Likewise.
39234 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39235 (BT_FN_UINT_UINT_UINT): Likewise.
39236 (BT_FN_VOID_UINT_PTR): Likewise.
39237 (BT_FN_UINT_UINT_PTR: Likewise.
39238 (BT_FN_UINT32_UINT64_PTR): Likewise.
39239 (BT_FN_INT_INT_UINT_UINT): Likewise.
39240 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39241 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39242 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39243 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39244 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39245 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39246 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39247 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39248 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39249 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39250 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39251 * doc/frontends.texi: List BRIG FE.
39252 * doc/install.texi (Testing): Add BRIG tesring requirements.
39253 * doc/invoke.texi (Overall Options): Mention BRIG.
39254 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39256 2017-01-24 Richard Biener <rguenther@suse.de>
39258 PR translation/79208
39259 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39261 2017-01-24 Martin Jambor <mjambor@suse.cz>
39264 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39265 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39266 and known_contexts.
39268 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39270 PR middle-end/79123
39271 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39272 casts from signed to unsigned really don't have a range.
39274 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39276 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39277 GMP_RNDx for compatiblity.
39279 2017-01-24 Martin Liska <mliska@suse.cz>
39282 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39283 that would prevent us to call alloca with -1 as argument.
39285 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39287 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39288 Avoid -Wformat-security warning.
39290 2017-01-23 Andrew Pinski <apinski@cavium.com>
39292 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39295 2017-01-23 Martin Sebor <msebor@redhat.com>
39297 PR middle-end/78703
39298 * gimple-ssa-sprintf.c (warn_level): New global.
39299 (format_integer): Use it here and throughout the rest of the file.
39300 Use the same switch to compute sign as base.
39301 (maybe_warn): New function.
39302 (format_directive): Factor out warnings into maybe_warn.
39303 Add debugging output. Use warn_level.
39304 (add_bytes): Use warn_level.
39305 (pass_sprintf_length::compute_format_length): Add debugging output.
39306 (try_substitute_return_value): Same.
39307 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39309 PR middle-end/78703
39310 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39311 (struct fmtresult, format_integer, format_floating): Adjust.
39312 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39313 (get_string_length, format_string,format_directive): Same.
39314 (pass_sprintf_length::compute_format_length): Same.
39315 (try_substitute_return_value): Simplify slightly.
39317 PR middle-end/78703
39318 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39319 (fmtresult::operator+=): Outlined.
39320 (struct fmtresult): Add ctors.
39321 (struct conversion_spec): Rename...
39322 (struct directive): ...to this. Add and remove data members.
39323 (directive::set_width, directive::set_precision): New functions.
39324 (format_percent): Use fmtresult ctor.
39325 (get_width_and_precision): Remove.
39326 (format_integer): Make naming changes. Avoid computing width and
39328 (format_floating): Same. Adjust indentation.
39329 (format_character, format_none): New functions.
39330 (format_string): Moved character handling to format_character.
39331 (format_directive): Remove arguments, change return type.
39332 (parse_directive): New function.
39333 (pass_sprintf_length::compute_format_length): Move directive
39334 parsing to parse_directive.
39336 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39338 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39339 (assign_assembler_name_if_needed): ... this.
39340 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39341 (assign_assembler_name_if_needed): ... this.
39342 (free_lang_data_in_cgraph): Adjust callers.
39343 * cgraphunit.c (cgraph_node::analyze): Likewise.
39344 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39347 2017-01-23 Richard Biener <rguenther@suse.de>
39349 PR tree-optimization/79088
39350 PR tree-optimization/79188
39351 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39352 resetting loop bounds after last path deletion. Reset loop
39353 bounds of the target loop, make code match the comments.
39354 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39355 Make sure loops need no fixups.
39357 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39359 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39360 exponent support with double type for first argument.
39361 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39362 type returned by __builtin_vec_extract_sig,
39363 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39364 functions from "vector int" to "vector unsigned int" or from
39365 "vector long long int" to "vector unsigned long long int".
39366 Changed type returned by __builtin_vec_extract_exp,
39367 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39368 functions from "vector int" to "vector unsigned int" or from
39369 "vector long long int" to "vector unsigned long long int".
39370 Changed return type of __builtin_vec_test_data_class,
39371 __builtin_vec_test_data_class_sp, and
39372 __builtin_vec_test_data_class_dp from "vector int" to
39373 "vector bool int" or from "vector long long int" to "vector bool
39374 long long int" and changed second argument type from "unsigned
39375 int" to "int". Added new overloaded function forms "vector float
39376 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39377 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39378 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39379 double, vector unsigned long long int)" and "vector double
39380 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39381 long int)". Changed return type of
39382 __builtin_scalar_test_data_class and
39383 __builtin_scalar_test_data_class_sp and
39384 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39385 int" and changed second argument from "unsigned int" to "int".
39386 Changed type returned by __builtin_scalar_test_neg,
39387 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39388 from "int" to "bool int". Added new overloaded function form
39389 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39390 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39391 exponent double-precision with floating point first argument.
39392 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39393 documentation of scalar_test_data_class, scalar_test_neg,
39394 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39395 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39396 vec_test_data_class built-in functions to reflect refinements in
39397 their type signatures.
39399 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39401 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39403 (aarch64_elf_asm_destructor): Likewise.
39405 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39407 PR rtl-optimization/78634
39408 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39409 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39410 * ifcvt.c (noce_try_cmove): Add missing cost check.
39412 PR rtl-optimization/71724
39413 * combine.c (if_then_else_cond): Look for situations where it is
39414 beneficial to undo the work of one of the recursive calls.
39416 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39418 PR tree-optimization/70754
39419 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39420 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39421 combined stmt before it if not NULL.
39422 (combine_chains): Process refs reversely and compute dominance point
39425 2017-01-23 Martin Liska <mliska@suse.cz>
39427 PR tree-optimization/79196
39428 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39429 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39431 (strlen_optimize_stmt): Call the renamed function.
39433 2017-01-23 Michael Matz <matz@suse.de>
39435 PR tree-optimization/78384
39436 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39438 2017-01-23 Richard Biener <rguenther@suse.de>
39440 PR tree-optimization/79186
39441 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39442 both incoming edges before moving an assert.
39444 2017-01-23 Martin Jambor <mjambor@suse.cz>
39446 * ipa-prop.c (load_from_param_1): Removed.
39447 (load_from_unmodified_param): Bits from load_from_param_1 put back
39449 (load_from_param): Removed.
39450 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39451 with stmt. Reverted back to use of load_from_unmodified_param.
39453 2017-01-23 Martin Jambor <mjambor@suse.cz>
39456 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39457 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39458 field a pointer to garbage collected vector, mark lattices and
39459 ipcp_orig_node with GTY((skip)).
39460 (ipa_get_param_count): Adjust to descriptors being a pointer.
39461 (ipa_get_param): Likewise.
39462 (ipa_get_type): Likewise.
39463 (ipa_get_param_move_cost): Likewise.
39464 (ipa_set_param_used): Likewise.
39465 (ipa_get_controlled_uses): Likewise.
39466 (ipa_set_controlled_uses): Likewise.
39467 (ipa_is_param_used): Likewise.
39468 (ipa_node_params_t): Move into garbage collector. New methods insert
39470 (ipa_node_params_sum): Annotate wth GTY(()).
39471 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39473 (ipa_load_from_parm_agg): Adjust declaration.
39474 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39475 * ipa-profile.c (ipa_profile): Likewise.
39476 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39477 (ipa_populate_param_decls): Make descriptors parameter garbage
39479 (ipa_dump_param): Adjust to descriptors being a pointer.
39480 (ipa_alloc_node_params): Likewise.
39481 (ipa_initialize_node_params): Likewise.
39482 (load_from_param_1): Make descriptors parameter garbage collected.
39483 (load_from_unmodified_param): Likewise.
39484 (load_from_param): Likewise.
39485 (ipa_load_from_parm_agg): Likewise.
39486 (ipa_node_params::~ipa_node_params): Removed.
39487 (ipa_free_all_node_params): Remove call to delete operator.
39488 (ipa_node_params_t::insert): New.
39489 (ipa_node_params_t::remove): Likewise.
39490 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39491 copy known_csts and known_contexts vectors.
39492 (ipa_read_node_info): Adjust to descriptors being a pointer.
39493 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39495 (ipcp_transform_function): Make descriptors variable garbage
39498 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39500 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39501 * config/i386/avx512dqintrin.h: Ditto.
39502 * config/i386/avx512fintrin.h: Ditto.
39503 * config/i386/i386.c: Handle new builtins.
39504 * config/i386/i386-builtin.def: Add new builtins.
39505 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39506 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39508 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39509 Martin Liska <mliska@suse.cz>
39511 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39512 * asan.c (asan_expand_poison_ifn): Support stores and use
39513 appropriate ASAN report function.
39514 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39515 * internal-fn.def (ASAN_POISON_USE): Declare.
39516 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39517 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39518 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39519 ASAN_POISON calls w/o LHS.
39520 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39521 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39522 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39523 * gimplify.c (asan_poison_variables): Add attribute
39524 use_after_scope_memory to variables that really needs to live
39526 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39527 having the attribute.
39529 2017-01-23 Martin Liska <mliska@suse.cz>
39531 * asan.c (create_asan_shadow_var): New function.
39532 (asan_expand_poison_ifn): Likewise.
39533 * asan.h (asan_expand_poison_ifn): New declaration.
39534 * internal-fn.c (expand_ASAN_POISON): Likewise.
39535 * internal-fn.def (ASAN_POISON): New builtin.
39536 * sanopt.c (pass_sanopt::execute): Expand
39537 asan_expand_poison_ifn.
39538 * tree-inline.c (copy_decl_for_dup_finish): Make function
39540 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39541 * tree-ssa.c (is_asan_mark_p): New function.
39542 (execute_update_addresses_taken): Rewrite local variables
39543 (identified just by use-after-scope as addressable) into SSA.
39545 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39547 * doc/install.texi (Specific): opensource.apple.com uses https
39548 now. Remove trailing slash.
39550 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39552 * README.Portability: Remove note on an Irix compatibility issue.
39554 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39556 * gcov.c (INCLUDE_ALGORITHM): Define.
39557 (INCLUDE_VECTOR): Define.
39558 No longer include <vector> and <algorithm> directly.
39560 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39562 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39564 * doc/invoke.texi (Code Gen Options): Ditto.
39566 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39569 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39571 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39573 rtl-optimization/79125
39574 * cprop.c (local_cprop_pass): Handle cases where we make an
39575 unconditional trap.
39577 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39581 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39582 read from, for big endian.
39584 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39586 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39587 register pauth builtins for LP64 only.
39589 2017-01-20 Marek Polacek <polacek@redhat.com>
39592 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39595 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39597 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39599 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39600 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39601 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39603 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39606 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39607 in big-endian mode when they are not a single duplicated value.
39609 2017-01-20 Richard Biener <rguenther@suse.de>
39611 * BASE-VER: Bump to 7.0.1.
39613 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39615 * omp-low.c (omplow_simd_context): New struct. Use it...
39616 (lower_rec_simd_input_clauses): ...here and...
39617 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39618 references to idx, lane, max_vf, is_simt.
39620 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39622 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39625 2017-01-20 Martin Jambor <mjambor@suse.cz>
39627 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39628 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39630 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39631 (GTFILES): Rename hsa.c to hsa-common.c.
39632 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39633 * hsa-dump.c: Likewise.
39634 * hsa-gen.c: Likewise.
39635 * hsa-regalloc.c: Likewise.
39636 * ipa-hsa.c: Likewise.
39637 * omp-expand.c: Likewise.
39638 * omp-low.c: Likewise.
39639 * toplev.c: Likewise.
39641 2017-01-20 Marek Polacek <polacek@redhat.com>
39644 * doc/invoke.texi: Document -Wduplicated-branches.
39645 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39646 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39647 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39648 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39649 return 0 only when not OEP_LEXICOGRAPHIC.
39650 (fold_build_cleanup_point_expr): Use the expression
39651 location when building CLEANUP_POINT_EXPR.
39652 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39653 * tree.c (add_expr): Handle error_mark_node.
39655 2017-01-20 Martin Liska <mliska@suse.cz>
39658 * tree-profile.c (init_ic_make_global_vars): Do not call
39660 (gimple_init_gcov_profiler): Likewise.
39662 2017-01-20 Martin Liska <mliska@suse.cz>
39665 * cgraph.h (maybe_create_reference): Remove argument and
39667 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39669 * ipa-cp.c (create_specialized_node): Likewise.
39670 * symtab.c (symtab_node::maybe_create_reference): Handle
39671 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39673 2017-01-20 Martin Liska <mliska@suse.cz>
39675 * read-rtl-function.c (function_reader::create_function): Use
39676 build_decl instread of build_decl_stat.
39678 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39680 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39681 * config/i386/avx512dqintrin.h: Ditto.
39682 * config/i386/avx512fintrin.h: Ditto.
39683 * config/i386/i386-builtin-types.def: Add new types.
39684 * config/i386/i386.c: Handle new types.
39685 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39686 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39687 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39688 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39689 (__builtin_ia32_kshiftridi): New.
39690 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39692 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39696 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39697 define to rs6000_init_stack_protect_guard.
39698 (rs6000_init_stack_protect_guard): New function.
39700 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39701 Yunqiang Su <yunqiang.su@imgtec.com>
39703 * config.gcc (supported_defaults): Add madd4.
39704 (with_madd4): Add validation.
39705 (all_defaults): Add madd4.
39706 * config/mips/mips.opt (mmadd4): New option.
39707 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39709 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39711 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39712 (ISA_HAS_FUSED_MADD4): Likewise.
39713 * doc/invoke.texi (-mmadd4): Document the new option.
39714 * doc/install.texi (--with-madd4): Document the new option.
39716 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39718 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39719 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39720 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39721 (aarch64_init_pauth_hint_builtins): New.
39722 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39723 (aarch64_expand_builtin): Expand new builtins.
39725 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39727 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39728 * combine-stack-adj.c (no_unhandled_cfa): Handle
39729 REG_CFA_TOGGLE_RA_MANGLE.
39730 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39731 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39732 info for return address signing.
39733 (aarch64_expand_epilogue): Likewise.
39735 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39737 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39738 * config/aarch64/aarch64-protos.h
39739 (aarch64_return_address_signing_enabled): New declaration.
39740 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39742 (aarch64_expand_prologue): Sign return address before it's pushed onto
39744 (aarch64_expand_epilogue): Authenticate return address fetched from
39746 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39747 (aarch64_attributes): New function attributes for "sign-return-address".
39748 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39749 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39750 ("*do_return"): Generate combined instructions according to key index.
39751 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39752 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39754 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39755 * config/aarch64/aarch64.opt (msign-return-address=): New.
39756 * doc/extend.texi (AArch64 Function Attributes): Documents
39757 "sign-return-address=".
39758 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39760 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39762 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39763 overall option summary.
39765 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39767 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39768 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39769 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39770 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39772 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39774 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39775 -mpower9-minmax by default for -mcpu=power9.
39776 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39777 128-bit floating point.
39779 2017-01-20 Alan Modra <amodra@gmail.com>
39781 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39782 optimizing for size.
39784 2017-01-20 Alan Modra <amodra@gmail.com>
39787 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39788 for strcmp and strncmp from corresponding builtin decl.
39790 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39792 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39793 instead of i386/rtems-64.h.
39794 * config/i386/rtems-64.h: Remove.
39796 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39800 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39802 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39804 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39806 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39807 Change int to HOST_WIDE_INT.
39808 * config/aarch64/aarch64-protos.h
39809 (aarch64_simd_gen_const_vector_dup): Likewise.
39810 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39812 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39814 * langhooks-def.h (lhd_type_for_size): New decl.
39815 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39816 * langhooks.c (lhd_type_for_size): New function, taken from
39819 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39821 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39822 define_bypass for CR latency.
39823 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39824 (power9-alu2): Add define_bypass for CR latency.
39826 (power9-mul): Update insn latency.
39827 (power9-mul-compare): Update insn latency, bypass latency and remove
39830 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39832 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39834 * config/aarch64/aarch64.md
39835 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39836 aarch64_nopcrelative_literal_loads.
39837 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39839 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39841 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39842 TARGET_LOONGSON_3A.
39843 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39845 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39848 * config.gcc (supported_defaults): Add lxc1-sxc1.
39849 (with_lxc1_sxc1): Add validation.
39850 (all_defaults): Add lxc1-sxc1.
39851 * config/mips/mips.opt (mlxc1-sxc1): New option.
39852 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39854 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39855 __mips_no_lxc1_sxc1.
39856 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39857 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39858 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39860 2017-01-19 Richard Biener <rguenther@suse.de>
39862 PR tree-optimization/72488
39863 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39864 sure to restore SSA info.
39865 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39867 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39869 PR rtl-optimization/79121
39870 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39871 of the inner type when shifting an extended value.
39873 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39876 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39877 interposable aliases.
39879 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39882 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39883 Use the evmergelohi instruction.
39884 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39885 (mov_sitf_e500_subreg8_2_be): Likewise.
39886 (mov_sitf_e500_subreg12_2_le): Likewise.
39887 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39888 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39889 (mov_sitf_e500_subreg8_2_le): Likewise.
39890 (mov_sitf_e500_subreg12_2_be): Likewise.
39892 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39894 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39895 attribute from vecsimple to vecperm.
39896 (altivec_vbpermq2): Likewise.
39898 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39901 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39903 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39904 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39905 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39906 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39907 case where N arg is SIZE_MAX.
39908 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39909 (cmpstrsi): Add pattern.
39911 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39913 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39914 __builtin_vec_revb builtins.
39915 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39916 built-in functions to support generation of the ISA 3.0 XXBR<x>
39917 vector byte reverse instructions.
39918 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39919 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39920 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39921 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39922 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39923 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39924 (P9V_BUILTIN_VEC_REVB): Likewise.
39925 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39926 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39927 (p9_xxbrq_v16qi): Likewise.
39928 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39929 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39930 (p9_xxbrh_v8hi): Likewise.
39931 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39932 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39933 vec_revb built-in functions.
39935 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39937 PR rtl-optimization/78952
39938 * config/i386/i386.md (any_extract): New code iterator.
39939 (*insvqi_2): Use any_extract for source operand.
39940 (*insvqi_3): Use any_shiftrt for source operand.
39942 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39944 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39946 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39948 2017-01-18 Matthias Klose <doko@ubuntu.com>
39950 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39952 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39954 * config/rs6000/altivec.h (vec_bperm): Change #define.
39955 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39956 (altivec_vbpermq2): New define_insn.
39957 (altivec_vbpermd): Likewise.
39958 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39959 function interface.
39960 (VBPERMD): Likewise.
39961 (VBPERM): New polymorphic function interface.
39962 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39963 Add entries for P9V_BUILTIN_VEC_VBPERM.
39964 * doc/extend.texi: Add interfaces for vec_bperm.
39966 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39968 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39969 first letter of error messages.
39970 (s390_resolve_overloaded_builtin): Likewise.
39971 * config/s390/s390.c (s390_expand_builtin): Likewise.
39972 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39973 (s390_valid_target_attribute_inner_p): Likewise.
39974 * config/s390/s390.md ("tabort"): Likewise.
39976 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39978 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39979 (ISA_AVOID_DIV_HILO): New macro.
39980 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39981 (ISA_HAS_DDIV): Likewise.
39983 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39985 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39987 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39989 * doc/invoke.texi (fabi-version): Spelling fix.
39991 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39994 * doc/invoke.texi (fabi-version): Mention mangling fix for
39997 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40000 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40002 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40005 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40006 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40008 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40010 (stack_protect_test): Ditto.
40011 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40012 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40014 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40015 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40016 -mstack-protector-guard-offset=.
40017 (RS/6000 and PowerPC Options): Ditto.
40019 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40021 * config/i386/i386.h (MASK_CLASS_P): New define.
40022 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40023 there are no registers from different register sets also when
40024 mask registers are used. Update function comment.
40025 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40026 to (*k/*r) and (*k/*km) alternatives.
40028 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40030 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40031 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40032 (EH_RETURN_HANDLER_RTX): New define.
40033 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40034 Force frame pointer in EH return functions.
40035 (aarch64_expand_epilogue): Add barrier for eh_return.
40036 (aarch64_final_eh_return_addr): Remove.
40037 (aarch64_eh_return_handler_rtx): New function.
40038 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40040 (aarch64_eh_return_handler_rtx): New prototype.
40042 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40044 * config/rs6000/altivec.h (vec_rlmi): New #define.
40045 (vec_vrlnm): Likewise.
40046 (vec_rlnm): Likewise.
40047 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40048 (UNSPEC_VRLNM): Likewise.
40049 (VIlong): New mode iterator.
40050 (altivec_vrl<VI_char>mi): New define_insn.
40051 (altivec_vrl<VI_char>nm): Likewise.
40052 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40054 (VRLDNM): Likewise.
40055 (RLNM): New polymorphic function entry.
40056 (VRLWMI): New monomorphic function entry.
40057 (VRLDMI): Likewise.
40058 (RLMI): New polymorphic function entry.
40059 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40060 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40061 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40064 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40067 * dwarf2out.c (field_byte_offset): Restore the
40068 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40069 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40070 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40073 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40076 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40078 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40081 * dwarf2out.c (add_data_member_location_attribute): For constant
40082 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40083 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40084 DW_AT_byte_size attributes.
40086 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40088 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40089 after forcing to constant memory when the code model is medium.
40091 2017-01-17 Julia Koval <julia.koval@intel.com>
40094 * config/i386/avx512fintrin.h
40095 (_mm512_i32gather_ps): Change __addr type to void const*.
40096 (_mm512_mask_i32gather_ps): Ditto.
40097 (_mm512_i32gather_pd): Ditto.
40098 (_mm512_mask_i32gather_pd): Ditto.
40099 (_mm512_i64gather_ps): Ditto.
40100 (_mm512_mask_i64gather_ps): Ditto.
40101 (_mm512_i64gather_pd): Ditto.
40102 (_mm512_mask_i64gather_pd): Ditto.
40103 (_mm512_i32gather_epi32): Ditto.
40104 (_mm512_mask_i32gather_epi32): Ditto.
40105 (_mm512_i32gather_epi64): Ditto.
40106 (_mm512_mask_i32gather_epi64): Ditto.
40107 (_mm512_i64gather_epi32): Ditto.
40108 (_mm512_mask_i64gather_epi32): Ditto.
40109 (_mm512_i64gather_epi64): Ditto.
40110 (_mm512_mask_i64gather_epi64): Ditto.
40111 (_mm512_i32scatter_ps): Change __addr type to void*.
40112 (_mm512_mask_i32scatter_ps): Ditto.
40113 (_mm512_i32scatter_pd): Ditto.
40114 (_mm512_mask_i32scatter_pd): Ditto.
40115 (_mm512_i64scatter_ps): Ditto.
40116 (_mm512_mask_i64scatter_ps): Ditto.
40117 (_mm512_i64scatter_pd): Ditto.
40118 (_mm512_mask_i64scatter_pd): Ditto.
40119 (_mm512_i32scatter_epi32): Ditto.
40120 (_mm512_mask_i32scatter_epi32): Ditto.
40121 (_mm512_i32scatter_epi64): Ditto.
40122 (_mm512_mask_i32scatter_epi64): Ditto.
40123 (_mm512_i64scatter_epi32): Ditto.
40124 (_mm512_mask_i64scatter_epi32): Ditto.
40125 (_mm512_i64scatter_epi64): Ditto.
40126 (_mm512_mask_i64scatter_epi64): Ditto.
40127 * config/i386/avx512pfintrin.h
40128 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40129 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40130 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40131 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40132 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40133 (_mm512_prefetch_i32scatter_ps): Ditto.
40134 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40135 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40136 (_mm512_prefetch_i64scatter_pd): Ditto.
40137 (_mm512_prefetch_i64scatter_ps): Ditto.
40138 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40139 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40140 * config/i386/avx512vlintrin.h
40141 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40142 (_mm_mmask_i32gather_ps): Ditto.
40143 (_mm256_mmask_i32gather_pd): Ditto.
40144 (_mm_mmask_i32gather_pd): Ditto.
40145 (_mm256_mmask_i64gather_ps): Ditto.
40146 (_mm_mmask_i64gather_ps): Ditto.
40147 (_mm256_mmask_i64gather_pd): Ditto.
40148 (_mm_mmask_i64gather_pd): Ditto.
40149 (_mm256_mmask_i32gather_epi32): Ditto.
40150 (_mm_mmask_i32gather_epi32): Ditto.
40151 (_mm256_mmask_i32gather_epi64): Ditto.
40152 (_mm_mmask_i32gather_epi64): Ditto.
40153 (_mm256_mmask_i64gather_epi32): Ditto.
40154 (_mm_mmask_i64gather_epi32): Ditto.
40155 (_mm256_mmask_i64gather_epi64): Ditto.
40156 (_mm_mmask_i64gather_epi64): Ditto.
40157 (_mm256_i32scatter_ps): Change __addr type to void*.
40158 (_mm256_mask_i32scatter_ps): Ditto.
40159 (_mm_i32scatter_ps): Ditto.
40160 (_mm_mask_i32scatter_ps): Ditto.
40161 (_mm256_i32scatter_pd): Ditto.
40162 (_mm256_mask_i32scatter_pd): Ditto.
40163 (_mm_i32scatter_pd): Ditto.
40164 (_mm_mask_i32scatter_pd): Ditto.
40165 (_mm256_i64scatter_ps): Ditto.
40166 (_mm256_mask_i64scatter_ps): Ditto.
40167 (_mm_i64scatter_ps): Ditto.
40168 (_mm_mask_i64scatter_ps): Ditto.
40169 (_mm256_i64scatter_pd): Ditto.
40170 (_mm256_mask_i64scatter_pd): Ditto.
40171 (_mm_i64scatter_pd): Ditto.
40172 (_mm_mask_i64scatter_pd): Ditto.
40173 (_mm256_i32scatter_epi32): Ditto.
40174 (_mm256_mask_i32scatter_epi32): Ditto.
40175 (_mm_i32scatter_epi32): Ditto.
40176 (_mm_mask_i32scatter_epi32): Ditto.
40177 (_mm256_i32scatter_epi64): Ditto.
40178 (_mm256_mask_i32scatter_epi64): Ditto.
40179 (_mm_i32scatter_epi64): Ditto.
40180 (_mm_mask_i32scatter_epi64): Ditto.
40181 (_mm256_i64scatter_epi32): Ditto.
40182 (_mm256_mask_i64scatter_epi32): Ditto.
40183 (_mm_i64scatter_epi32): Ditto.
40184 (_mm_mask_i64scatter_epi32): Ditto.
40185 (_mm256_i64scatter_epi64): Ditto.
40186 (_mm256_mask_i64scatter_epi64): Ditto.
40187 (_mm_i64scatter_epi64): Ditto.
40188 (_mm_mask_i64scatter_epi64): Ditto.
40189 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40190 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40191 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40192 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40193 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40194 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40195 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40196 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40197 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40198 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40199 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40200 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40201 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40202 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40203 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40204 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40205 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40206 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40207 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40208 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40209 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40210 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40211 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40212 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40213 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40214 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40215 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40216 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40217 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40218 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40219 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40220 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40221 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40222 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40223 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40224 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40225 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40226 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40227 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40228 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40229 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40230 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40231 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40232 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40233 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40234 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40235 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40236 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40237 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40238 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40239 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40240 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40241 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40242 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40243 definitions accordingly.
40245 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40246 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40249 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40252 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40255 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40256 hard regs for inner regno.
40258 2017-01-17 Martin Liska <mliska@suse.cz>
40261 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40262 assumption and add comment.
40264 2017-01-17 Nathan Sidwell <nathan@acm.org>
40266 * ipa-visibility.c (localize_node): New function, broken out of ...
40267 (function_and_variable_visibility): ... here. Call it.
40269 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40271 PR middle-end/77445
40272 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40273 correctly set frequency of oudgoing edge.
40274 (duplicate_thread_path): Fix profile updating.
40276 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40279 * configure.ac: Add GCC_BASE_VER.
40280 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40281 version from BASE-VER file.
40282 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40283 (gcc.o): Depend on $(BASEVER).
40284 * common.opt (dumpfullversion): New option.
40285 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40286 * doc/invoke.texi: Document -dumpfullversion.
40287 * doc/install.texi: Document --with-gcc-major-version-only.
40288 * configure: Regenerated.
40290 2017-01-17 Richard Biener <rguenther@suse.de>
40292 PR tree-optimization/71433
40293 * tree-vrp.c (register_new_assert_for): Merge same asserts
40294 on all incoming edges.
40295 (process_assert_insertions_for): Handle insertions at the
40298 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40300 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40301 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40303 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40306 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40309 2017-01-17 Alan Modra <amodra@gmail.com>
40312 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40313 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40314 symbolic stack limit when pic.
40316 2017-01-16 Martin Sebor <msebor@redhat.com>
40318 PR tree-optimization/78608
40319 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40321 2017-01-16 Jeff Law <law@redhat.com>
40324 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40325 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40326 for several include directories that may be relative to sysroot.
40327 * config/i386/x-mingw32 (gplus_includedir): Define.
40328 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40329 (native_system_includedir): Likewise.
40330 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40331 override if TARGET_SYSTEM_ROOT is defined.
40332 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40334 PR tree-optimization/79090
40335 PR tree-optimization/33562
40336 PR tree-optimization/61912
40337 PR tree-optimization/77485
40338 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40339 and computed trims into the dump file.
40341 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40343 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40345 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40348 * gimplify.c (gimplify_init_constructor): If want_value and
40349 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40353 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40354 sequence. Formatting fixes.
40355 (doloop_optimize): Formatting fixes.
40358 * gcc.c (debug_level_greater_than_spec_func): New function.
40359 (static_spec_functions): Add debug-level-gt spec function.
40360 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40362 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40363 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40364 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40365 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40366 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40367 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40369 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40371 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40372 QImode fixups to general and mask registers only.
40374 2017-01-16 Carl Love <cel@us.ibm.com>
40376 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40377 for built-in functions
40378 vector signed char vec_nabs (vector signed char)
40379 vector signed short vec_nabs (vector signed short)
40380 vector signed int vec_nabs (vector signed int)
40381 vector signed long long vec_nabs (vector signed long long)
40382 vector float vec_nabs (vector float)
40383 vector double vec_nabs (vector double)
40384 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40386 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40387 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40388 * doc/extend.texi: Update the documentation file for the new built-in
40391 2017-01-16 Martin Sebor <msebor@redhat.com>
40393 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40396 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40398 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40399 UNSPEC_VSX__XXSPLTD to require special splat handling.
40401 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40404 * system.h: Poison strndup.
40406 2017-01-16 Alan Modra <amodra@gmail.com>
40409 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40412 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40414 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40416 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40418 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40419 call recog here. Assert that INSN_CODE (insn) is non-negative.
40421 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40424 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40426 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40427 in the currently scheduled RTL fragment.
40429 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40431 PR rtl-optimization/78751
40432 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40435 2017-01-14 Jeff Law <law@redhat.com>
40437 PR tree-optimization/79090
40438 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40439 variable length stores.
40440 (compute_trims): Delete dead assignment to *trim_tail.
40441 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40444 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40446 PR rtl-optimization/78626
40447 PR rtl-optimization/78727
40448 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40449 of a block, and split such blocks after everything else is finished.
40451 2017-01-14 Alan Modra <amodra@gmail.com>
40454 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40455 target legitimate_combined_insn.
40456 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40457 (rs6000_legitimate_combined_insn): New function.
40458 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40460 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40461 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40462 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40464 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40466 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40468 2017-01-13 Jeff Law <law@redhat.com>
40470 PR tree-optimization/33562
40471 PR tree-optimization/61912
40472 PR tree-optimization/77485
40473 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40475 (delete_dead_assignment): Likewise.
40476 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40477 statement to delete_dead_call and delete_dead_assignment.
40479 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40482 * substring-locations.c (format_warning_va): Strengthen case 1 so
40483 that both endpoints of the substring must be within the format
40484 range for just the substring to be printed.
40486 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40488 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40489 * config/i386/i386.c (ix86_target_string): Add missing options
40490 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40491 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40492 flags_other and ix86_target_other to flags2_other. Display unknown
40494 (ix86_valid_target_attribute_inner_p): Add missing options and
40495 reorder options by implied ISAs, as in ix86_target_string.
40497 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40499 * hash-table.h (hash_table::too_empty_p): New function.
40500 (hash_table::expand): Use it.
40501 (hash_table::traverse): Likewise.
40502 (hash_table::empty_slot): Use sizeof (value_type) instead of
40503 sizeof (PTR) to convert bytes to elements. Shrink the table
40504 if the current size is excessive for the current number of
40507 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40509 * ira-costs.c (record_reg_classes): Break from the inner loop
40510 early once alt_fail is known to be true. Update outer loop
40511 handling accordingly.
40513 2017-01-13 Jeff Law <law@redhat.com>
40515 * tree-ssa-dse.c (decrement_count): New function.
40516 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40517 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40518 when we know the partially dead statement is a mem* function.
40520 PR tree-optimization/61912
40521 PR tree-optimization/77485
40522 * tree-ssa-dse.c: Include expr.h.
40523 (maybe_trim_constructor_store): New function.
40524 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40526 PR tree-optimization/33562
40527 PR tree-optimization/61912
40528 PR tree-optimization/77485
40529 * doc/invoke.texi: Document new dse-max-object-size param.
40530 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40531 * tree-ssa-dse.c: Include params.h.
40532 (dse_store_status): New enum.
40533 (initialize_ao_ref_for_dse): New, partially extracted from
40535 (valid_ao_ref_for_dse, normalize_ref): New.
40536 (setup_live_bytes_from_ref, compute_trims): Likewise.
40537 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40538 (maybe_trim_partially_dead_store): Likewise.
40539 (maybe_trim_complex_store): Likewise.
40540 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40541 Track what bytes live from the original store. Return tri-state
40542 for dead, partially dead or live.
40543 (dse_dom_walker): Add constructor, destructor and new private members.
40544 (delete_dead_call, delete_dead_assignment): New extracted from
40546 (dse_optimize_stmt): Make a member of dse_dom_walker.
40547 Use initialize_ao_ref_for_dse.
40549 PR tree-optimization/33562
40550 PR tree-optimization/61912
40551 PR tree-optimization/77485
40552 * sbitmap.h (bitmap_count_bits): Prototype.
40553 (bitmap_clear_range, bitmap_set_range): Likewise.
40554 * sbitmap.c (bitmap_clear_range): New function.
40555 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40557 2017-01-13 Martin Liska <mliska@suse.cz>
40560 * function.c (set_cfun): Add new argument force.
40561 * function.h (set_cfun): Likewise.
40562 * ipa-inline-transform.c (inline_call): Use the function when
40563 strict alising from is dropped for function we inline to.
40565 2017-01-13 Richard Biener <rguenther@suse.de>
40567 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40568 for dumping GIMPLE INTEGER_CSTs.
40570 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40572 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40573 to 201112L since C++17.
40575 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40578 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40579 if -fsanitize=kernel-address is present.
40581 2017-01-13 Richard Biener <rguenther@suse.de>
40583 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40584 as _Literal ( type ) number in case usual suffixes do not
40585 preserve all information.
40587 2017-01-13 Richard Biener <rguenther@suse.de>
40589 PR tree-optimization/77283
40590 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40591 and ssa-iterators.h.
40592 (is_feasible_trace): Implement a cost model based on joiner
40595 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40598 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40599 char or short to __float128/_Float128 directly.
40601 2017-01-12 Martin Sebor <msebor@redhat.com>
40603 to -Wformat-overflow.
40604 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40605 (min_bytes_remaining): Same.
40606 (get_string_length): Same.
40607 (format_string): Same.
40608 (format_directive): Same.
40610 (pass_sprintf_length::handle_gimple_call): Same.
40612 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40614 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40615 info.nowrite calls with no lhs that can't throw. Return bool
40616 whether gsi_remove has been called or not.
40617 (pass_sprintf_length::handle_gimple_call): Return bool whether
40618 try_substitute_return_value called gsi_remove. Formatting fix.
40619 (pass_sprintf_length::execute): Don't use gsi_remove if
40620 handle_gimple_call returned true.
40623 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40624 be removed due to side-effects, don't remove following barrier nor
40625 turn the successor edge into fallthru edge.
40627 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40630 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40631 element-reversing loads and stores as not swappable.
40633 2017-01-12 Nathan Sidwell <nathan@acm.org>
40634 Nicolai Stange <nicstange@gmail.com>
40636 * combine.c (try_combine): Don't ignore result of overlap checking
40637 loop. Combine overlap & asm check into single loop.
40639 2017-01-12 Richard Biener <rguenther@suse.de>
40641 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40642 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40644 2017-01-12 Richard Biener <rguenther@suse.de>
40646 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40647 and TS_TARGET_OPTION directly derive from TS_BASE.
40648 * tree-core.h (tree_optimization_option): Derive from tree_base.
40649 (tree_target_option): Likewise.
40651 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40653 * config/i386/i386.c (memory_address_length): Increase len
40654 only when rip_relative_addr_p returns false.
40656 2017-01-11 Julia Koval <julia.koval@intel.com>
40658 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40659 (OPTION_MASK_ISA_SGX_SET): New.
40660 (ix86_handle_option): Handle OPT_msgx.
40661 * config.gcc: Added sgxintrin.h.
40662 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40663 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40664 * config/i386/i386.c (ix86_target_string): Add -msgx.
40666 (ix86_option_override_internal): Handle new options.
40667 (ix86_valid_target_attribute_inner_p): Add sgx.
40668 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40669 * config/i386/i386.opt: Add msgx.
40670 * config/i386/sgxintrin.h: New file.
40671 * config/i386/x86intrin.h: Add sgxintrin.h.
40673 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40676 * fold-const.c (maybe_nonzero_address): Return 1 for function
40678 (tree_single_nonzero_warnv_p): Don't handle function local objects
40682 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40685 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40688 * opts.c: Include "spellcheck.h"
40689 (struct string_fragment): New struct.
40690 (struct edit_distance_traits<const string_fragment &>): New
40692 (get_closest_sanitizer_option): New function.
40693 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40695 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40697 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40699 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40700 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40701 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40702 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40703 for initial die_offset if dwarf_split_debug_info.
40704 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40705 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40707 (output_skeleton_debug_sections): Formatting fix. Use
40708 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40709 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40711 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40713 * config/arm/cortex-a53.md: Add bypasses for
40714 cortex_a53_r2f_cvt.
40715 (cortex_a53_r2f): Only use for transfers.
40716 (cortex_a53_f2r): Likewise.
40717 (cortex_a53_r2f_cvt): Add reservation for conversions.
40718 (cortex_a53_f2r_cvt): Likewise.
40720 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40722 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40723 to all inlined functions, change static to extern.
40725 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40728 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40730 (arm_assemble_integer): Likewise.
40732 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40734 * config.gcc: Use new awk script to check CPU, FPU and architecture
40735 parameters for --with-... options.
40736 * config/arm/parsecpu.awk: New file
40737 * config/arm/arm-cpus.in: New file.
40738 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40740 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40742 * config/arm/t-arm: Update dependency rules.
40743 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40744 of processing .def files.
40745 * config/arm/genopt.sh: Deleted.
40746 * config/arm/gentune.sh: Deleted.
40747 * config/arm/arm-cores.def: Deleted.
40748 * config/arm/arm-arches.def: Deleted.
40749 * config/arm/arm-fpus.def: Deleted.
40750 * config/arm/arm-tune.md: Regenerated.
40751 * config/arm/arm-tables.opt: Regenerated.
40752 * config/arm/arm-cpu.h: New generated file.
40753 * config/arm/arm-cpu-data.h: New generated file.
40754 * config/arm/arm-cpu-cdata.h: New generated file.
40756 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40759 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40761 (input_varpool_node): Unpack dynamically_initialized bit.
40763 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40765 PR rtl-optimization/79032
40766 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40767 the alignment of the adjusted memory reference against that of MODE,
40768 instead of the alignment of the original memory reference.
40770 2017-01-11 Martin Jambor <mjambor@suse.cz>
40772 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40774 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40775 decorated functions.
40777 2017-01-11 Richard Biener <rguenther@suse.de>
40779 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40780 set range/nonnull info for PHI results. Do not set it on
40781 stmts marked for removal.
40783 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40785 * expr.c (store_field): In the bitfield case, fetch the return value
40786 from the registers before applying a single big-endian adjustment.
40787 Always do a final load for a BLKmode value not larger than a word.
40789 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40792 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40793 that we correctly handle column numbers greater than
40794 LINE_MAP_MAX_COLUMN_NUMBER.
40796 2017-01-10 Martin Sebor <msebor@redhat.com>
40798 PR middle-end/78245
40799 * gimple-ssa-sprintf.c (get_destination_size): Call
40800 {init,fini}object_sizes.
40801 * tree-object-size.c (addr_object_size): Adjust.
40802 (pass_through_call): Adjust.
40803 (pass_object_sizes::execute): Adjust.
40804 * tree-object-size.h (fini_object_sizes): Declare.
40806 2017-01-10 Martin Sebor <msebor@redhat.com>
40808 PR tree-optimization/78775
40809 * builtins.c (get_size_range): Move...
40810 * calls.c: ...to here.
40811 (alloc_max_size): Accept zero argument.
40812 (operand_signed_p): Remove.
40813 (maybe_warn_alloc_args_overflow): Call get_size_range.
40814 * calls.h (get_size_range): Declare.
40816 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40818 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40819 from TI's devices.csv file as of September 2016.
40820 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40822 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40824 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40825 * doc/invoke.texi: Likewise.
40826 * doc/md.texi: Likewise.
40827 * doc/objc.texi: Likewise.
40829 2017-01-10 Joshua Conner <joshconner@google.com>
40831 * config/arm/fuchsia-elf.h: New file.
40832 * config/fuchsia.h: New file.
40833 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40834 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40836 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40838 2016-01-10 Richard Biener <rguenther@suse.de>
40840 PR tree-optimization/79034
40841 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40842 Propagate out degenerate PHIs in the joiner.
40844 2017-01-10 Martin Liska <mliska@suse.cz>
40846 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40847 (sort_congruence_classes_by_decl_uid): Likewise.
40848 (sort_congruence_class_groups_by_decl_uid): Likewise.
40849 (sem_item_optimizer::merge_classes): Sort class, groups in these
40850 classes and members in the groups by DECL_UID of declarations.
40851 This would make merge operations stable.
40853 2017-01-10 Martin Liska <mliska@suse.cz>
40855 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40856 usage of m_classes_vec.
40857 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40858 (sem_item_optimizer::get_group_by_hash): Likewise.
40859 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40860 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40861 (sem_item_optimizer::verify_classes): Likewise.
40862 (sem_item_optimizer::process_cong_reduction): Likewise.
40863 (sem_item_optimizer::dump_cong_classes): Likewise.
40864 (sem_item_optimizer::merge_classes): Likewise.
40865 * ipa-icf.h (congruence_class_hash): Rename from
40866 congruence_class_group_hash. Remove declaration of m_classes_vec.
40868 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40870 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40871 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40872 * config.gcc: Add avx512vpopcntdqintrin.h.
40873 * config/i386/avx512vpopcntdqintrin.h: New.
40874 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40875 * config/i386/i386-builtin-types.def: Add new types.
40876 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40877 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40878 __builtin_ia32_vpopcountq_v8di_mask): New.
40879 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40880 __AVX512VPOPCNTDQ__.
40881 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40882 (PTA_AVX512VPOPCNTDQ): Define.
40883 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40884 TARGET_AVX512VPOPCNTDQ_P): Define.
40885 * config/i386/i386.opt: Add mavx512vpopcntdq.
40886 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40887 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40889 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40891 PR middle-end/77484
40892 * predict.def (PRED_CALL): Set to 67.
40894 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40896 * expr.c (store_field): In the bitfield case, if the value comes from
40897 a function call and is of an aggregate type returned in registers, do
40898 not modify the field mode; extract the value in all cases if the mode
40899 is BLKmode and the size is not larger than a word.
40901 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40904 * config/i386/cpuid.h: Fix undefined behavior.
40906 2017-01-04 Jeff Law <law@redhat.com>
40908 PR tree-optimization/79007
40909 PR tree-optimization/67955
40910 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40911 conservative for pt.null when flag_non_call_exceptions is on.
40913 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40915 PR translation/79019
40916 PR translation/79020
40917 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40918 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40919 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40921 * config/avr/avr.opt (maccumulate-args): Likewise.
40922 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40923 * common.opt (freport-bug): Likewise.
40924 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40925 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40926 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40927 translatable string.
40928 * config/i386/i386.c (function_value_32): Likewise.
40929 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40930 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40932 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40933 * common/config/msp430/msp430-common.c (msp430_handle_option):
40935 * symtab.c (symtab_node::verify_base): Likewise.
40936 * opts.c (set_debug_level): Likewise.
40937 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40938 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40939 missing whitespace to translatable strings.
40940 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40941 * config/sh/superh.h: Likewise.
40942 * config/i386/xopintrin.h: Likewise.
40943 * config/i386/znver1.md: Likewise.
40944 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40945 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40946 * double-int.h (struct double_int): Likewise.
40947 * double-int.c (div_and_round_double): Likewise.
40948 * wide-int.cc: Likewise.
40949 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40950 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40951 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40952 (crossjumps_occurred): ... this.
40953 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40956 PR tree-optimization/78899
40957 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40958 returning bool return struct loop *, NULL for failure and the new
40960 (versionable_outer_loop_p): Don't version outer loop if it has
40961 dont_vectorized bit set.
40962 (tree_if_conversion): When versioning outer loop, ensure
40963 tree_if_conversion is performed also on the inner loop of the
40964 non-vectorizable outer loop copy.
40965 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40966 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40967 prevent vectorization of it.
40968 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40969 the outer loop vectorization of the non-scalar version is attempted
40970 before vectorization of the inner loop in scalar version. If
40971 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40972 vectorization of its inner loop.
40973 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40974 has 2 inner loops, rename also on edges from bb whose single pred
40975 is outer_loop->header. Fix typo in function comment.
40977 2017-01-09 Martin Sebor <msebor@redhat.com>
40980 * asan.c (asan_emit_stack_protection): Increase local buffer size
40981 to avoid snprintf truncation warning.
40983 2017-01-09 Andrew Pinski <apinski@cavium.com>
40985 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40986 to reference thunderx2t99 for the tuning structure
40987 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40989 (thunderx2t99_extra_costs): This.
40990 * config/aarch64/aarch64-tune.md: Regenerate.
40991 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40992 (vulcan_addrcost_table): This.
40993 (vulcan_regmove_cost): Rename to ...
40994 (thunderx2t99_regmove_cost): This.
40995 (vulcan_vector_cost): Rename to ...
40996 (thunderx2t99_vector_cost): this.
40997 (vulcan_branch_cost): Rename to ...
40998 (thunderx2t99_branch_cost): This.
40999 (vulcan_tunings): Rename to ...
41000 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41001 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41003 2017-01-09 Martin Jambor <mjambor@suse.cz>
41007 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41008 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41009 (propagate_vr_accross_jump_function): Use the above function for all
41010 value range computations for pass-through jump functions and type
41011 converasion from explicit value range values.
41012 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41013 parameters from TYPE_ARG_TYPES.
41014 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41015 (ipa_write_node_info): Stream type of the actual argument.
41016 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41018 2017-01-09 Martin Liska <mliska@suse.cz>
41021 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41022 (lookup_compiler): Do not show error message with have_E.
41024 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41026 PR tree-optimization/78938
41027 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41028 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41029 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41030 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41033 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41035 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41038 2017-01-09 Richard Biener <rguenther@suse.de>
41040 PR tree-optimization/78997
41041 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41042 name condition properly.
41044 2017-01-09 Richard Biener <rguenther@suse.de>
41047 * dwarf2out.c (is_cxx): New overload with context.
41048 (is_naming_typedef_decl): Use it.
41050 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41052 * invoke.texi (Option Summary): Correct spacing in option lists
41053 and add line breaks to fix over-long lines.
41055 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41057 PR middle-end/17660
41059 * extend.texi (Common Variable Attributes): Add xref to GCC
41060 Internals manual to explain mode attribute keywords.
41062 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41065 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41066 and Preprocessor Options.
41067 (Options for Linking): Document -pthread here....
41068 (RS/6000 and PowerPC Options): ...not here.
41069 (Solaris 2 Options): ...or here.
41070 * doc/cppopts.texi: Document -pthread.
41072 2017-01-08 Martin Sebor <msebor@redhat.com>
41074 PR middle-end/77708
41075 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41076 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41077 New member functions.
41078 (format_directive): Used them.
41080 (pass_sprintf_length::handle_gimple_call): Same.
41081 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41082 to avoid truncation for any argument.
41083 (extract_affine_mul): Same.
41084 * tree.c (get_file_function_name): Same.
41086 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41088 PR middle-end/77484
41089 * predict.def (PRED_INDIR_CALL): Set to 86.
41091 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41093 PR preprocessor/54124
41094 * doc/cppopts.texi: Reformat -d subtable to list the full name
41095 of the options. Add cross-reference to the docs for the general
41096 compiler -d options.
41097 * doc/invoke.texi (Developer Options): Add cross-reference to the
41098 preprocessor-specific -d option documentation.
41100 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41102 PR preprocessor/13498
41103 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41104 redudant material, and reflect new command-line options.
41105 (System Headers): Likewise.
41107 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41109 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41110 -isystem, and -idirafter. Copy-edit.
41111 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41112 default for -ftrack-macro-expansion. Delete obsolete and
41113 badly-formatted implementation details about -fdebug-cpp output.
41114 * doc/cppwarnopts.texi: Copy-edit.
41116 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41119 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41120 that the transition from a max line width >= 1<<10 to narrower
41121 lines works correctly.
41123 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41125 * doc/options.texi (PerFunction): New.
41126 * opt-functions.awk (switch_flags): Map both Optimization and
41127 PerFunction to CL_OPTIMIZATION.
41128 * opth-gen.awk: Test for PerFunction flag along with
41130 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41131 it only when the latter is present. Skip those that don't in
41132 the hash function generator.
41133 * common.opt (fvar-tracking): Mark as PerFunction instead of
41135 (fvar-tracking-assignments): Likewise.
41136 (fvar-tracking-assignments-toggle): Likewise.
41137 (fvar-tracking-uninit): Likewise.
41139 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41141 PR translation/79018
41142 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41145 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41148 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41149 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41150 TARGET_LONG_JUMP_TABLE_OFFSETS.
41151 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41152 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41153 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41154 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41155 * config/m68k/m68k.md (tablejump expander): Likewise.
41156 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41157 TARGET_LONG_JUMP_TABLE_OFFSETS.
41158 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41159 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41161 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41162 David Holsgrove <david.holsgrove@xilinx.com>
41164 * common/config/microblaze/microblaze-common.c
41165 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41166 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41168 * config/microblaze/microblaze.c (microblaze_must_save_register)
41169 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41171 (microblaze_eh_return): New function.
41172 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41173 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41174 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41175 * config/microblaze/microblaze.md (eh_return): New pattern.
41177 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41179 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41180 GCC_DIAGNOSTIC_STRINGIFY): Define.
41182 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41184 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41186 * config/arm/arm.md (<mcrr>): New.
41188 * config/arm/arm.c (arm_arch5te): New.
41189 (arm_option_override): Set arm_arch5te.
41190 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41192 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41193 (arm_mcrr_qualifiers): ... this. New.
41194 (MRRC_QUALIFIERS): Define to...
41195 (arm_mrrc_qualifiers): ... this. New.
41196 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41198 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41199 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41200 (MRRCI, mrrc, MRRC): New.
41201 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41202 VUNSPEC_MRRC2): New.
41204 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41206 * config/arm/arm.md (<mcr>): New.
41208 * config/arm/arm.c (arm_coproc_builtin_available): Add
41209 support for mcr, mrc, mcr2 and mrc2.
41210 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41211 (arm_mcr_qualifiers): ... this. New.
41212 (MRC_QUALIFIERS): Define to ...
41213 (arm_mrc_qualifiers): ... this. New.
41214 (MCR_QUALIFIERS): Define to ...
41215 (arm_mcr_qualifiers): ... this. New.
41216 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41218 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41219 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41220 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41221 VUNSPEC_MRC2): New.
41223 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41225 * config/arm/arm.md (*ldc): New.
41229 * config/arm/arm.c (arm_coproc_builtin_available): Add
41230 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41231 (arm_coproc_ldc_stc_legitimate_address): New.
41232 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41233 'qualifier_const_pointer'.
41234 (LDC_QUALIFIERS): Define to...
41235 (arm_ldc_qualifiers): ... this. New.
41236 (STC_QUALIFIERS): Define to...
41237 (arm_stc_qualifiers): ... this. New.
41238 * config/arm/arm-protos.h
41239 (arm_coproc_ldc_stc_legitimate_address): New.
41240 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41241 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41242 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41243 stc2, stcl, stc2l): New.
41244 * config/arm/constraints.md (Uz): New.
41245 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41246 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41247 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41248 VUNSPEC_STC2L): New.
41250 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41252 * config/arm/arm.md (<cdp>): New.
41253 * config/arm/arm.c (neon_const_bounds): Rename this ...
41254 (arm_const_bounds): ... this.
41255 (arm_coproc_builtin_available): New.
41256 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41257 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41258 (CDP_QUALIFIERS): Define to...
41259 (arm_cdp_qualifiers): ... this. New.
41261 (arm_expand_builtin_args): Add case for 6 arguments.
41262 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41263 (arm_const_bounds): ... this.
41264 (arm_coproc_builtin_available): New.
41265 * config/arm/arm_acle.h (__arm_cdp): New.
41267 * config/arm/arm_acle_builtins.def (cdp): New.
41269 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41270 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41271 'arm_const_bounds'.
41272 * config/arm/types.md (coproc): New.
41273 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41274 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41275 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41276 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41278 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41280 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41281 (UBINOP_QUALIFIERS): New.
41283 (acle_builtin_data): New. Change comment.
41284 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41285 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41286 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41287 arm_acle_builtins.def.
41288 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41289 (arm_init_acle_builtins): New.
41290 (CRC32_BUILTIN): Remove.
41291 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41292 crc32cb, crc32ch and crc32cw.
41293 (arm_init_crc32_builtins): Remove.
41294 (arm_init_builtins): Use arm_init_acle_builtins rather
41295 than arm_init_crc32_builtins.
41296 (arm_expand_acle_builtin): New.
41297 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41298 * config/arm/arm_acle_builtins.def: New.
41300 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41302 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41303 (arm_builtin_datum): ... this.
41304 (arm_init_neon_builtin): Rename to ...
41305 (arm_init_builtin): ... this. Add a new parameters PREFIX
41306 and USE_SIG_IN_NAME.
41307 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41308 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41309 'arm_builtin_datum'.
41310 (arm_init_vfp_builtins): Likewise.
41311 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41312 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41313 (arm_expand_neon_args): Rename to ...
41314 (arm_expand_builtin_args): ... this. Rename builtin_arg
41315 enum values and differentiate between ARG_BUILTIN_MEMORY
41316 and ARG_BUILTIN_NEON_MEMORY.
41317 (arm_expand_neon_builtin_1): Rename to ...
41318 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41319 values, arm_expand_builtin_args and add bool parameter NEON.
41320 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41321 (arm_expand_vfp_builtin): Likewise.
41322 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41324 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41326 PR middle-end/77484
41327 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41328 * predict.c (tree_estimate_probability_bb): Reverse direction of
41329 polymorphic call predictor.
41331 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41333 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41334 (determine_pass_name_match): ...this new function and...
41335 (should_skip_pass_p): ...this new function.
41337 2017-01-06 Nathan Sidwell <nathan@acm.org>
41339 * ipa-visibility.c (function_and_variable_visibility): Reformat
41340 comments and long lines. Remove extrneous if.
41341 * symtab.c (symtab_node::make_decl_local): Fix code format.
41342 (symtab_node::set_section_for_node): Fix comment typo.
41344 2017-01-06 Martin Liska <mliska@suse.cz>
41347 * lra-constraints.c: Rename invariant to lra_invariant.
41348 * predict.c (set_even_probabilities): Initialize e to NULL.
41350 2017-01-05 Martin Sebor <msebor@redhat.com>
41352 PR tree-optimization/78910
41353 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41354 (format_integer): Correct off-by-one error in the handling
41355 of precision with negative numbers in signed conversions..
41357 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41359 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41361 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41363 PR tree-optimization/71016
41364 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41365 factor_out_conditional_conversion. Formatting fix.
41366 (factor_out_conditional_conversion): Add cond_stmt argument.
41367 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41368 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41371 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41373 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41374 read-rtl-function.o, and selftest-rtl.o.
41375 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41376 (selftest::aarch64_test_loading_full_dump): New function.
41377 (selftest::aarch64_run_selftests): New function.
41378 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41379 selftest::aarch64_run_selftests.
41380 * config/i386/i386.c
41381 (selftest::ix86_test_loading_dump_fragment_1): New function.
41382 (selftest::ix86_test_loading_call_insn): New function.
41383 (selftest::ix86_test_loading_full_dump): New function.
41384 (selftest::ix86_test_loading_unspec): New function.
41385 (selftest::ix86_run_selftests): Call the new functions.
41386 * emit-rtl.c (maybe_set_max_label_num): New function.
41387 * emit-rtl.h (maybe_set_max_label_num): New decl.
41388 * function.c (instantiate_decls): Guard call to
41389 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41390 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41392 * gensupport.c (gen_reader::gen_reader): Pass "false"
41393 for new "compact" param of rtx_reader.
41394 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41395 rather than an empty string for NULL strings.
41396 * read-md.c: Potentially include config.h rather than bconfig.h.
41397 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41398 (have_error): New global, copied from errors.c.
41399 (md_reader::read_name): Rename to...
41400 (md_reader::read_name_1): ...this, adding "out_loc" param,
41401 and converting "missing name or number" to returning false, rather
41403 (md_reader::read_name): Reimplement in terms of read_name_1.
41404 (md_reader::read_name_or_nil): New function.
41405 (md_reader::read_string): Handle "(nil)" by returning NULL.
41406 (md_reader::md_reader): Add new param "compact".
41407 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41408 (md_reader::read_file): New method.
41409 * read-md.h (md_reader::md_reader): Add new param "compact".
41410 (md_reader::read_file): New method.
41411 (md_reader::is_compact): New accessor.
41412 (md_reader::read_name): Convert return type from void to file_location.
41413 (md_reader::read_name_or_nil): New decl.
41414 (md_reader::read_name_1): New decl.
41415 (md_reader::m_compact): New field.
41416 (noop_reader::noop_reader): Pass "false" for new "compact" param
41418 (rtx_reader::rtx_reader): Add new "compact" param.
41419 (rtx_reader::read_rtx_operand): Make virtual and convert return
41420 type from void to rtx.
41421 (rtx_reader::read_until): New decl.
41422 (rtx_reader::handle_any_trailing_information): New virtual function.
41423 (rtx_reader::postprocess): New virtual function.
41424 (rtx_reader::finalize_string): New virtual function.
41425 (rtx_reader::m_in_call_function_usage): New field.
41426 (rtx_reader::m_reuse_rtx_by_id): New field.
41427 * read-rtl-function.c: New file.
41428 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41429 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41430 (selftest::verify_three_block_rtl_cfg): New decl.
41431 * read-rtl-function.h: New file.
41432 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41433 For host, include function.h, memmodel.h, and emit-rtl.h.
41434 (one_time_initialization): New function.
41435 (struct compact_insn_name): New struct.
41436 (compact_insn_names): New array.
41437 (find_code): Handle insn codes in compact dumps.
41438 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41439 (bind_subst_iter_and_attr): Likewise.
41440 (add_condition_to_string): Likewise.
41441 (add_condition_to_rtx): Likewise.
41442 (apply_attribute_uses): Likewise.
41443 (add_current_iterators): Likewise.
41444 (apply_iterators): Likewise.
41445 (initialize_iterators): Guard usage of apply_subst_iterator with
41446 #ifdef GENERATOR_FILE.
41447 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41448 (md_reader::read_mapping): Likewise.
41449 (add_define_attr_for_define_subst): Likewise.
41450 (add_define_subst_attr): Likewise.
41451 (read_subst_mapping): Likewise.
41452 (check_code_iterator): Likewise.
41453 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41455 (one_time_initialization): New function.
41456 (rtx_reader::read_until): New method.
41457 (read_flags): New function.
41458 (parse_reg_note_name): New function.
41459 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41460 Handle reuse_rtx ids.
41461 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41462 Add parsing support for RTL dumps, mirroring the special-cases in
41463 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41464 values, and calling handle_any_trailing_information.
41465 (rtx_reader::read_rtx_operand): Convert return type from void
41466 to rtx, returning return_rtx. Handle case 'e'. Call
41467 finalize_string on XSTR and XTMPL fields.
41468 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41469 "(nil)" values were omitted. Call the postprocess vfunc on the
41471 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41472 class ctor. Initialize m_in_call_function_usage. Call
41473 one_time_initialization.
41474 * rtl-tests.c (selftest::test_uncond_jump): Call
41475 set_new_first_and_last_insn.
41476 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41477 * selftest-rtl.c: New file.
41478 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41479 (selftest::get_insn_by_uid): New decl.
41480 * selftest-run-tests.c (selftest::run_tests): Call
41481 read_rtl_function_c_tests.
41482 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41483 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41486 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41488 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41489 operands in a special way. Assert that pos+len <= mode precision.
41491 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41493 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41494 3 argument Alias with unlimited for the negative form.
41495 (fno-vect-cost-model): Removed.
41497 2017-01-05 Martin Liska <mliska@suse.cz>
41499 * hsa-gen.c (gen_hsa_divmod): New function.
41500 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41502 2017-01-05 Martin Liska <mliska@suse.cz>
41505 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41508 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41510 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41511 small constant length operands.
41513 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41515 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41516 between loop iterations.
41518 2017-01-05 Martin Liska <mliska@suse.cz>
41521 * gimplify.c (gimplify_decl_expr): Compare to
41522 asan_poisoned_variables instread of checking flags.
41523 (gimplify_target_expr): Likewise.
41524 (gimplify_expr): Likewise.
41525 (gimplify_function_tree): Conditionally initialize
41526 asan_poisoned_variables.
41528 2017-01-04 Jeff Law <law@redhat.com>
41530 PR tree-optimizatin/78812
41531 * rtl.h (contains_mem_rtx_p): Prototype.
41532 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41533 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41534 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41535 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41536 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41538 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41540 * input.c (assert_char_at_range): Default-initialize actual_range.
41542 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41544 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41545 to match the caller.
41547 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41549 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41550 insns after final jump in test to emit dummy move.
41552 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41554 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41555 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41557 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41559 * multiple_target.c (create_dispatcher_calls): Init e_next.
41560 * tree-ssa-loop-split.c (split_loop): Init border.
41561 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41564 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41569 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41570 (altivec_register_operand): Do not return true if the operand
41571 contains a SUBREG mixing SImode and SFmode.
41572 (vsx_register_operand): Likewise.
41573 (vsx_reg_sfsubreg_ok): New predicate.
41574 (vfloat_operand): Do not return true if the operand contains a
41575 SUBREG mixing SImode and SFmode.
41576 (vint_operand): Likewise.
41577 (vlogical_operand): Likewise.
41578 (gpc_reg_operand): Likewise.
41579 (int_reg_operand): Likewise.
41580 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41581 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41582 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41584 (rs6000_emit_move_si_sf_subreg): New helper function.
41585 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41586 fixup SUBREGs involving SImode and SFmode.
41587 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41588 numbers for the new peephole2 optimization.
41589 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41590 the GLIBC math library that do AND/IOR/XOR operations on single
41591 precision floating point.
41592 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41593 target macros to say whether we need to avoid SUBREGs mixing
41595 (TARGET_ALLOW_SF_SUBREG): Likewise.
41596 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41597 (UNSPEC_SI_FROM_SF): Likewise.
41598 (iorxor): Change spacing.
41599 (and_ior_xor): New iterator for AND, IOR, and XOR.
41600 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41601 (movdi_from_sf_zero_ext): Likewise.
41602 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41603 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41604 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41605 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41606 (fms<mode>4): Likewise.
41607 (fnma<mode>4): Likewise.
41608 (fnms<mode>4): Likewise.
41609 (nfma<mode>4): Likewise.
41610 (nfms<mode>4): Likewise.
41612 2017-01-04 Marek Polacek <polacek@redhat.com>
41615 * doc/invoke.texi: Document -Wpointer-compare.
41617 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41619 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41622 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41623 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41626 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41629 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41630 documentation of the powerpc_popcntb_ok attribute.
41631 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41632 code to issue warning messages if a requested CPU configuration is
41633 not supported by the binary (assembler and loader) toolchain.
41634 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41635 made to define a built-in function that has been disabled.
41636 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41637 made to define a built-in function that has been disabled.
41638 (altivec_init_builtins): Add comment explaining why definition
41639 of the DST built-in functions is not preceded by an assertion
41640 check. Add assertions to prevent ICE if attempts are made to
41641 define an altivec predicate or an abs* built-in function that has
41643 (htm_init_builtins): Add comment explaining why definition of the
41644 htm built-in functions is not preceded by an assertion check.
41646 2017-01-04 Jeff Law <law@redhat.com>
41648 PR tree-optimizatin/67955
41649 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41650 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41651 the points-to solution does not include pt_null. Use DECL_PT_UID
41654 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41656 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41657 Use gen_int_mode instead of gen_lopwart for const_int operands.
41659 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41661 PR tree-optimization/71563
41662 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41663 out of range value - has low bits known to be zero.
41665 2017-01-04 Alan Modra <amodra@gmail.com>
41667 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41668 * configure: Regenerate.
41669 * config.in: Regenerate.
41671 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41674 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41675 a substring of the message, but strcmp with the whole message. Ifdef
41676 ENABLE_NLS, translate the message first using dgettext.
41678 2017-01-03 Jeff Law <law@redhat.com>
41680 PR tree-optimizatin/78856
41681 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41682 (mark_threaded_blocks): Remove code to truncate thread paths that
41683 cross multiple loop headers. Instead invalidate the cached loop
41684 iteration information and handle case of a thread path walking
41685 into an irreducible region.
41687 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41690 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41691 assertions. Add support for doing the signbit if the IEEE 128-bit
41692 floating point value is in a GPR.
41693 * config/rs6000/rs6000.md (Fsignbit): Delete.
41694 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41695 Update the length attribute if the value is in a GPR.
41696 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41697 the sign or zero extension instruction, since the value is always 0/1.
41698 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41701 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41702 extracting SImode to a GPR register so that we can generate a
41703 store, limit the vector to be in a traditional Altivec register
41704 for the vextuwrx instruction.
41706 2017-01-03 Ian Lance Taylor <iant@google.com>
41708 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41710 2017-01-03 Martin Sebor <msebor@redhat.com>
41712 PR tree-optimization/78696
41713 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41714 precision. Use MPFR for %f for greater fidelity. Correct handling
41716 (pass_sprintf_length::compute_format_length): Set width and precision
41717 specified by asrerisk to void_node for vararg functions.
41718 (try_substitute_return_value): Adjust dump output.
41720 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41722 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41724 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41726 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41727 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41728 -mlra/-mno-lra was passed to the compiler.
41730 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41732 PR rtl-optimization/65618
41733 * emit-rtl.c (try_split): Move initialization of "before" and
41734 "after" to just before the call to emit_insn_after_setloc.
41736 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41738 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41740 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41742 * dwarf2out.c (gen_enumeration_type_die): When
41743 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41745 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41747 PR tree-optimization/78965
41748 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41749 Change first argument from const call_info & to call_info &. For %n
41750 set info.nowrite to false.
41752 PR middle-end/78901
41753 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41754 possibly throwing calls.
41756 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41757 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41758 and fns handling, rather than in a separate case SSA_NAME.
41760 2017-01-02 Jeff Law <law@redhat.com>
41762 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41763 fixes for first_period and second_period variables.
41765 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41768 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41769 (*insvqi_1): New insn pattern.
41770 (*insvqi_1_mem_rex64): Ditto.
41771 (*insvqi_2): Ditto.
41772 (*insvqi_3): Rename from *insvqi.
41774 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41776 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41778 * doc/cfg.texi (Edges): Remove reference to Java.
41779 (Maintaining the CFG): Ditto.
41781 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41783 PR middle-end/77674
41784 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41785 transparent aliases.
41787 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41789 PR middle-end/77484
41790 * predict.def (PRED_CALL): Update hitrate.
41791 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41792 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41793 into direct/indirect/polymorphic variants.
41795 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41797 Update copyright years.
41799 * gcc.c (process_command): Update copyright notice dates.
41800 * gcov-dump.c (print_version): Ditto.
41801 * gcov.c (print_version): Ditto.
41802 * gcov-tool.c (print_version): Ditto.
41803 * gengtype.c (create_file): Ditto.
41804 * doc/cpp.texi: Bump @copying's copyright year.
41805 * doc/cppinternals.texi: Ditto.
41806 * doc/gcc.texi: Ditto.
41807 * doc/gccint.texi: Ditto.
41808 * doc/gcov.texi: Ditto.
41809 * doc/install.texi: Ditto.
41810 * doc/invoke.texi: Ditto.
41812 Copyright (C) 2017 Free Software Foundation, Inc.
41814 Copying and distribution of this file, with or without modification,
41815 are permitted in any medium without royalty provided the copyright
41816 notice and this notice are preserved.