1 2018-01-01 Jakub Jelinek <jakub@redhat.com>
3 PR tree-optimization/83581
4 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
5 TODO_cleanup_cfg if any changes have been made.
8 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
9 convert_modes if target mode has the right side, but different mode
13 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
14 last argument when extracting from CONCAT. If either from_real or
15 from_imag is NULL, use expansion through memory. If result is not
16 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
17 the parts directly to inner mode, if even that fails, use expansion
21 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
22 check for bswap in mode rather than HImode and use that in expand_unop
25 2017-12-31 Jakub Jelinek <jakub@redhat.com>
28 * config/i386/i386.c (ix86_attribute_table): Remove excess
29 initializer for "shared" attribute.
31 2017-12-30 Tom de Vries <tom@codesourcery.com>
34 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
36 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
37 offload_funcs entries that no longer have a corresponding cgraph_node.
38 Mark the remaining ones as DECL_PRESERVE_P.
39 (output_lto): Call prune_offload_funcs.
41 2017-12-30 Jakub Jelinek <jakub@redhat.com>
43 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
44 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
45 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
46 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
47 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
48 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
49 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
50 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
51 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
52 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
53 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
54 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
55 vaesenc_<mode>, vpclmulqdq_<mode>,
56 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
58 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
60 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
61 _Float<N> and _Float<N>X variants for rounding built-in
63 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
64 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
65 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
66 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
67 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
68 * builtins.c (mathfn_built_in_2): Likewise.
69 * internal-fn.def (CEIL): Likewise.
71 (NEARBYINT): Likewise.
75 * convert.c (convert_to_integer_1): Likewise.
76 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
77 (integer_valued_real_call_p): Likewise.
78 * fold-const-call.c (fold_const_call_ss): Likewise.
79 * gencfn-macros.c (print_case_cfn): Change CFN and operator
80 printers to take a const char * suffix instead of a bool.
81 (print_define_operator_list): Likewise.
82 (fltall_suffixes): New list of suffixes, that include the
83 traditional suffixes as well as all of the _Float<N> and
85 (main): For _Float<N> and _Float<N>X functions, emit both
86 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
87 has the _Float<N> and _Float<N>X case names or operators. The
88 <name>_ALL variant has both the traditional and the
89 _Float<N>/_Float<N>X case names or operators.
90 * match.pd (COPYSIGN optimizations): Provide optimizations for
91 _Float<N> and _Float<N>X types where possible.
92 (MIN/MAX optimizations): Likewise.
93 (sqrt optimizations): Likewise.
94 (rounding optimizations): Likewise.
96 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
98 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
99 gen_int_mode rather than GEN_INT.
100 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
102 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
104 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
105 instead of CONSTANT_P.
106 (gen_vec_series): Likewise.
107 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
109 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
111 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
114 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
116 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
117 (aarch64_ld1x2<VDC:mode>): Likewise.
118 (aarch64_simd_ld1<mode>_x2): Likewise.
119 (aarch64_simd_ld1<mode>_x2): Likewise.
120 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
121 (vld1_s8_x2): Likewise.
122 (vld1_u16_x2): Likewise.
123 (vld1_s16_x2): Likewise.
124 (vld1_u32_x2): Likewise.
125 (vld1_s32_x2): Likewise.
126 (vld1_u64_x2): Likewise.
127 (vld1_s64_x2): Likewise.
128 (vld1_f16_x2): Likewise.
129 (vld1_f32_x2): Likewise.
130 (vld1_f64_x2): Likewise.
131 (vld1_p8_x2): Likewise.
132 (vld1_p16_x2): Likewise.
133 (vld1_p64_x2): Likewise.
134 (vld1q_u8_x2): Likewise.
135 (vld1q_s8_x2): Likewise.
136 (vld1q_u16_x2): Likewise.
137 (vld1q_s16_x2): Likewise.
138 (vld1q_u32_x2): Likewise.
139 (vld1q_s32_x2): Likewise.
140 (vld1q_u64_x2): Likewise.
141 (vld1q_s64_x2): Likewise.
142 (vld1q_f16_x2): Likewise.
143 (vld1q_f32_x2): Likewise.
144 (vld1q_f64_x2): Likewise.
145 (vld1q_p8_x2): Likewise.
146 (vld1q_p16_x2): Likewise.
147 (vld1q_p64_x2): Likewise.
149 2017-12-27 Martin Liska <mliska@suse.cz>
151 PR tree-optimization/83552
152 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
153 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
155 2017-12-27 Tom de Vries <tom@codesourcery.com>
157 * config/nvptx/nvptx.c (nvptx_option_override): Disable
158 -gstatement-frontiers.
160 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
162 PR rtl-optimization/83513
163 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
164 before priority comparison.
166 2017-12-25 Jakub Jelinek <jakub@redhat.com>
169 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
170 ix86_isa_flags2 to ix86_isa_flags.
171 * config/i386/i386-c.c (ix86_target_macros_internal): Test
172 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
173 isa_flags rather than isa_flags2.
174 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
175 and -mavx512bitalg from isa2_opts to isa_opts.
176 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
177 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
178 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
180 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
181 OPTION_MASK_ISA_AVX512F ored with another option similarly to
182 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
183 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
184 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
185 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
186 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
187 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
188 builtins from bdesc_args2 to bdesc_args section.
189 (__builtin_ia32_compressstoreuqi512_mask,
190 __builtin_ia32_compressstoreuhi512_mask,
191 __builtin_ia32_compressstoreuqi256_mask,
192 __builtin_ia32_expandloadqi512_mask,
193 __builtin_ia32_expandloadqi512_maskz,
194 __builtin_ia32_expandloadhi512_mask,
195 __builtin_ia32_expandloadhi512_maskz,
196 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
197 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
198 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
199 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
200 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
201 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
202 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
203 __builtin_ia32_vpshldv_v32hi_maskz,
204 __builtin_ia32_vpopcountb_v64qi_mask,
205 __builtin_ia32_vpopcountw_v32hi_mask,
206 __builtin_ia32_vpshufbitqmb512_mask,
207 __builtin_ia32_vpshufbitqmb256_mask): Add
208 " | OPTION_MASK_ISA_AVX512BW".
209 (__builtin_ia32_expandloadqi256_mask,
210 __builtin_ia32_expandloadqi256_maskz,
211 __builtin_ia32_vpopcountb_v32qi_mask): Add
212 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
213 (__builtin_ia32_expandloadhi256_mask,
214 __builtin_ia32_expandloadhi256_maskz,
215 __builtin_ia32_expandloadqi128_mask,
216 __builtin_ia32_expandloadqi128_maskz,
217 __builtin_ia32_expandloadhi128_mask,
218 __builtin_ia32_expandloadhi128_maskz,
219 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
220 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
221 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
222 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
223 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
224 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
225 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
226 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
227 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
228 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
229 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
230 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
231 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
232 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
233 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
234 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
235 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
236 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
237 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
238 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
239 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
240 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
241 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
242 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
243 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
244 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
245 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
246 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
247 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
248 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
249 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
250 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
251 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
252 __builtin_ia32_vpopcountw_v8hi_mask): Add
253 " | OPTION_MASK_ISA_AVX512VL".
254 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
255 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
256 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
257 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
258 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
259 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
260 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
261 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
262 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
263 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
264 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
265 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
266 _mm_mask_bitshuffle_epi64_mask): Likewise.
267 * common/config/i386/i386-common.c
268 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
269 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
270 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
271 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
272 OPTION_MASK_ISA_AVX512BITALG_UNSET.
273 (OPTION_MASK_ISA2_AVX512F_UNSET,
274 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
275 (ix86_handle_option): For -mno-general-regs-only, clear from
276 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
277 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
278 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
279 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
280 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
281 for moving from ix86_isa_flags2 to ix86_isa_flags.
283 2017-12-23 Jakub Jelinek <jakub@redhat.com>
286 * fold-const.c (struct contains_label_data): New type.
287 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
288 inside of a SWITCH_BODY seen during the walk.
289 (contains_label_p): Use walk_tree instead of
290 walk_tree_without_duplicates, prepare data for contains_label_1 and
293 2017-12-22 Jakub Jelinek <jakub@redhat.com>
296 * tree-iterator.c (alloc_stmt_list): Start with cleared
297 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
301 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
302 enable avx512f explicitly in #pragma GCC target.
303 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
304 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
305 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
306 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
307 __builtin_ia32_vpdpbusds_v8si_mask,
308 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
309 __builtin_ia32_vpdpbusds_v4si_mask,
310 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
311 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
312 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
313 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
314 __builtin_ia32_vpdpwssds_v8si_mask,
315 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
316 __builtin_ia32_vpdpwssds_v4si_mask,
317 __builtin_ia32_vpdpwssds_v4si_maskz): Use
318 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
319 just OPTION_MASK_ISA_AVX512VNNI.
321 2017-12-22 Martin Jambor <mjambor@suse.cz>
324 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
327 2017-12-22 Julia Koval <julia.koval@intel.com>
328 Sebastian Peryt <sebastian.peryt@intel.com>
330 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
331 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
332 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
333 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
334 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
335 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
336 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
337 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
338 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
339 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
340 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
341 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
342 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
343 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
344 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
345 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
346 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
347 _mm256_maskz_popcnt_epi64): New intrinsics.
348 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
349 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
350 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
351 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
352 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
353 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
354 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
355 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
356 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
357 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
358 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
359 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
360 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
361 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
362 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
363 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
364 __builtin_ia32_vpshufbitqmb256_mask,
365 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
366 * config/i386/i386-c.c (__AVX512BITALG__): New.
367 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
368 (ix86_valid_target_attribute_inner_p): Ditto.
369 (ix86_expand_args_builtin): Handle new types.
370 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
371 * config/i386/i386.opt: Add -mavx512bitalg.
372 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
373 avx512bitalgintrin.h.
374 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
375 (vpopcount<mode><mask_name>): Add more types.
376 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
377 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
379 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
381 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
382 Or in OPTION_MASK_ISA_AVX512F_SET.
383 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
384 OPTION_MASK_ISA_AVX512VNNI_UNSET.
385 (ix86_handle_option): Adjust for
386 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
387 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
389 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
390 isa_flag instead of isa_flag2.
391 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
392 isa_opts2 to isa_opts.
393 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
396 2017-12-22 Mike Stump <mikestump@comcast.net>
397 Eric Botcazou <ebotcazou@adacore.com>
399 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
401 2017-12-21 Jakub Jelinek <jakub@redhat.com>
404 * config/i386/i386.c (ix86_function_arg_boundary): Return
405 PARM_BOUNDARY for TYPE_EMPTY_P types.
408 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
409 if navail is >= dir.len.
411 2017-12-21 Steve Ellcey <sellcey@cavium.com>
413 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
416 2017-12-21 Jakub Jelinek <jakub@redhat.com>
418 PR rtl-optimization/80747
419 PR rtl-optimization/83512
420 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
421 succ edge from ENTRY, copy partition from e->dest to the newly
423 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
424 ENTRY, use BB_PARTITION of its successor block as current_partition.
425 Don't copy partition when splitting succ edge from ENTRY.
427 PR tree-optimization/83523
428 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
429 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
430 (convert_mult_to_fma): Likewise.
432 PR tree-optimization/83521
433 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
434 gimple_build_assign without code on result of
435 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
438 2017-12-21 Andrew Pinski <apinski@cavium.com>
439 Steve Ellcey <sellcey@cavium.com>
441 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
442 multi-arch for ilp32.
444 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
447 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
448 constraints to operand 2.
449 (*ashl<mode>3_mask_1): Ditto.
450 (*<shift_insn><mode>3_mask): Ditto.
451 (*<shift_insn><mode>3_mask_1): Ditto.
452 (*<rotate_insn><mode>3_mask): Ditto.
453 (*<rotate_insn><mode>3_mask_1): Ditto.
455 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
457 * reorg.c (make_return_insns): Reemit each insn with its own location.
459 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
461 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
462 generation for cases where splatting a value is not useful.
463 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
464 across a vec_duplicate and a paradoxical subreg forming a vector
465 mode to a vec_concat.
467 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
469 * combine.c (simplify_set): Do not transform subregs to zero_extends
470 if the destination is not a scalar int mode.
472 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
475 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
476 shared zero if the input has overflowed.
478 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
480 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
481 support for Cortex-A55 and Cortex-A75.
483 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
485 * common/config/arm/arm-common.c (compare_opt_names): Add function
486 comment. Use strcmp instead of manual loop.
488 2017-12-21 Martin Liska <mliska@suse.cz>
490 PR gcov-profile/83509
491 * gcov-dump.c (dump_gcov_file): Do not read info about
492 support_unexecuted_blocks for gcda files.
494 2017-12-21 Jakub Jelinek <jakub@redhat.com>
496 PR rtl-optimization/82973
497 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
498 (valid_for_const_vector_p): ... this.
499 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
500 (valid_for_const_vector_p): ... this. Adjust function comment.
501 (gen_vec_duplicate): Adjust caller.
502 * optabs.c (expand_vector_broadcast): Likewise.
503 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
504 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
506 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
507 instead of GET_CODE == CONST_FIXED.
508 (simplify_subreg): Use CONST_FIXED_P macro instead of
509 GET_CODE == CONST_FIXED.
512 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
513 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
514 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
515 (ix86_option_override_internal): Adjust for
516 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
517 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
518 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
519 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
520 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
521 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
522 def_builtin2 for CET builtins.
523 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
524 up formatting in IX86_BUILTIN_RDPID code.
525 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
526 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
527 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
528 ix86_isa_flags2 to ix86_isa_flags.
529 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
531 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
532 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
533 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
535 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
536 Or in OPTION_MASK_ISA_AVX512F_SET.
537 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
538 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
539 (ix86_handle_option): Adjust for
540 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
541 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
543 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
544 Alan Hayward <alan.hayward@arm.com>
545 David Sherwood <david.sherwood@arm.com>
547 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
548 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
549 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
550 Track polynomial offsets.
552 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
553 Alan Hayward <alan.hayward@arm.com>
554 David Sherwood <david.sherwood@arm.com>
556 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
557 Treat drb->init as a poly_int. Fail if its misalignment wrt
558 vector_alignment isn't known.
560 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
561 Alan Hayward <alan.hayward@arm.com>
562 David Sherwood <david.sherwood@arm.com>
564 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
566 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
567 (vect_loop_versioning): Take the loop versioning threshold as a
569 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
570 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
571 versioning_threshold.
572 (vect_analyze_loop_2): Compute the loop versioning threshold
573 whenever loop versioning is needed, and store it in the new
574 field rather than combining it with the cost model threshold.
575 (vect_transform_loop): Update call to vect_loop_versioning.
576 Try to combine the loop versioning and cost thresholds here.
578 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
579 Alan Hayward <alan.hayward@arm.com>
580 David Sherwood <david.sherwood@arm.com>
582 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
583 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
584 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
585 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
586 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
587 (group_compare_offset): Update accordingly.
588 (split_small_address_groups_p): Likewise.
589 (record_use): Take addr_offset as a poly_uint64 rather than
590 an unsigned HOST_WIDE_INT.
591 (strip_offset): Return the offset as a poly_uint64 rather than
592 an unsigned HOST_WIDE_INT.
593 (record_group_use, split_address_groups): Track polynomial offsets.
594 (add_iv_candidate_for_use): Likewise.
595 (addr_offset_valid_p): Take the offset as a poly_int64 rather
596 than a HOST_WIDE_INT.
597 (strip_offset_1): Return the offset as a poly_int64 rather than
600 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
601 Alan Hayward <alan.hayward@arm.com>
602 David Sherwood <david.sherwood@arm.com>
604 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
605 rather than a HOST_WIDE_INT.
606 * tree.c (get_binfo_at_offset): Likewise.
608 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
609 Alan Hayward <alan.hayward@arm.com>
610 David Sherwood <david.sherwood@arm.com>
612 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
613 rather than a HOST_WIDE_INT.
614 * tree-sra.c (build_ref_for_offset): Likewise.
616 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
617 Alan Hayward <alan.hayward@arm.com>
618 David Sherwood <david.sherwood@arm.com>
620 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
622 * tree.c (mem_ref_offset): Likewise.
623 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
624 * builtins.c (get_object_alignment_2): Likewise.
625 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
626 * gimple-fold.c (get_base_constructor): Likewise.
627 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
628 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
630 * ipa-polymorphic-call.c
631 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
632 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
633 (get_ancestor_addr_info): Likewise.
634 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
635 * match.pd: Likewise.
636 * tree-data-ref.c (dr_analyze_innermost): Likewise.
637 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
638 * tree-eh.c (tree_could_trap_p): Likewise.
639 * tree-object-size.c (addr_object_size): Likewise.
640 * tree-ssa-address.c (copy_ref_info): Likewise.
641 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
642 (indirect_refs_may_alias_p): Likewise.
643 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
644 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
645 (non_rewritable_mem_ref_base): Likewise.
646 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
647 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
648 * varasm.c (decode_addr_const): Likewise.
650 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
651 Alan Hayward <alan.hayward@arm.com>
652 David Sherwood <david.sherwood@arm.com>
654 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
655 offsets for MEM_REFs.
657 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
658 Alan Hayward <alan.hayward@arm.com>
659 David Sherwood <david.sherwood@arm.com>
661 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
662 as a poly_uint64 rather than an unsigned int.
663 * tree-ssanames.c (adjust_ptr_info_misalignment): 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 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
671 (decode_addr_const): Update accordingly.
673 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
674 Alan Hayward <alan.hayward@arm.com>
675 David Sherwood <david.sherwood@arm.com>
677 * tree.h (bit_field_size, bit_field_offset): New functions.
678 * hsa-gen.c (gen_hsa_addr): Use them.
679 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
680 (simplify_vector_constructor): Likewise.
681 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
682 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
683 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
684 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
687 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
688 Alan Hayward <alan.hayward@arm.com>
689 David Sherwood <david.sherwood@arm.com>
691 * expr.h (emit_group_load, emit_group_load_into_temps)
692 (emit_group_store): Take the size as a poly_int64 rather than an int.
693 * expr.c (emit_group_load_1, emit_group_load): Likewise.
694 (emit_group_load_into_temp, emit_group_store): Likewise.
696 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
697 Alan Hayward <alan.hayward@arm.com>
698 David Sherwood <david.sherwood@arm.com>
700 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
701 an unsigned int to a poly_uint64.
702 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
703 sizes as poly_uint64s rather than unsigned ints.
704 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
707 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
708 Alan Hayward <alan.hayward@arm.com>
709 David Sherwood <david.sherwood@arm.com>
711 * lra-constraints.c (emit_inc): Change inc_amount from an int
714 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
715 Alan Hayward <alan.hayward@arm.com>
716 David Sherwood <david.sherwood@arm.com>
718 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
720 (add_stack_var, stack_var_cmp, partition_stack_vars)
721 (dump_stack_var_partition): Update accordingly.
722 (alloc_stack_frame_space): Take the size as a poly_int64 rather
723 than a HOST_WIDE_INT.
724 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
725 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
726 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
727 rather than a HOST_WIDE_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 * target.def (return_pops_args): Treat both the input and output
734 sizes as poly_int64s rather than HOST_WIDE_INTS.
735 * targhooks.h (default_return_pops_args): Update accordingly.
736 * targhooks.c (default_return_pops_args): Likewise.
737 * doc/tm.texi: Regenerate.
738 * emit-rtl.h (incoming_args): Change pops_args, size and
739 outgoing_args_size from int to poly_int64_pod.
740 * function.h (expr_status): Change x_pending_stack_adjust and
741 x_stack_pointer_delta from int to poly_int64.
742 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
743 (ARGS_SIZE_RTX): Update accordingly.
744 * calls.c (highest_outgoing_arg_in_use): Change from int to
746 (stack_usage_watermark, stored_args_watermark): New variables.
747 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
748 (emit_call_1): Change the stack_size and rounded_stack_size
749 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
751 (save_fixed_argument_area): Check stack_usage_watermark.
752 (initialize_argument_information): Change old_pending_adj from
753 a HOST_WIDE_INT * to a poly_int64_pod *.
754 (compute_argument_block_size): Return the size as a poly_int64
756 (finalize_must_preallocate): Track polynomial argument sizes.
757 (compute_argument_addresses): Likewise.
758 (internal_arg_pointer_based_exp): Track polynomial offsets.
759 (mem_overlaps_already_clobbered_arg_p): Rename to...
760 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
761 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
762 Check stored_args_used_watermark.
763 (load_register_parameters): Update accordingly.
764 (check_sibcall_argument_overlap_1): Likewise.
765 (combine_pending_stack_adjustment_and_call): Take the unadjusted
766 args size as a poly_int64 rather than an int. Return a bool
767 indicating whether the optimization was possible and return
768 the new adjustment by reference.
769 (check_sibcall_argument_overlap): Track polynomail argument sizes.
770 Update stored_args_watermark.
771 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
772 (expand_call): Likewise. Maintain stack_usage_watermark and
773 stored_args_watermark. Update calls to
774 combine_pending_stack_adjustment_and_call.
775 (emit_library_call_value_1): Handle polynomial argument sizes.
776 Call stack_region_maybe_used_p and mark_stack_region_used.
777 Maintain stack_usage_watermark.
778 (store_one_arg): Likewise. Update call to
779 mem_overlaps_already_clobbered_arg_p.
780 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
782 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
783 * config/microblaze/microblaze.c (microblaze_function_prologue):
785 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
786 TARGET_RETURN_POPS_ARGS interface.
787 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
789 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
790 * config/i386/i386.c (ix86_return_pops_args): Update for new
791 TARGET_RETURN_POPS_ARGS interface.
792 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
793 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
794 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
795 TARGET_RETURN_POPS_ARGS interface.
796 * config/vax/vax.c (vax_return_pops_args): Likewise.
797 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
798 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
799 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
800 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
801 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
802 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
803 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
804 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
805 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
806 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
807 and x_stack_pointer_delta from int to poly_int64.
808 * dojump.c (do_pending_stack_adjust): Update accordingly.
809 * explow.c (allocate_dynamic_stack_space): Handle polynomial
810 stack_pointer_deltas.
811 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
812 (pad_to_arg_alignment): Track polynomial offsets.
813 (assign_parm_find_stack_rtl): Likewise.
814 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
815 * toplev.c (output_stack_usage): Update reference to
816 current_function_pushed_stack_size.
818 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
819 Alan Hayward <alan.hayward@arm.com>
820 David Sherwood <david.sherwood@arm.com>
822 * function.c (in_arg_offset, var_offset, dynamic_offset)
823 (out_arg_offset, cfa_offset): Change from int to poly_int64.
824 (instantiate_new_reg): Return the new offset as a poly_int64_pod
825 rather than a HOST_WIDE_INT.
826 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
827 (instantiate_virtual_regs_in_insn): Likewise.
829 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
830 Alan Hayward <alan.hayward@arm.com>
831 David Sherwood <david.sherwood@arm.com>
833 * rtl.h (get_args_size, add_args_size_note): New functions.
834 (find_args_size_adjust): Return a poly_int64 rather than a
836 (fixup_args_size_notes): Likewise. Make the same change to the
837 end_args_size parameter.
838 * rtlanal.c (get_args_size, add_args_size_note): New functions.
839 * builtins.c (expand_builtin_trap): Use add_args_size_note.
840 * calls.c (emit_call_1): Likewise.
841 * explow.c (adjust_stack_1): Likewise.
842 * cfgcleanup.c (old_insns_match_p): Update use of
843 find_args_size_adjust.
844 * combine.c (distribute_notes): Track polynomial arg sizes.
845 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
846 end_true_args_size, beg_delay_args_size and end_delay_args_size
847 from HOST_WIDE_INT to poly_int64.
848 (add_cfi_args_size): Take the args_size as a poly_int64 rather
849 than a HOST_WIDE_INT.
850 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
851 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
852 polynomial arg sizes.
853 * emit-rtl.c (try_split): Use get_args_size.
854 * recog.c (peep2_attempt): Likewise.
855 * reload1.c (reload_as_needed): Likewise.
856 * expr.c (find_args_size_adjust): Return the adjustment as a
857 poly_int64 rather than a HOST_WIDE_INT.
858 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
859 to a poly_int64 and change the return type in the same way.
860 (emit_single_push_insn): Track polynomial arg sizes.
862 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
863 Alan Hayward <alan.hayward@arm.com>
864 David Sherwood <david.sherwood@arm.com>
866 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
867 from HOST_WIDE_INT to poly_int64.
868 * expr.c (push_block, emit_push_insn): Likewise.
870 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
871 Alan Hayward <alan.hayward@arm.com>
872 David Sherwood <david.sherwood@arm.com>
874 * function.h (frame_space): Change start and length from HOST_WIDE_INT
876 (get_frame_size): Return the size as a poly_int64 rather than a
878 (frame_offset_overflow): Take the offset as a poly_int64 rather
879 than a HOST_WIDE_INT.
880 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
881 (assign_stack_temp): Likewise for the size.
882 * function.c (get_frame_size): Return a poly_int64 rather than
884 (frame_offset_overflow): Take the offset as a poly_int64 rather
885 than a HOST_WIDE_INT.
886 (try_fit_stack_local): Take the start, length and size as poly_int64s
887 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
888 rather than a HOST_WIDE_INT.
889 (add_frame_space): Take the start and end as poly_int64s rather than
891 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
892 (assign_stack_temp): Likewise for the size.
893 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
895 (find_temp_slot_from_address): Handle polynomial offsets.
896 (combine_temp_slots): Likewise.
897 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
899 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
900 poly_int64 rather than a HOST_WIDE_INT.
901 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
902 than a HOST_WIDE_INT.
903 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
904 polynomial frame offsets.
905 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
906 as a poly_int64 rather than an int.
907 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
908 * config/v850/v850-protos.h (compute_frame_size): Likewise.
909 * config/v850/v850.c (compute_frame_size): Likewise.
910 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
911 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
912 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
913 * config/pa/pa.c (pa_compute_frame_size): Likewise.
914 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
915 rather than a HOST_WIDE_INT.
916 * explow.c (get_dynamic_stack_base): Likewise.
917 * final.c (final_start_function): Use the constant lower bound
918 of the frame size for -Wframe-larger-than.
919 * ira.c (do_reload): Adjust for new get_frame_size return type.
920 * lra.c (lra): Likewise.
921 * reload1.c (reload): Likewise.
922 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
923 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
924 * rtlanal.c (get_initial_register_offset): Return the offset as
925 a poly_int64 rather than a HOST_WIDE_INT.
927 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
928 Alan Hayward <alan.hayward@arm.com>
929 David Sherwood <david.sherwood@arm.com>
931 * reload1.c (elim_table): Change initial_offset, offset and
932 previous_offset from HOST_WIDE_INT to poly_int64_pod.
933 (offsets_at): Change the target array's element type from
934 HOST_WIDE_INT to poly_int64_pod.
935 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
936 (elimination_costs_in_insn, update_eliminable_offsets)
937 (verify_initial_elim_offsets, set_offsets_for_label)
938 (init_eliminable_invariants): Update after above changes.
940 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
941 Alan Hayward <alan.hayward@arm.com>
942 David Sherwood <david.sherwood@arm.com>
944 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
945 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
946 (decomposition): Change start and end from HOST_WIDE_INT
948 (decompose, immune_p): Update accordingly.
949 (find_inc_amount): Return a poly_int64 rather than an int.
950 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
953 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
954 Alan Hayward <alan.hayward@arm.com>
955 David Sherwood <david.sherwood@arm.com>
957 * tree.h (get_inner_reference): Return the bitsize and bitpos
958 as poly_int64_pods rather than HOST_WIDE_INT.
959 * fold-const.h (ptr_difference_const): Return the pointer difference
960 as a poly_int64_pod rather than a HOST_WIDE_INT.
961 * expr.c (get_inner_reference): Return the bitsize and bitpos
962 as poly_int64_pods rather than HOST_WIDE_INT.
963 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
965 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
966 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
967 (optimize_bit_field_compare): Update call to get_inner_reference.
968 (decode_field_reference): Likewise.
969 (fold_unary_loc): Track polynomial offsets and sizes.
970 (split_address_to_core_and_offset): Return the bitpos as a
971 poly_int64_pod rather than a HOST_WIDE_INT.
972 (ptr_difference_const): Likewise for the pointer difference.
973 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
974 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
975 * dbxout.c (dbxout_expand_expr): Likewise.
976 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
977 (loc_list_from_tree_1, fortran_common): Likewise.
978 * gimple-laddress.c (pass_laddress::execute): Likewise.
979 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
980 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
981 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
982 * tree-affine.c (tree_to_aff_combination): Likewise.
983 (get_inner_reference_aff): Likewise.
984 * tree-data-ref.c (split_constant_offset_1): Likewise.
985 (dr_analyze_innermost): Likewise.
986 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
987 * tree-sra.c (ipa_sra_check_caller): Likewise.
988 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
989 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
990 (instrument_bool_enum_load, instrument_object_size): Likewise.
991 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
992 to get_inner_reference.
993 * hsa-gen.c (gen_hsa_addr): Likewise.
994 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
995 * tsan.c (instrument_expr): Likewise.
996 * match.pd: Update call to ptr_difference_const.
998 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
999 Alan Hayward <alan.hayward@arm.com>
1000 David Sherwood <david.sherwood@arm.com>
1002 * fold-const.c (fold_comparison): Track sizes and offsets as
1003 poly_int64s rather than HOST_WIDE_INTs when folding address
1006 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1007 Alan Hayward <alan.hayward@arm.com>
1008 David Sherwood <david.sherwood@arm.com>
1010 * expr.h (get_bit_range): Return the bitstart and bitend as
1011 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1012 as a poly_int64 rather than a HOST_WIDE_INT.
1013 * expr.c (get_bit_range): Likewise.
1014 (expand_assignment): Update call accordingly.
1015 * fold-const.c (optimize_bit_field_compare): Likewise.
1017 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1018 Alan Hayward <alan.hayward@arm.com>
1019 David Sherwood <david.sherwood@arm.com>
1021 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1022 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1023 as an optional final parameter.
1024 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1025 (aarch64_legitimate_address_p): Likewise.
1026 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1027 instead of an rtx code.
1028 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1029 (aarch64_legitimate_address_hook_p): Likewise.
1030 (aarch64_print_ldpstp_address): Likewise.
1031 (aarch64_print_operand_address): Likewise.
1032 (aarch64_address_cost): Likewise.
1033 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1034 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1035 (aarch64_mem_pair_lanes_operand): Likewise.
1037 2017-12-20 Richard Biener <rguenther@suse.de>
1039 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1040 update_stmt_if_modified.
1042 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1044 PR tree-optimization/83491
1045 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1046 before walking uses. Improve coding style and comments.
1048 2017-12-20 Tom de Vries <tom@codesourcery.com>
1050 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1052 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1055 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1057 (pass_ipa_free_fn_summary): Add small_p private data member,
1058 initialize to false in the ctor.
1059 (pass_ipa_free_fn_summary::clone,
1060 pass_ipa_free_fn_summary::set_pass_param,
1061 pass_ipa_free_fn_summary::gate): New methods.
1062 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1063 | TODO_dump_symtab if small_p.
1064 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1065 entry and add another instance of the pass with false parm after
1067 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1068 ipa_free_fn_summary here.
1070 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1072 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1074 2017-12-20 Martin Sebor <msebor@redhat.com>
1077 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1078 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1079 (expand_builtin_strncmp): Same.
1081 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1084 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1085 allowed debug stmts before labels.
1086 (expand_gimple_basic_block): Likewise.
1087 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1088 * gimple-iterator.h (gsi_after_labels): Likewise.
1089 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1090 rename reused variable, and simplify using gsi_move_before.
1091 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1092 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1093 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1094 (gimple_verify_flow_info, gimple_block_label): Likewise.
1095 (make_blocks): Move debug markers after adjacent labels.
1096 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1097 allowed debug insns outside blocks.
1098 * df-scan.c (df_insn_delete): Likewise.
1099 * lra-constraints.c (update_ebb_live_info): Likewise.
1100 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1101 (vt_initialize, delete_vta_debug_insns): Likewise.
1102 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1104 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1105 Alan Hayward <alan.hayward@arm.com>
1106 David Sherwood <david.sherwood@arm.com>
1108 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1109 (round_up_to_byte_boundary): Likewise.
1110 * expr.h (get_bit_range): Add temporary shim.
1111 * gimple-ssa-store-merging.c (store_operand_info): Change the
1112 bitsize, bitpos, bitregion_start and bitregion_end fields from
1113 unsigned HOST_WIDE_INT to poly_uint64.
1114 (merged_store_group): Likewise load_align_base.
1115 (compatible_load_p, compatible_load_p): Update accordingly.
1116 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1117 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1118 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1119 bitregion_start and bitregion_end as poly_uint64s rather than
1120 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1121 (handled_load): Take the bitsize, bitpos,
1122 bitregion_start and bitregion_end as poly_uint64s rather than
1123 unsigned HOST_WIDE_INTs.
1124 (pass_store_merging::process_store): Update call to
1125 mem_valid_for_store_merging.
1127 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1128 Alan Hayward <alan.hayward@arm.com>
1129 David Sherwood <david.sherwood@arm.com>
1131 * builtins.c (get_object_alignment_2): Track polynomial offsets
1132 and sizes. Update the alignment handling.
1134 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1135 Alan Hayward <alan.hayward@arm.com>
1136 David Sherwood <david.sherwood@arm.com>
1138 * tree.h (get_inner_reference): Add a version that returns the
1139 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1140 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1141 the case in which bitpos is not associated with the first byte.
1143 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1144 Alan Hayward <alan.hayward@arm.com>
1145 David Sherwood <david.sherwood@arm.com>
1147 * tree-affine.h (get_inner_reference_aff): Return the size as a
1149 * tree-affine.c (get_inner_reference_aff): Likewise.
1150 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1151 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1153 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1154 Alan Hayward <alan.hayward@arm.com>
1155 David Sherwood <david.sherwood@arm.com>
1157 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1158 HOST_WIDE_INT rather than a poly_int64.
1160 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1161 Alan Hayward <alan.hayward@arm.com>
1162 David Sherwood <david.sherwood@arm.com>
1164 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1165 HOST_WIDE_INT to poly_int64_pod.
1166 (perform_symbolic_merge): Update accordingly.
1167 (bswap_replace): Likewise.
1169 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1170 Alan Hayward <alan.hayward@arm.com>
1171 David Sherwood <david.sherwood@arm.com>
1173 * tree-affine.h (aff_tree::offset): Change from widest_int
1175 (wide_int_ext_for_comb): Delete.
1176 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1177 constants as poly_widest_int rather than widest_int.
1178 (aff_combination_constant_multiple_p): Return the multiplier
1179 as a poly_widest_int.
1180 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1182 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1183 version static and add an overload for poly_widest_int.
1184 (aff_combination_const, aff_combination_add_cst)
1185 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1186 the constants as poly_widest_int rather than widest_int.
1187 (tree_to_aff_combination): Generalize INTEGER_CST case to
1189 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1190 (aff_combination_add_product, aff_combination_mult): Handle
1192 (aff_combination_constant_multiple_p): Return the multiplier
1193 as a poly_widest_int.
1194 * tree-predcom.c (determine_offset): Return the offset as a
1196 (split_data_refs_to_components, suitable_component_p): Update
1198 (valid_initializer_p): Update call to
1199 aff_combination_constant_multiple_p.
1200 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1201 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1202 as a poly_int64 rather than a HOST_WIDE_INT.
1203 (get_address_cost): Handle polynomial offsets.
1204 (iv_elimination_compare_lt): Likewise.
1205 (rewrite_use_nonlinear_expr): Likewise.
1207 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1208 Alan Hayward <alan.hayward@arm.com>
1209 David Sherwood <david.sherwood@arm.com>
1211 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1212 as a poly_int64_pod rather than a HOST_WIDE_INT.
1213 (get_addr_base_and_unit_offset): Likewise.
1214 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1215 (get_addr_base_and_unit_offset): Likewise.
1216 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1217 to poly_int64 in example.
1218 * fold-const.c (fold_binary_loc): Update call to
1219 get_addr_base_and_unit_offset.
1220 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1221 (maybe_canonicalize_mem_ref_addr): Likewise.
1222 (gimple_fold_stmt_to_constant_1): Likewise.
1223 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1225 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1226 * match.pd: Likewise.
1227 * omp-low.c (lower_omp_target): Likewise.
1228 * tree-sra.c (build_ref_for_offset): Likewise.
1229 (build_debug_ref_for_model): Likewise.
1230 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1231 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1232 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1233 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1234 (constant_pointer_difference): Likewise.
1235 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1236 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1237 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1238 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1239 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1240 (set_ssa_val_to): Likewise.
1241 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1242 (maybe_diag_stxncpy_trunc): Likewise.
1243 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1244 * tree.c (build_simple_mem_ref_loc): Likewise.
1245 (array_at_struct_end_p): Likewise.
1247 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1248 Alan Hayward <alan.hayward@arm.com>
1249 David Sherwood <david.sherwood@arm.com>
1251 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1252 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1253 (get_ref_base_and_extent_hwi): Declare.
1254 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1255 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1256 (get_ref_base_and_extent_hwi): New function.
1257 * cfgexpand.c (expand_debug_expr): Update call to
1258 get_ref_base_and_extent.
1259 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1260 * gimple-fold.c (get_base_constructor): Return the offset as a
1261 poly_int64_pod rather than a HOST_WIDE_INT.
1262 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1263 * ipa-polymorphic-call.c
1264 (ipa_polymorphic_call_context::set_by_invariant)
1265 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1266 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1267 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1268 rather than get_ref_base_and_extent.
1269 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1270 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1271 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1273 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1274 call to get_ref_base_and_extent.
1275 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1276 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1277 (stmt_kills_ref_p): Likewise.
1278 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1279 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1281 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1282 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1283 when calling native_encode_expr.
1284 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1285 call to get_ref_base_and_extent.
1286 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1287 get_ref_base_and_extent.
1288 * var-tracking.c (track_expr_p): Likewise.
1290 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1291 Alan Hayward <alan.hayward@arm.com>
1292 David Sherwood <david.sherwood@arm.com>
1294 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1295 HOST_WIDE_INT to poly_int64_pod.
1296 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1297 polynomail parameter offsets.
1299 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1300 Alan Hayward <alan.hayward@arm.com>
1301 David Sherwood <david.sherwood@arm.com>
1303 * gengtype.c (main): Handle poly_int64_pod.
1304 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1305 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1306 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1307 from HOST_WIDE_INT to poly_int64_pod.
1308 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1309 (copy_cfa): New function.
1310 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1311 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1312 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1313 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1314 (initial_return_save): Treat offsets as poly_ints.
1315 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1316 offset is nonconstant.
1317 (reg_save): Take the offset as a poly_int64. Fall back to
1318 DW_CFA_expression for nonconstant offsets.
1319 (queue_reg_save): Take the offset as a poly_int64.
1320 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1322 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1323 Alan Hayward <alan.hayward@arm.com>
1324 David Sherwood <david.sherwood@arm.com>
1326 * rtl.h (operand_subword, operand_subword_force): Take the offset
1327 as a poly_uint64 an unsigned int.
1328 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1330 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1331 Alan Hayward <alan.hayward@arm.com>
1332 David Sherwood <david.sherwood@arm.com>
1334 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1335 'p' format code. Use INT_LIST rather than SUBREG as the example of
1336 a code with an XINT and an XEXP. Remove the implication that
1337 accessing an rtx field using XINT is expected to work.
1338 * rtl.def (SUBREG): Change format from "ei" to "ep".
1339 * rtl.h (rtunion::rt_subreg): New field.
1340 (XCSUBREG): New macro.
1341 (SUBREG_BYTE): Use it.
1342 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1343 Update constructor accordingly.
1344 (subreg_shape::operator ==): Update accordingly.
1345 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1346 than an unsigned int.
1347 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1348 a poly_uint64 rather than an unsigned int.
1349 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1350 than an unsigned int.
1351 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1352 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1353 an unsigned int. Take the sizes as poly_uint64s.
1354 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1355 an unsigned int. Take the shift as a poly_uint64 rather than
1357 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1358 as a poly_uint64 rather than an unsigned int.
1359 (simplify_subreg_regno): Likewise.
1360 (byte_lowpart_offset): Return the memory offset as a poly_int64
1362 (subreg_memory_offset): Likewise. Take the subreg offset as a
1363 poly_uint64 rather than an unsigned int.
1364 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1365 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1366 poly_uint64 rather than an unsigned int.
1367 * rtl.c (rtx_format): Describe 'p' in comment.
1368 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1369 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1370 offset as a poly_uint64 rather than an unsigned int.
1371 (byte_lowpart_offset): Return the memory offset as a poly_int64
1373 (subreg_memory_offset): Likewise. Take the subreg offset as a
1374 poly_uint64 rather than an unsigned int.
1375 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1376 mode sizes as poly_uint64s rather than unsigned ints. Return a
1377 poly_uint64 rather than an unsigned int.
1378 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1379 (copy_insn_1): Handle 'p'.
1380 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1381 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1382 an unsigned int. Return the shift in the same way.
1383 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1385 (subreg_size_offset_from_lsb): Take the sizes and shift as
1386 poly_uint64s rather than unsigned ints. Return the offset as
1388 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1389 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1391 * rtlhash.c (add_rtx): Handle 'p'.
1392 * genemit.c (gen_exp): Likewise.
1393 * gengenrtl.c (type_from_format, gendef): Likewise.
1394 * gensupport.c (subst_pattern_match, get_alternatives_number)
1395 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1396 (subst_dup): Likewise.
1397 * gengtype.c (adjust_field_rtx_def): Likewise.
1398 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1399 (match_pattern_2): Likewise.
1400 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1401 (rtx_test::subreg_field): New function.
1402 (operator ==, safe_to_hoist_p, transition_parameter_type)
1403 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1404 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1405 * genpeep.c (match_rtx): Likewise.
1406 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1407 (rtx_writer::print_rtx_operand): Handle 'p'.
1408 (print_value): Handle SUBREG.
1409 * read-rtl.c (apply_int_iterator): Likewise.
1410 (rtx_reader::read_rtx_operand): Handle 'p'.
1411 * alias.c (rtx_equal_for_memref_p): Likewise.
1412 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1413 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1415 * calls.c (expand_call): Likewise.
1416 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1417 (make_extraction, gen_lowpart_for_combine): Likewise.
1418 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1420 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1421 rather than an unsigned int. Treat subreg offsets as poly_ints.
1422 (exp_equiv_p): Handle 'p'.
1423 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1424 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1425 * dse.c (find_shift_sequence): Likewise.
1426 * dwarf2out.c (rtl_for_decl_location): Likewise.
1427 * expmed.c (extract_low_bits): Likewise.
1428 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1429 (expand_expr_real_2): Likewise.
1430 * final.c (alter_subreg): Likewise.
1431 (leaf_renumber_regs_insn): Handle 'p'.
1432 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1433 Treat subreg offsets as poly_ints.
1434 * fwprop.c (forward_propagate_and_simplify): Likewise.
1435 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1436 * ira.c (get_subreg_tracking_sizes): Likewise.
1437 * ira-conflicts.c (go_through_subreg): Likewise.
1438 * ira-lives.c (process_single_reg_class_operands): Likewise.
1439 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1440 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1441 as a poly_uint64 rather than an unsigned int.
1442 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1443 subreg offsets as poly_ints.
1444 * lra-constraints.c (operands_match_p): Handle 'p'.
1445 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1446 * lra-spills.c (assign_mem_slot): Likewise.
1447 * postreload.c (move2add_valid_value_p): Likewise.
1448 * recog.c (general_operand, indirect_operand): Likewise.
1449 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1450 (copyprop_hardreg_forward_1): Likewise.
1451 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1452 (record_subregs_of_mode): Likewise.
1453 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1454 * reload.c (operands_match_p): Handle 'p'.
1455 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1456 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1457 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1458 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1459 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1460 (simplify_const_poly_int_tests<N>::run): Likewise.
1461 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1462 a poly_uint64 rather than an unsigned int.
1463 * valtrack.c (debug_lowpart_subreg): Likewise.
1464 * var-tracking.c (var_lowpart): Likewise.
1465 (loc_cmp): Handle 'p'.
1467 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1468 Alan Hayward <alan.hayward@arm.com>
1469 David Sherwood <david.sherwood@arm.com>
1471 * ira.c (get_subreg_tracking_sizes): New function.
1472 (init_live_subregs): Take an integer size rather than a register.
1473 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1474 to init_live_subregs.
1476 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1477 Alan Hayward <alan.hayward@arm.com>
1478 David Sherwood <david.sherwood@arm.com>
1480 * expr.c (store_constructor_field): Change bitsize from a
1481 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1482 HOST_WIDE_INT to a poly_int64.
1483 (store_constructor): Change size from a HOST_WIDE_INT to
1485 (store_field): Likewise bitsize and bitpos.
1487 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1488 Alan Hayward <alan.hayward@arm.com>
1489 David Sherwood <david.sherwood@arm.com>
1491 * expmed.h (store_bit_field): Change bitregion_start and
1492 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1493 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1494 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1495 (store_fixed_bit_field, store_split_bit_field): Likewise.
1496 * expr.c (store_constructor_field, store_field): Likewise.
1497 (optimize_bitfield_assignment_op): Likewise. Make the same change
1498 to bitsize and bitpos.
1499 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1500 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1501 same change in the constructor arguments.
1502 (get_best_mode): Change bitregion_start and bitregion_end from
1503 unsigned HOST_WIDE_INT to poly_uint64.
1504 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1505 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1507 (bit_field_mode_iterator::next_mode): Update for new types
1508 of m_bitregion_start and m_bitregion_end.
1509 (get_best_mode): Change bitregion_start and bitregion_end from
1510 unsigned HOST_WIDE_INT to poly_uint64.
1512 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1513 Alan Hayward <alan.hayward@arm.com>
1514 David Sherwood <david.sherwood@arm.com>
1516 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1517 accepts poly_uint64 offsets.
1518 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1519 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1520 * expmed.c (lowpart_bit_field_p): Likewise.
1521 (extract_bit_field_as_subreg): New function, split out from...
1522 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1523 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1524 extractions, check that BITSIZE matches the size of the extracted
1525 value and that BITNUM is an exact multiple of that size.
1526 If all else fails, try forcing the value into memory if
1527 BITNUM is variable, and adjusting the address so that the
1528 offset is constant. Split the part that can only handle constant
1529 bitsize and bitnum out into...
1530 (extract_integral_bit_field): ...this new function.
1531 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1532 rather than unsigned HOST_WIDE_INTs.
1534 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1535 Alan Hayward <alan.hayward@arm.com>
1536 David Sherwood <david.sherwood@arm.com>
1538 * expmed.h (store_bit_field): Take bitsize and bitnum as
1539 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1540 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1541 that returns the byte size.
1542 (store_bit_field_1): Take bitsize and bitnum as
1543 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1544 to simple_mem_bitfield_p. Split the part that can only handle
1545 constant bitsize and bitnum out into...
1546 (store_integral_bit_field): ...this new function.
1547 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1548 than unsigned HOST_WIDE_INTs.
1549 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1551 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1552 Alan Hayward <alan.hayward@arm.com>
1553 David Sherwood <david.sherwood@arm.com>
1555 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1556 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1558 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1559 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1560 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1561 offset as a poly_int64 rather than an int.
1562 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1563 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1564 * lra-constraints.c (equiv_address_substitution): Track offsets
1566 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1567 (curr_insn_transform): Handle the new form of sp_offset.
1568 * lra-eliminations.c (lra_elim_table): Change previous_offset
1569 and offset from HOST_WIDE_INT to poly_int64.
1570 (print_elim_table, update_reg_eliminate): Update accordingly.
1571 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1572 (get_elimination): Update accordingly.
1573 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1574 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1575 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1576 poly_int64 offsets generally.
1577 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1578 (mark_not_eliminable, init_elimination): Update accordingly.
1579 (remove_reg_equal_offset_note): Return a bool and pass the new
1580 offset back by pointer as a poly_int64.
1581 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1582 rather than a HOST_WIDE_INT.
1583 (do_remat): Track offsets poly_int64s.
1584 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1586 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1587 Alan Hayward <alan.hayward@arm.com>
1588 David Sherwood <david.sherwood@arm.com>
1590 * rtl.h (mem_attrs): Add a default constructor. Change size and
1591 offset from HOST_WIDE_INT to poly_int64.
1592 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1593 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1594 (widen_memory_access): Take the sizes and offsets as poly_int64s
1595 rather than HOST_WIDE_INTs.
1596 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1597 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1599 (adjust_offset_for_component_ref): Change the offset from a
1600 HOST_WIDE_INT to a poly_int64.
1601 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1602 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1603 * dce.c (find_call_stack_args): Likewise.
1604 * dse.c (record_store): Likewise.
1605 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1606 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1607 * read-rtl-function.c (test_loading_mem): Likewise.
1608 * rtlanal.c (may_trap_p_1): Likewise.
1609 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1610 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1611 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1612 (mem_attrs::mem_attrs): New function.
1613 (set_mem_attributes_minus_bitpos): Change bitpos from a
1614 HOST_WIDE_INT to poly_int64.
1615 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1616 (clear_mem_offset, clear_mem_size, change_address)
1617 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1618 initialize mem_attrs.
1619 (set_mem_offset, set_mem_size, adjust_address_1)
1620 (adjust_automodify_address_1, offset_address, widen_memory_access):
1621 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1623 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1624 Alan Hayward <alan.hayward@arm.com>
1625 David Sherwood <david.sherwood@arm.com>
1627 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1628 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1629 rather than 0 to represent an unknown size. Assert that the size
1630 is known when the mode isn't BLKmode.
1631 (may_trap_p_1): Use -1 for unknown sizes.
1632 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1634 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1635 Alan Hayward <alan.hayward@arm.com>
1636 David Sherwood <david.sherwood@arm.com>
1638 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1639 to poly_int64. Update commentary for positions_needed.large.
1640 (read_info_type): Change offset and width from HOST_WIDE_INT
1642 (set_usage_bits): Likewise.
1643 (canon_address): Return the offset as a poly_int64 rather than
1644 a HOST_WIDE_INT. Use strip_offset_and_add.
1645 (set_all_positions_unneeded, any_positions_needed_p): Use
1646 positions_needed.large to track stores with non-constant widths.
1647 (all_positions_needed_p): Likewise. Take the offset and width
1648 as poly_int64s rather than ints. Assert that rhs is nonnull.
1649 (record_store): Cope with non-constant offsets and widths.
1650 Nullify the rhs of an earlier store if we can't tell which bytes
1652 (find_shift_sequence): Take the access_size and shift as poly_int64s
1654 (get_stored_val): Take the read_offset and read_width as poly_int64s
1655 rather than HOST_WIDE_INTs.
1656 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1657 non-constant offsets and widths.
1659 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1660 Alan Hayward <alan.hayward@arm.com>
1661 David Sherwood <david.sherwood@arm.com>
1663 * inchash.h (inchash::hash::add_poly_int): New function.
1664 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1665 Use poly_int64 rather than HOST_WIDE_INT.
1666 (ao_ref::max_size_known_p): New function.
1667 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1668 rather than HOST_WIDE_INT.
1669 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1670 to temporaries until its interface is adjusted to match.
1671 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1672 (aliasing_component_refs_p, decl_refs_may_alias_p)
1673 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1674 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1675 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1677 * alias.c (ao_ref_from_mem): Likewise.
1678 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1679 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1680 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1681 (maybe_trim_complex_store, maybe_trim_constructor_store)
1682 (live_bytes_read, dse_classify_store): Likewise.
1683 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1684 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1685 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1686 (vn_reference_lookup_3): Likewise.
1687 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1689 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1690 Alan Hayward <alan.hayward@arm.com>
1691 David Sherwood <david.sherwood@arm.com>
1693 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1694 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1695 instead of ranges_overlap_p.
1697 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1698 Alan Hayward <alan.hayward@arm.com>
1699 David Sherwood <david.sherwood@arm.com>
1701 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1702 sizes as poly_int64s rather than HOST_WIDE_INTs.
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 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1709 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1710 * gimple-fold.c (fold_ctor_reference): Likewise.
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 * target.def (dwarf_poly_indeterminate_value): New hook.
1717 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1718 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1719 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1720 * doc/tm.texi: Regenerate.
1721 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1722 offset as a poly_int64.
1723 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1724 offset as a poly_int64.
1725 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1726 Take the offset as a poly_int64.
1727 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1728 (frame_pointer_fb_offset): Change to a poly_int64.
1729 (int_loc_descriptor): Take the offset as a poly_int64. Use
1730 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1731 (based_loc_descr): Take the offset as a poly_int64.
1732 Use strip_offset_and_add to handle (plus X (const)).
1733 Use new_reg_loc_descr instead of an open-coded version of the
1734 previous implementation.
1735 (mem_loc_descriptor): Handle CONST_POLY_INT.
1736 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1737 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1739 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1740 Alan Hayward <alan.hayward@arm.com>
1741 David Sherwood <david.sherwood@arm.com>
1743 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1744 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1745 * inchash.h (inchash::hash::add_poly_hwi): New function.
1746 * gengtype.c (main): Register poly_int64.
1747 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1748 offset as a poly_int.
1749 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1750 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1751 offset as a poly_int64.
1752 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1753 * print-rtl.c (print_poly_int): New function.
1754 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1756 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1758 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1759 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1760 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1761 Enforce track_offset_p for parts of a PARALLEL.
1762 (vt_add_function_parameter): Use const_offset for the final
1763 offset to track. Use get_tracked_reg_offset for the parts
1766 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1767 Alan Hayward <alan.hayward@arm.com>
1768 David Sherwood <david.sherwood@arm.com>
1770 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1771 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1772 * doc/tm.texi: Regenerate.
1773 * hooks.h (hook_bool_uint_uint_true): Delete.
1774 (hook_bool_puint64_puint64_true): Declare.
1775 * hooks.c (hook_bool_uint_uint_true): Delete.
1776 (hook_bool_puint64_puint64_true): New function.
1777 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1778 instead of unsigned ints.
1779 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1780 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1782 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1783 Alan Hayward <alan.hayward@arm.com>
1784 David Sherwood <david.sherwood@arm.com>
1786 * optabs.h (expand_operand): Add an int_value field.
1787 (create_expand_operand): Add an int_value parameter and use it
1788 to initialize the new expand_operand field.
1789 (create_integer_operand): Replace with a declaration of a function
1790 that accepts poly_int64s. Move the implementation to...
1791 * optabs.c (create_integer_operand): ...here.
1792 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1793 the mode preserves the value of int_value, instead of calling
1794 const_int_operand on the rtx. Use gen_int_mode to generate
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 * dumpfile.h (dump_dec): Declare.
1802 * dumpfile.c (dump_dec): New function.
1803 * pretty-print.h (pp_wide_integer): Turn into a function and
1804 declare a poly_int version.
1805 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1807 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1808 Alan Hayward <alan.hayward@arm.com>
1809 David Sherwood <david.sherwood@arm.com>
1811 * doc/generic.texi (POLY_INT_CST): Document.
1812 * tree.def (POLY_INT_CST): New tree code.
1813 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1814 * tree-core.h (tree_poly_int_cst): New struct.
1815 (tree_node): Add a poly_int_cst field.
1816 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1817 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1818 instead of a wide_int_ref.
1819 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1821 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1822 instead of an unsigned HOST_WIDE_INT.
1823 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1824 (ptrdiff_tree_p): Declare.
1825 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1826 extern inline implementations if the target doesn't use POLY_INT_CST.
1827 (poly_int_tree_p): New function.
1828 (wi::unextended_tree): New class.
1829 (wi::int_traits <unextended_tree>): New override.
1830 (wi::extended_tree): Add a default constructor.
1831 (wi::extended_tree::get_tree): New function.
1832 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1833 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1834 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1835 (wi::tree_to_poly_wide_ref): New typedefs.
1836 (wi::ints_for): Provide overloads for extended_tree and
1838 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1839 (wi::to_wide): New functions.
1840 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1841 * tree.c (poly_int_cst_hasher): New struct.
1842 (poly_int_cst_hash_table): New variable.
1843 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1844 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1846 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1847 (init_ttree): Initialize poly_int_cst_hash_table.
1848 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1849 a poly_int64 instead of a HOST_WIDE_INT.
1850 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1851 instead of an unsigned HOST_WIDE_INT.
1852 (wide_int_to_tree): Rename to...
1853 (wide_int_to_tree_1): ...this.
1854 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1855 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1856 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1857 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1858 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1859 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1861 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1862 (streamer_read_tree_body): Likewise.
1863 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1864 (streamer_write_tree_body): Likewise.
1865 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1866 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1867 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1868 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1869 * gimple-expr.h (is_gimple_constant): Likewise.
1870 * gimplify.c (maybe_with_size_expr): Likewise.
1871 * print-tree.c (print_node): Likewise.
1872 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1873 * tree-pretty-print.c (dump_generic_node): Likewise.
1874 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1875 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1876 * tree-vrp.c (compare_values_warnv): Likewise.
1877 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1878 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1879 (force_expr_to_var_cost): Likewise.
1880 * tree-ssa-loop.c (for_each_index): Likewise.
1881 * fold-const.h (build_invariant_address, size_int_kind): Take a
1882 poly_int64 instead of a HOST_WIDE_INT.
1883 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1884 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1886 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1887 (int_const_binop_2): New function, split out from...
1888 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1889 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1890 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1891 * gimple-ssa-strength-reduction.c (slsr_process_add)
1892 (slsr_process_mul): Check for INTEGER_CSTs before using them
1894 * stor-layout.c (bits_from_bytes): New function.
1895 (bit_from_pos): Use it.
1896 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1897 by BITS_PER_UNIT to get the TYPE_SIZE.
1898 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1899 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1901 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1902 Alan Hayward <alan.hayward@arm.com>
1903 David Sherwood <david.sherwood@arm.com>
1905 * doc/rtl.texi (const_poly_int): Document. Also document the
1906 rtl sharing behavior.
1907 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1908 * rtl.h (const_poly_int_def): New struct.
1909 (rtx_def::u): Add a cpi field.
1910 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1911 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1912 (wi::rtx_to_poly_wide_ref): New typedef
1913 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1914 (poly_int_rtx_p): New functions.
1915 (trunc_int_for_mode): Declare a poly_int64 version.
1916 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1917 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1919 (strip_offset): Declare.
1920 (strip_offset_and_add): New function.
1921 * rtl.def (CONST_POLY_INT): New rtx code.
1922 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1923 (shared_const_p): Use poly_int_rtx_p.
1924 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1926 (gen_int_shift_amount): Likewise.
1927 * emit-rtl.c (const_poly_int_hasher): New class.
1928 (const_poly_int_htab): New variable.
1929 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1930 (const_poly_int_hasher::hash): New function.
1931 (const_poly_int_hasher::equal): Likewise.
1932 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1933 (immed_wide_int_const): Rename to...
1934 (immed_wide_int_const_1): ...this and make static.
1935 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1936 instead of a wide_int_ref.
1937 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1938 (gen_lowpart_common): Handle CONST_POLY_INT.
1939 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1940 * cselib.c (cselib_hash_rtx): Likewise.
1941 * dwarf2out.c (const_ok_for_output_1): Likewise.
1942 * expr.c (convert_modes): Likewise.
1943 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1944 * rtlhash.c (add_rtx): Likewise.
1945 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1946 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1947 Handle existing CONST_POLY_INT rtxes.
1948 * expmed.h (expand_shift): Take a poly_int64 instead of a
1950 * expmed.c (expand_shift): Likewise.
1951 * rtlanal.c (strip_offset): New function.
1952 (commutative_operand_precedence): Give CONST_POLY_INT the same
1953 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1955 * rtl-tests.c (const_poly_int_tests): New struct.
1956 (rtl_tests_c_tests): Use it.
1957 * simplify-rtx.c (simplify_const_unary_operation): Handle
1959 (simplify_const_binary_operation): Likewise.
1960 (simplify_binary_operation_1): Fold additions of symbolic constants
1961 and CONST_POLY_INTs.
1962 (simplify_subreg): Handle extensions and truncations of
1964 (simplify_const_poly_int_tests): New struct.
1965 (simplify_rtx_c_tests): Use it.
1966 * wide-int.h (storage_ref): Add default constructor.
1967 (wide_int_ref_storage): Likewise.
1968 (trailing_wide_ints): Use GTY((user)).
1969 (trailing_wide_ints::operator[]): Add a const version.
1970 (trailing_wide_ints::get_precision): New function.
1971 (trailing_wide_ints::extra_size): Likewise.
1973 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1974 Alan Hayward <alan.hayward@arm.com>
1975 David Sherwood <david.sherwood@arm.com>
1977 * emit-rtl.h (gen_int_shift_amount): Declare.
1978 * emit-rtl.c (gen_int_shift_amount): New function.
1979 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1981 * calls.c (shift_return_value): Likewise.
1982 * cse.c (fold_rtx): Likewise.
1983 * dse.c (find_shift_sequence): Likewise.
1984 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1985 (expand_shift, expand_smod_pow2): Likewise.
1986 * lower-subreg.c (shift_cost): Likewise.
1987 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1988 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1989 (expand_vec_perm_var): Likewise.
1990 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1991 (simplify_binary_operation_1): Likewise.
1992 * combine.c (try_combine, find_split_point, force_int_to_mode)
1993 (simplify_shift_const_1, simplify_shift_const): Likewise.
1994 (change_zero_ext): Likewise. Use simplify_gen_binary.
1996 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1998 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2000 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2002 * doc/invoke.texi (ARM Options): Document accepted extension options
2003 for -march=armv8.3-a.
2005 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2008 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2009 or --with-float=softfp, set the default CPU to arm10e.
2011 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2013 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2014 * config/visium/predicates.md (const_shift_operand): Likewise.
2015 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2016 (visium_legitimize_reload_address): Likewise.
2018 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2021 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2023 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2026 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2028 2017-12-20 Julia Koval <julia.koval@intel.com>
2030 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2031 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2032 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2033 * config.gcc: Include vpclmulqdqintrin.h.
2034 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2035 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2037 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2038 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2039 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2040 * config/i386/i386.c (isa2_opts): Add -mcx16.
2041 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2042 (ix86_option_override_internal): Move mcx16 to flags2.
2043 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2044 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2045 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2046 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2047 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2048 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2049 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2050 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2051 * doc/invoke.texi: Add -mvpclmulqdq.
2053 2017-12-20 Tom de Vries <tom@codesourcery.com>
2056 * config/i386/i386.c (ix86_static_chain): Move
2057 DECL_STATIC_CHAIN test ...
2058 * calls.c (rtx_for_static_chain): ... here. New function.
2059 * calls.h (rtx_for_static_chain): Declare.
2060 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2061 instead of targetm.calls.static_chain.
2062 * df-scan.c (df_get_entry_block_def_set): Same.
2064 2017-12-19 Tom de Vries <tom@codesourcery.com>
2066 PR tree-optimization/83493
2067 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2070 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2072 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2073 inform with hardcoded english plural handling.
2075 2017-12-18 Jeff Law <law@redhat.com>
2077 PR tree-optimization/83477
2078 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2079 a non-virtual PHI, always push a new range.
2081 2017-12-19 Martin Sebor <msebor@redhat.com>
2084 * builtins.c (compute_objsize): Handle non-constant offsets.
2086 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2088 PR tree-optimization/83444
2089 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2090 character load case, if get_stridx on MEM_REF's operand doesn't
2091 look usable, retry with get_addr_stridx.
2093 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2096 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2097 (variable_tracking_main_1): Keep markers even when VTA fails.
2100 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2101 even if there are markers before it.
2102 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2104 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2106 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2107 typical order conditions.
2108 * sel-sched.c (extract_new_fences_from): Likewise.
2109 * config/visium/constraints.md (J, K, L): Likewise.
2110 * config/visium/predicates.md (const_shift_operand): Likewise.
2111 * config/visium/visium.c (visium_legitimize_address,
2112 visium_legitimize_reload_address): Likewise.
2113 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2114 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2115 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2116 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2117 SET_DUMP_DETAIL): Likewise.
2118 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2119 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2120 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2121 avr_set_core_architecture, avr_set_current_function,
2122 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2123 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2124 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2125 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2126 avr_map_decompose, avr_fold_builtin): Likewise.
2127 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2128 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2129 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2130 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2131 * config/m32c/m32c.c (m32c_conditional_register_usage,
2132 m32c_address_cost): Likewise.
2133 * config/m32c/predicates.md (shiftcount_operand,
2134 longshiftcount_operand): Likewise.
2135 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2136 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2137 can_use_cdx_ldstw): Likewise.
2138 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2139 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2141 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2142 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2143 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2144 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2145 * config/vax/vax.c (adjacent_operands_p): Likewise.
2146 * config/ft32/constraints.md (L, b, KA): Likewise.
2147 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2149 * cfgexpand.c (expand_stack_alignment): Likewise.
2150 * gcse.c (insert_expr_in_table): Likewise.
2151 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2152 * cgraphunit.c (cgraph_node::expand): Likewise.
2153 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2154 * emit-rtl.c (add_insn): Likewise.
2155 * input.c (dump_location_info): Likewise.
2156 * passes.c (NEXT_PASS): Likewise.
2157 * read-rtl-function.c (parse_note_insn_name,
2158 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2160 * sched-rgn.c (sched_rgn_init): Likewise.
2161 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2162 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2163 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2165 * explow.c (eliminate_constant_term): Likewise.
2166 * final.c (leaf_renumber_regs_insn): Likewise.
2167 * cfgrtl.c (print_rtl_with_bb): Likewise.
2168 * genhooks.c (emit_init_macros): Likewise.
2169 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2170 * tree-data-ref.c (conflict_fn): Likewise.
2171 * selftest.c (assert_streq): Likewise.
2172 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2173 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2174 fold_binary_loc, multiple_of_p): Likewise.
2175 * reload.c (push_reload, find_equiv_reg): Likewise.
2176 * et-forest.c (et_nca, et_below): Likewise.
2177 * dbxout.c (dbxout_symbol_location): Likewise.
2178 * reorg.c (relax_delay_slots): Likewise.
2179 * dojump.c (do_compare_rtx_and_jump): Likewise.
2180 * gengtype-parse.c (type): Likewise.
2181 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2182 simplify_const_relational_operation): Likewise.
2183 * reload1.c (do_output_reload): Likewise.
2184 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2185 * gengtype.c (type_for_name): Likewise.
2186 * gimple-ssa-sprintf.c (format_directive): Likewise.
2188 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2191 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2192 accessing it. Adjust comment.
2194 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2197 * predict.c (zero_one_minusone): New function.
2198 (apply_return_prediction): Avoid return prediction for functions
2199 returning only -1, 0 and 1 values, unless they only return -1 and 0
2202 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2204 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2205 fall-through warning.
2206 (arc_compute_frame_size): Remove unused variables.
2207 (arc_print_operand): Fix fprintif format.
2208 (arc_can_follow_jump): Clean fall-through warning.
2210 2017-12-19 Marek Polacek <polacek@redhat.com>
2213 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2216 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2218 * config/arc/arc.c (overriderregs): New variable.
2219 (arc_override_options): Track fixed/call saved/call options.
2220 (arc_conditional_register_usage): Check against overrideregs
2221 variable whenever we change register properties.
2223 2017-12-19 Nathan Sidwell <nathan@acm.org>
2225 * opts.c (finish_options): Don't prefix dump_base_name if it
2226 already contains directories.
2228 2017-12-19 Martin Liska <mliska@suse.cz>
2230 PR rtl-optimization/82675
2231 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2232 more element in sbitmap.
2234 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2236 * gimplify.c (gimplify_expr): Use error_operand_p.
2238 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2241 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2244 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2246 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2247 the jump is to a label.
2249 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2251 PR tree-optimization/83444
2252 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2257 * ipa-inline.c (flatten_remove_node_hook): New function.
2258 (ipa_inline): Keep only nodes with flatten attribute at the end of
2259 the array in the order from ipa_reverse_postorder, only walk that
2260 portion of array for flattening, if there is more than one such
2261 node, temporarily register a removal hook and ignore removed nodes.
2263 PR tree-optimization/80631
2264 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2265 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2266 IFN_REDUC_MAX or IFN_REDUC_MIN.
2268 2017-12-18 Jeff Law <law@redhat.com>
2270 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2271 record symbolic equivalences from backedges in the CFG.
2274 2017-11-19 Jeff Law <law@redhat.com>
2276 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2277 of degenerates resulting from ignoring an edge.
2279 2017-12-18 Martin Sebor <msebor@redhat.com>
2282 PR tree-optimization/78450
2283 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2284 (handle_builtin_strlen): Call it.
2286 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2288 PR rtl-optimization/83424
2289 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2291 2017-12-18 Marek Polacek <polacek@redhat.com>
2294 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2295 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2298 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2300 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2301 (umaddsidi4, umaddsidi_split): Likewise.
2303 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2305 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2308 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2311 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2313 2017-12-18 Richard Biener <rguenther@suse.de>
2315 PR tree-optimization/81877
2316 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2317 (outermost_indep_loop): Adjust.
2318 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2319 (can_sm_ref_p): Adjust.
2321 2017-12-18 Richard Biener <rguenther@suse.de>
2324 * tree.c (array_at_struct_end_p): Return true if the underlying
2325 object has space for at least one element in excess of what
2326 the array domain specifies.
2328 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2330 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2331 example to avoid overfull hbox.
2332 * doc/invoke.texi (Option Summary): Add missing @gol.
2333 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2336 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2337 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2339 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2340 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2341 (AArch64 Options): Document -mverbose-cost-dump.
2342 (ARM Options): Likewise, plus -mflip-thumb.
2344 2017-12-17 Martin Sebor <msebor@redhat.com>
2347 * gimple-ssa-warn-restrict.c
2348 (builtin_memref::offset_out_of_bounds): Correct the handling of
2351 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2353 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2355 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2357 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2359 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2361 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2364 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2366 * doc/invoke.texi: Fix some typos.
2368 2017-12-16 Martin Sebor <msebor@redhat.com>
2370 PR tree-optimization/78918
2371 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2372 * builtins.c (check_sizes): Rename...
2373 (check_access): ...to this. Rename function arguments for clarity.
2374 (check_memop_sizes): Adjust names.
2375 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2376 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2377 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2378 (check_strncat_sizes, expand_builtin_strncat): Same.
2379 (expand_builtin_strncpy, expand_builtin_memset): Same.
2380 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2381 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2382 (maybe_emit_sprintf_chk_warning): Same.
2383 (expand_builtin_strcpy): Adjust.
2384 (expand_builtin_stpcpy): Same.
2385 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2386 in pointer-checking forms of memcpy, memmove, and mempcpy.
2387 (gcall_to_tree_minimal, max_object_size): Define new functions.
2388 * builtins.h (max_object_size): Declare.
2389 * calls.c (alloc_max_size): Call max_object_size instead of
2390 hardcoding ssizetype limit.
2391 (get_size_range): Handle new argument.
2392 * calls.h (get_size_range): Add a new argument.
2393 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2394 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2395 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2397 (gimple_fold_builtin_memory_chk): Same.
2398 (gimple_fold_builtin_stxcpy_chk): New function.
2399 * gimple-ssa-warn-restrict.c: New source.
2400 * gimple-ssa-warn-restrict.h: New header.
2401 * gimple.c (gimple_build_call_from_tree): Propagate location.
2402 * passes.def (pass_warn_restrict): Add new pass.
2403 * tree-pass.h (make_pass_warn_restrict): Declare.
2404 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2406 (handle_builtin_strcat): Same.
2407 (strlen_optimize_stmt): Rename...
2408 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2409 stpncpy, strncpy, and their checking forms.
2411 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2413 PR rtl-optimization/82849
2414 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2415 and get_max_loop_iterations_int.
2417 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2418 Alan Hayward <alan.hayward@arm.com>
2419 David Sherwood <david.sherwood@arm.com>
2421 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2422 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2423 size as a poly_uint64.
2424 (mode_for_vector, mode_for_int_vector): Take the number of vector
2425 elements as a poly_uint64.
2426 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2427 size as a poly_uint64.
2428 (mode_for_vector, mode_for_int_vector): Take the number of vector
2429 elements as a poly_uint64.
2431 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2432 Alan Hayward <alan.hayward@arm.com>
2433 David Sherwood <david.sherwood@arm.com>
2435 * machmode.h (MACRO_MODE): New macro.
2436 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2437 * targhooks.c (default_libcall_value, default_secondary_reload)
2438 (default_memory_move_cost, default_register_move_cost)
2439 (default_class_max_nregs): Likewise.
2441 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2442 Alan Hayward <alan.hayward@arm.com>
2443 David Sherwood <david.sherwood@arm.com>
2445 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2446 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2448 * genautomata.c (main): Likewise.
2449 * genconditions.c (write_header): Likewise.
2450 * genemit.c (main): Likewise.
2451 * genextract.c (print_header): Likewise.
2452 * genopinit.c (main): Likewise.
2453 * genoutput.c (output_prologue): Likewise.
2454 * genpeep.c (main): Likewise.
2455 * genpreds.c (write_insn_preds_c): Likewise.
2456 * genrecog.c (writer_header): Likewise.
2457 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2458 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2459 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2460 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2461 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2462 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2463 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2464 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2465 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2466 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2467 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2468 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2469 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2470 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2471 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2472 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2473 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2474 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2475 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2476 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2477 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2478 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2479 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2480 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2481 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2482 * config/darwin.c (IN_TARGET_CODE): Likewise.
2483 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2484 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2485 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2486 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2487 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2488 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2489 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2490 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2491 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2492 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2493 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2494 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2495 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2496 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2497 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2498 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2499 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2500 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2501 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2502 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2503 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2504 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2505 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2506 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2507 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2508 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2509 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2510 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2511 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2512 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2513 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2514 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2515 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2516 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2517 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2518 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2519 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2520 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2521 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2522 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2523 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2524 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2525 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2526 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2527 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2528 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2529 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2530 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2531 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2532 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2533 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2534 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2535 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2536 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2537 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2538 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2539 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2540 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2541 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2542 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2543 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2544 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2545 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2546 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2547 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2548 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2549 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2550 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2551 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2552 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2553 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2554 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2555 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2556 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2557 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2558 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2559 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2560 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2561 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2562 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2563 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2564 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2565 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2566 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2567 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2568 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2569 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2570 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2571 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2572 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2573 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2574 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2575 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2576 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2577 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2578 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2579 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2580 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2581 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2582 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2583 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2584 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2585 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2586 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2587 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2588 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2590 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2591 Alan Hayward <alan.hayward@arm.com>
2592 David Sherwood <david.sherwood@arm.com>
2594 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2597 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2598 Alan Hayward <alan.hayward@arm.com>
2599 David Sherwood <david.sherwood@arm.com>
2601 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2602 * doc/md.texi (vec_series@var{m}): Document.
2603 * tree.def (VEC_SERIES_EXPR): New tree code.
2604 * tree.h (build_vec_series): Declare.
2605 * tree.c (build_vec_series): New function.
2606 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2607 * tree-pretty-print.c (dump_generic_node): Likewise.
2608 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2609 * tree-inline.c (estimate_operator_cost): Likewise.
2610 * expr.c (expand_expr_real_2): Likewise.
2611 * optabs-tree.c (optab_for_tree_code): Likewise.
2612 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2613 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2614 * expmed.c (make_tree): Handle VEC_SERIES.
2615 * optabs.def (vec_series_optab): New optab.
2616 * optabs.h (expand_vec_series_expr): Declare.
2617 * optabs.c (expand_vec_series_expr): New function.
2618 * tree-vect-generic.c (expand_vector_operations_1): Check that
2619 the operands also have vector type.
2621 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2622 Alan Hayward <alan.hawyard@arm.com>
2623 David Sherwood <david.sherwood@arm.com>
2625 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2626 (VEC_COND_EXPR): Add missing @tindex.
2627 * doc/md.texi (vec_duplicate@var{m}): Document.
2628 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2629 * tree.c (build_vector_from_val): Add stubbed-out handling of
2630 variable-length vectors, using VEC_DUPLICATE_EXPR.
2631 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2632 * cfgexpand.c (expand_debug_expr): Likewise.
2633 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2634 * tree-inline.c (estimate_operator_cost): Likewise.
2635 * tree-pretty-print.c (dump_generic_node): Likewise.
2636 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2637 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2638 (test_vec_duplicate_folding): New function.
2639 (fold_const_c_tests): Call it.
2640 * optabs.def (vec_duplicate_optab): New optab.
2641 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2642 * optabs.h (expand_vector_broadcast): Declare.
2643 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2644 vec_duplicate_optab.
2645 * expr.c (store_constructor): Try using vec_duplicate_optab for
2647 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2649 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2652 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2653 div/mod latencies a bit.
2655 2017-12-15 Jeff Law <law@redhat.com>
2657 PR tree-optimization/36550
2658 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2659 (mark_threaded_blocks): Rewrite code to avoid block copying when
2660 optimizing for size. Don't pessimize blocks which will be
2661 copied, but all the statements will be dead.
2663 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2665 PR tree-optimization/81165
2666 * tree-ssa-threadupdate.c (uses_in_bb): New.
2667 (estimate_threading_killed_stmts): New.
2668 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2669 * tree-ssa-threadedge.c
2670 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2673 2017-12-15 Jeff Law <law@redhat.com>
2675 PR tree-optimization/83410
2676 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2677 threads when parallelizing loops.
2679 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2681 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2683 * config/alpha/alpha.c (vms_attribute_table): Swap
2684 affects_type_identity and handler fields, adjust comments.
2685 * config/mips/mips.c (mips_attribute_table): Likewise.
2686 * config/visium/visium.c (visium_attribute_table): Likewise.
2687 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2688 * config/microblaze/microblaze.c (microblaze_attribute_table):
2690 * config/spu/spu.c (spu_attribute_table): Likewise.
2691 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2692 * config/arc/arc.c (arc_attribute_table): Likewise.
2693 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2694 * config/v850/v850.c (v850_handle_interrupt_attribute,
2695 v850_handle_data_area_attribute): Formatting fixes.
2696 (v850_attribute_table): Swap affects_type_identity and handler
2697 fields, adjust comments.
2698 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2699 * config/arm/arm.c (arm_attribute_table): Likewise.
2700 * config/avr/avr.c (avr_attribute_table): Likewise.
2701 * config/s390/s390.c (s390_attribute_table): Likewise.
2702 * config/sh/sh.c (sh_attribute_table): Likewise.
2703 * config/i386/i386.c (ix86_handle_cconv_attribute,
2704 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2705 (ix86_attribute_table): Swap affects_type_identity and handler
2706 fields, adjust comments.
2707 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2708 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2709 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2710 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2711 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2712 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2713 * config/rx/rx.c (rx_attribute_table): Likewise.
2714 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2715 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2716 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2717 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2718 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2719 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2720 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2721 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2722 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2723 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2724 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2726 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2728 (register_scoped_attributes, decl_attributes): Formatting fixes.
2730 PR tree-optimization/83269
2731 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2732 subtraction in arg0's type if type is signed and arg0 is unsigned.
2736 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2737 convert? on @0 instead of convert. Check type of @1, not @0.
2738 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2739 convert? on @0 instead of convert. Check type of @1, not @0.
2740 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2741 only optimize if either both @1 and @2 types are narrower
2742 precision, or both are wider or equal precision, and in the former
2743 case only if both have undefined overflow.
2745 2017-12-15 Richard Biener <rguenther@suse.de>
2748 * internal-fn.def (IFN_NOP): Add.
2749 * internal-fn.c (expand_NOP): Do nothing.
2750 * lto-streamer-in.c (input_function): Instead of removing
2751 sanitizer calls replace them with IFN_NOP calls.
2753 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2754 Alan Hayward <alan.hayward@arm.com>
2755 David Sherwood <david.sherwood@arm.com>
2757 * dse.c (store_info, read_info_type): Replace begin and end with
2759 (print_range): New function.
2760 (set_all_positions_unneeded, any_positions_needed_p)
2761 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2763 (record_store): Likewise. Optimize the case in which all positions
2765 (get_stored_val): Replace read_begin and read_end with read_offset
2767 (replace_read): Update call accordingly.
2769 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2771 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2772 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2773 (loop_cand::loop_cand): Initialize above members.
2774 (loop_cand::supported_operations): Delete.
2775 (loop_cand::can_interchange_p): Inline above function.
2776 (loop_cand::classify_simple_reduction): Record number of constant
2777 initialized simple reductions.
2778 (should_interchange_loops): New parameters. Check stmt cost of loops
2780 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2781 Update call to should_interchange_loops.
2782 (should_interchange_loop_nest): Update call to
2783 should_interchange_loops.
2785 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2788 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2789 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2790 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2792 2017-12-15 Julia Koval <julia.koval@intel.com>
2794 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2795 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2797 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2798 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2799 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2801 2017-12-15 Julia Koval <julia.koval@intel.com>
2803 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2804 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2805 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2806 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2807 _mm_aesenc_epi128): New intrinsics.
2809 2017-12-15 Julia Koval <julia.koval@intel.com>
2811 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2812 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2814 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2815 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2816 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2818 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2820 * gimple-ssa-strength-reduction.c (analyze_increments):
2821 Distinguish replacement costs for constant strides from those for
2824 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2826 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2828 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2830 * doc/invoke.texi: Document -Wcast-function-type.
2831 * recog.h (stored_funcptr): Change signature.
2832 * tree-dump.c (dump_node): Avoid warning.
2833 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2835 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2838 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2841 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2843 PR tree-optimization/83312
2844 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2845 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2846 "val" param, and to cope with arbitrary basic blocks.
2847 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2848 handle NULL_TREE for "val", dropping "bb" param.
2849 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2850 drop "bb" param. Handle NULL_TREE for "val".
2851 (find_case_label_for_value): Make "switch_stmt" param const.
2852 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2854 (vrp_prop::check_all_array_refs): Reimplement as...
2855 (check_array_bounds_dom_walker::before_dom_children): ...this new
2856 vfunc. Replace linear search through BB block list, excluding
2857 those with non-executable in-edges via dominator walk.
2859 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2861 * config/arm/arm.opt (mverbose-cost-dump): New option.
2862 * config/arm/arm.c (arm_rtx_costs): Use it.
2864 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2867 * reload1.c (emit_input_reload_insns): Skip debug markers.
2869 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2871 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2875 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2876 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2877 * config/sh/sh.c: Skip debug insns besides notes.
2878 * config/sh/sh.md: Likewise.
2879 * config/sh/sh_treg_combine.cc: Likewise.
2880 * config/sh/sync.md: Likewise.
2882 2017-12-14 Tom de Vries <tom@codesourcery.com>
2884 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2885 Add item for weak_undefined.
2887 2017-12-14 Richard Biener <rguenther@suse.de>
2889 PR tree-optimization/67842
2890 * sese.h (bb_in_region): Remove #if 0'ed code.
2892 2017-12-14 Richard Biener <rguenther@suse.de>
2894 PR tree-optimization/83326
2895 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2896 may_be_zero parameter and handle it by not marking the first
2897 peeled copy as not exiting the loop.
2898 (try_peel_loop): Likewise.
2899 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2900 to handle the case of constant or zero iterations and perform
2901 loop header copying on-the-fly.
2903 2017-12-14 Richard Biener <rguenther@suse.de>
2905 PR tree-optimization/83418
2906 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2907 Instead of asserting we don't get unfolded comparisons deal with them.
2909 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2912 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2913 basic blocks. Assert debug bind insns don't appear outside of bbs,
2914 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2917 PR tree-optimization/83198
2918 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2919 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2920 value if arg is a REAL_CST with incompatible type.
2922 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2923 Bin Cheng <bin.cheng@arm.com>
2926 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2928 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2931 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2933 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2934 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2936 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2937 Alan Hayward <alan.hayward@arm.com>
2938 David Sherwood <david.sherwood@arm.com>
2940 * poly-int.h: New file.
2941 * poly-int-types.h: Likewise.
2942 * coretypes.h: Include them.
2943 (POLY_INT_CONVERSION): Define.
2944 * target.def (estimated_poly_value): New hook.
2945 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2946 * doc/tm.texi: Regenerate.
2947 * doc/poly-int.texi: New file.
2948 * doc/gccint.texi: Include it.
2949 * doc/rtl.texi: Describe restrictions on subreg modes.
2950 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2951 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2952 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2953 * targhooks.h (default_estimated_poly_value): Declare.
2954 * targhooks.c (default_estimated_poly_value): New function.
2955 * target.h (estimated_poly_value): Likewise.
2956 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2957 (wi::unary_traits): Delete.
2958 (wi::binary_traits::signed_shift_result_type): Define for
2959 offset_int << HOST_WIDE_INT, etc.
2960 (generic_wide_int::operator <<=): Define for all types and use
2961 wi::lshift instead of <<.
2962 (wi::hwi_with_prec): Add a default constructor.
2963 (wi::ints_for): New class.
2964 (operator <<): Define for all wide-int types.
2965 (operator /): New function.
2966 (operator %): Likewise.
2967 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2968 (ASSERT_MAYBE_NE_AT): New macros.
2970 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2971 Dominik Vogt <vogt@linux.vnet.ibm.com>
2974 * emit-rtl.c (init_emit): Remove ??? comment.
2975 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2976 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2977 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2978 alignment of 3 virtual registers to BITS_PER_WORD.
2980 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2982 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2984 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2985 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2987 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2990 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2992 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2994 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2995 with the exception of debug begin stmt markers appear before
2999 * final.c (rest_of_handle_final): Call variable_tracking_main only
3000 if !flag_var_tracking.
3002 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3003 Jakub Jelinek <jakub@redhat.com>
3007 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3008 labels debug stmts that can only appear after labels.
3010 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3012 PR rtl-optimization/82398
3013 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3014 EXPR_USEFULNESS in priority comparison.
3016 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3018 PR rtl-optimization/83393
3019 * combine.c (move_deaths): If reg_stat points to a too new insn in
3020 last_death, do not use it: find the proper insn instead.
3022 2017-12-12 Jeff Law <law@redhat.com>
3024 PR tree-optimization/83298
3025 PR tree-optimization/83362
3026 PR tree-optimization/83383
3027 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3028 push_value_range a public interface. Add new argument to
3029 record_ranges_from_stmt.
3030 * gimple-ssa-evrp-analyze.c
3031 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3032 Update comments. Handle recording temporary equivalences.
3033 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3034 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3035 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3036 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3037 gimple-ssa-evrp-analyze.h.
3038 (record_temporary_equivalences_from_phis): Add new argument. When
3039 the PHI arg is an SSA_NAME, set the result's range to the range
3041 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3042 from statements too.
3043 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3044 Pass it down to children as needed.
3045 (thread_outgoing_edges): Likewise.
3046 (thread_across_edge): Likewise. Push/pop range state as needed.
3047 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3049 2017-12-12 Julia Koval <julia.koval@intel.com>
3051 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3052 (PTA_CANNONLAKE): Remove PTA_CLWB.
3054 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3057 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3058 (vcondv2div2df): Likewise.
3059 (vconduv2dfv2di): Likewise.
3061 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3063 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3064 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3065 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3066 ATTR_MATHFN_FPROUNDING_ERRNO.
3068 2017-12-12 Richard Biener <rguenther@suse.de>
3070 PR tree-optimization/83385
3071 * graphite-scop-detection.c (get_order, order): Remove.
3072 (bb_to_rpo): New global.
3074 (build_scops): Sort pbbs in RPO order.
3076 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3078 * combine.c (simplify_set): Do not transform subregs to zero_extends
3079 if the destination mode is a vector mode.
3081 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3083 PR tree-optimization/83359
3084 * tree-cfg.h (fold_loop_internal_call): Declare.
3085 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3086 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3087 (find_loop_dist_alias): New function.
3088 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3089 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3092 PR tree-optimization/80631
3093 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3094 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3095 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3096 hardcoding zero as the value if COND_EXPR is never true. For
3097 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3098 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3099 hardcoding MAX_EXPR as the reduction operation.
3100 (is_nonwrapping_integer_induction): Allow negative step.
3101 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3102 vect_create_epilog_for_reduction, if no value is suitable, don't
3103 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3105 2017-12-12 Richard Biener <rguenther@suse.de>
3107 PR tree-optimization/81889
3108 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3109 range info from the non-wrapping IV instead of just the range
3112 2017-12-12 Julia Koval <julia.koval@intel.com>
3114 * config.gcc: Add vaesintrin.h.
3115 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3117 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3118 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3120 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3121 * config/i386/immintrin.h: Include vaesintrin.h.
3122 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3123 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3124 _mm_aesdec_epi128): New intrinsics.
3126 2017-12-12 Julia Koval <julia.koval@intel.com>
3128 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3129 OPTION_MASK_ISA_VAES_UNSET): New.
3130 (ix86_handle_option): Handle -mvaes.
3131 * config/i386/cpuid.h: Define bit_VAES.
3132 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3133 * config/i386/i386-c.c (__VAES__): New.
3134 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3135 (ix86_valid_target_attribute_inner_p): Ditto.
3136 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3137 * config/i386/i386.opt: Add -mvaes.
3138 * doc/invoke.texi: Ditto.
3140 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3142 * debug.h (gcc_debug_hooks): Add inline_entry.
3143 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3144 * debug.c (do_nothing_debug_hooks): Likewise.
3145 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3146 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3147 (dwarf2_lineno_debug_hooks): Likewise.
3149 * common.opt (gstatement-frontiers): New, setting
3150 debug_nonbind_markers_p.
3151 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3152 * toplev.c (process_options): Autodetect value for debug statement
3154 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3155 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3157 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3158 markers. Integrate source bind into debug stmt expand loop.
3159 (pass_expand::execute): Check debug marker limit. Avoid deep
3160 TER and expand debug locations for debug bind insns only.
3161 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3163 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3164 * final.c (reemit_insn_block_notes): Take current block from
3165 nonbind markers. Declare note where it's first set.
3166 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3167 begin stmt markers if enabled.
3168 (notice_source_line): Handle nonbind markers. Fail if their
3169 location is unknown or that of builtins.
3170 (rest_of_handle_final): Convert begin stmt markers to notes if
3171 var-tracking didn't run.
3172 (rest_of_clean_state): Skip begin stmt markers.
3173 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3175 * function.c (allocate_struct_function): Set begin_stmt_markers.
3176 * function.h (struct function): Add debug_marker_count counter
3177 and debug_nonbind_markers flag.
3178 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3179 * gimple-low.c (lower_function_body): Adjust
3180 debug_nonbind_markers.
3181 (lower_stmt): Drop or skip gimple debug stmts.
3182 (lower_try_catch): Skip debug stmts.
3183 * gimple.c (gimple_build_debug_begin_stmt): New.
3184 (gimple_copy): Increment debug_marker_count if copying one.
3185 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3186 * gimplify.c (rexpr_location): New.
3187 (rexpr_has_location): New.
3188 (warn_switch_unreachable_r): Handle gimple debug stmts.
3189 (shortcut_cond_r): Call expr_location.
3191 (find_goto_label): New.
3192 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3194 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3196 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3197 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3198 (LANG_HOOKS_INITIALIZER): ... this.
3199 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3200 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3202 (update_ebb_live_info): Skip debug insn markers.
3203 * lra.c (debug_insn_static_data): Rename to...
3204 (debug_bind_static_data): ... this.
3205 (debug_marker_static_data): New.
3206 (lra_set_insn_recog_data): Select one of the above depending
3208 (lra_update_isn_regno_info): Don't assume debug insns have
3210 (push_insns): Skip debug insns.
3211 * lto-streamer-in.c (input_function): Drop debug stmts
3212 depending on active options. Adjust debug_nonbind_markers.
3213 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3214 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3215 begin stmt marker notes.
3216 (print_insn): Likewise.
3217 * recog.c (extract_insn): Recognize rtl for debug markers.
3218 * rtl.def (DEBUG_MARKER): New.
3219 * tree-inline.c: Include params.h.
3220 (remap_gimple_stmt): Handle nonbind markers.
3221 (maybe_move_debug_stmts_to_successors): Likewise.
3222 (copy_debug_stmt): Likewise.
3223 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3224 markers regardless of no side effects.
3225 (tsi_link_before): Don't update container's side effects when adding
3226 a begin stmt marker.
3227 (tsi_link_after): Likewise.
3228 (expr_first): Skip begin stmt markers.
3229 (expr_last): Likewise.
3230 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3231 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3232 Disregard nonbind markers.
3233 * tree.c (make_node_stat): Don't set side effects for begin stmt
3235 (build1_stat): Likewise.
3236 * tree.def (DEBUG_BEGIN_STMT): New.
3237 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3238 * var-tracking.c (delete_debug_insns): Renamed to...
3239 (delete_vta_debug_insns): ... this.
3240 (reemit_marker_as_note): New.
3241 (vt_initialize): Reemit markers.
3242 (delete_vta_debug_insns): Likewise.
3243 (vt_debug_insns_local): Reemit or delete markers.
3244 (variable_tracking_main_1): Likewise.
3245 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3246 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3247 (gimple_debug_nonbind_marker_p): New.
3248 (gimple_build_debug_bind): Adjust.
3249 (gimple_build_debug_begin_stmt): New.
3250 * doc/invoke.texi (max-debug-marker-count): New param.
3251 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3252 (debug_parameter_ref, debug_marker): New.
3253 (NOTE_INSN_BEGIN_STMT): New.
3254 (DEBUG_INSN): Describe begin stmt markers.
3256 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3257 without debug insns, we wouldn't, but clean up debug insns
3258 after a control flow insn nevertheless.
3260 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3261 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3262 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3263 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3264 (rtl_tidy_fallthru_edge): Likewise.
3265 (rtl_verify_fallthru): Likewise.
3266 (rtl_verify_bb_layout): Likewise.
3267 (skip_insns_after_block): Likewise.
3268 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3269 * dwarf2out.c: Include print-rtl.h.
3270 (dwarf2out_next_real_insn): New.
3271 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3272 Dump debug binds in asm comments.
3273 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3274 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3275 callers to use gsi_start_nondebug_bb instead.
3276 (gsi_after_labels): Skip gimple debug stmts.
3277 (gsi_start_nondebug): New.
3278 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3279 (proper_loop_form_for_interchange): Adjust.
3280 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3281 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3282 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3283 (collect_fallthrough_labels): Likewise.
3284 (should_warn_for_implicit_fallthrough): Likewise.
3285 (warn_implicit_fallthrough_r): Likewise.
3286 (expand_FALLTHROUGH_r): Likewise.
3287 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3288 (graphite_copy_stmts_from_block): Skip nonbind markers.
3289 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3290 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3291 * jump.c (clean_barriers): Skip debug insns.
3292 * omp-expand.c (expand_parallel_call): Skip debug insns.
3293 (expand_task_call): Likewise.
3294 (remove_exit_barrier): Likewise.
3295 (expand_omp_taskreg): Likewise.
3296 (expand_omp_for_init_counts): Likewise.
3297 (expand_omp_for_generic): Likewise.
3298 (expand_omp_for_static_nochunk): Likewise.
3299 (expand_omp_for_static_chunk): Likewise.
3300 (expand_omp_simd): Likewise.
3301 (expand_omp_taskloop_for_outer): Likewise.
3302 (expand_omp_taskloop_for_inner): Likewise.
3303 (expand_oacc_for): Likewise.
3304 (expand_omp_sections): Likewise.
3305 (expand_omp_single): Likewise.
3306 (expand_omp_synch): Likewise.
3307 (expand_omp_atomic_load): Likewise.
3308 (expand_omp_atomic_store): Likewise.
3309 (expand_omp_atomic_fetch_op): Likewise.
3310 (expand_omp_atomic_pipeline): Likewise.
3311 (expand_omp_atomic_mutex): Likewise.
3312 (expand_omp_target): Likewise.
3313 (grid_expand_omp_for_loop): Likewise.
3314 (grid_expand_target_grid_body): Likewise.
3315 (build_omp_regions_1): Likewise.
3316 * omp-low.c (check_combined_parallel): Skip debug stmts.
3317 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3318 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3319 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3321 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3322 (make_edges): Likewise.
3323 (cleanup_dead_labels): Likewise.
3324 (gimple_can_merge_blocks_p): Likewise.
3325 (stmt_starts_bb_p): Likewise.
3326 (gimple_block_label): Likewise.
3327 (gimple_redirect_edge_and_branch): Likewise.
3328 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3330 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3332 * tree-pretty-print (print_declaration): Omit initializer in slim
3334 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3336 (eliminate_unnecessary_stmts): Stabilize block removal order.
3337 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3338 * var-tracking.c (get_first_insn): New.
3339 (vt_emit_notes): Call it.
3340 (vt_initialize): Walk any insns before the first BB.
3341 (delete_debug_insns): Likewise.
3343 * gimple.h (enum gimple_debug_subcode): Add
3344 GIMPLE_DEBUG_BEGIN_STMT.
3345 (gimple_debug_begin_stmt_p): New.
3346 (gimple_debug_nonbind_marker_p): New.
3347 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3348 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3349 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3350 * insn-notes.def (BEGIN_STMT): New.
3351 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3352 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3353 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3354 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3355 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3356 (INSN_DEBUG_MARKER_KIND): New.
3357 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3358 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3359 (INSN_VAR_LOCATION_PTR): New.
3360 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3362 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3363 deps only if debug bind insns are enabled.
3364 (pass_expand::execute): Avoid deep TER and expand
3365 debug locations for debug bind insns only.
3366 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3367 debug stmts special handling down to debug bind stmts.
3368 * combine.c (try_combine): Narrow debug insns special handling
3369 down to debug bind insns.
3370 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3371 Narrow debug insns preexisting special handling down to debug
3373 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3374 handling down to debug bind insns.
3375 * function.c (instantiate_virtual_regs): Skip debug markers,
3376 adjust handling of debug binds.
3377 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3378 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3379 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3380 insns to debug bind insns.
3381 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3382 special handling of debug stmts to debug bind stmts.
3383 * ipa-split.c (split_function): Likewise.
3384 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3385 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3387 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3388 * regrename.c (build_def_use): Likewise.
3389 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3390 (pass_cprop_hardreg): Narrow special casing of debug insns to
3392 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3393 * reload1.c (reload): Likewise.
3394 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3395 casing of debug stmts to debug bind stmts.
3396 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3397 * ssa-iterators.h (num_imm_uses): Likewise.
3398 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3399 debug stmts to debug bind stmts.
3400 * tree-inline.c (tree_function_versioning): Narrow special casing
3401 of debug stmts to debug bind stmts.
3402 * tree-loop-distribution.c (generate_loops_for_partition):
3403 Narrow special casing of debug stmts to debug bind stmts.
3404 * tree-sra.c (analyze_access_subtree): Narrow special casing
3405 of debug stmts to debug bind stmts.
3406 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3407 stmts to debug bind stmts.
3408 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3409 casing of debug stmts to debug bind stmts.
3410 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3411 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3412 casing of debug stmts to debug bind stmts.
3413 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3415 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3416 debug stmts to debug bind stmts.
3417 (gimple_replace_ssa_lhs): Likewise.
3418 (insert_debug_temp_for_var_def): Likewise.
3419 (insert_debug_temps_for_defs): Likewise.
3420 (reset_debug_uses): Likewise.
3421 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3422 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3423 (adjust_debug_stmts): Likewise.
3424 (adjust_phi_and_debug_stmts): Likewise.
3425 (vect_do_peeling): Likewise.
3426 * tree-vect-loop.c (vect_transform_loop): Likewise.
3427 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3428 * var-tracking.c (adjust_mems): Narrow special casing of debug
3429 insns to debug bind insns.
3430 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3431 (compute_bb_dataflow, vt_find_locations): Likewise.
3432 (vt_expand_loc, emit_notes_for_changes): Likewise.
3433 (vt_init_cfa_base): Likewise.
3434 (vt_emit_notes): Likewise.
3435 (vt_initialize): Likewise.
3436 (vt_finalize): Likewise.
3438 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3439 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3440 (next_nonnote_nondebug_insn_bb): New.
3441 (prev_nonnote_nondebug_insn_bb): New.
3442 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3443 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3444 (prev_nonnote_nondebug_insn_bb): Declare.
3445 (next_nonnote_nondebug_insn_bb): Declare.
3446 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3447 * cfgrtl.c (get_last_bb_insn): Likewise.
3448 * lra.c (push_insns): Likewise.
3450 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3453 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3455 * selftest.c (selftest::test_runner::test_runner): New ctor.
3456 (selftest::test_runner::~test_runner): New dtor.
3457 * selftest.h (class selftest::test_runner): New class.
3459 2017-12-11 Carl Love <cel@us.ibm.com>
3461 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3462 vec_extract_fp32_from_shortl]): Add #defines.
3463 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3464 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3465 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3466 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3467 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3468 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3469 * doc/extend.texi: Add documentation for the added builtins.
3471 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3473 PR rtl-optimization/80693
3474 PR rtl-optimization/81019
3475 PR rtl-optimization/81020
3476 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3477 are not mentioned in i3. Place the REG_UNUSED note on i2,
3478 possibly modified to REG_DEAD, if it did not originate in i3.
3480 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3482 * recog.c (store_data_bypass_p_1): New function.
3483 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3484 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3486 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3488 PR rtl-optimization/83361
3489 * ifcvt.c (if_convert): Call fixup_partitions.
3491 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3493 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3494 early folding of splat_u{8,16,32}.
3496 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3498 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3499 output_operand_lossage first argument with capital letter.
3500 (aarch64_override_options): Don't start error and sorry first argument
3501 with capital letter.
3503 2017-12-11 Andi Kleen <ak@linux.intel.com>
3505 PR gcov-profile/83355
3506 * auto-profile.c (string_table::get_index_by_decl): Don't
3507 recurse when abstract origin points to itself.
3509 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3511 PR tree-optimization/83320
3512 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3513 (prune_datarefs_not_in_loop): Ditto.
3515 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3517 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3519 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3521 PR tree-optimization/83337
3522 * gimple-loop-interchange.cc (compute_access_stride): Handle
3523 bitfield DRs properly.
3525 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3527 PR tree-optimization/83338
3528 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3529 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3530 vector of unsigned integers to vector of signed integers.
3532 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3534 PR rtl-optimization/83317
3535 * lra-constraints.c (process_address_1): Add insn code check.
3537 2017-12-08 Michael Matz <matz@suse.de>
3539 Fix PR tree-optimization/83323
3540 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3541 head-controlled loops and loop BBs.
3542 * common.opt (funroll-and-jam): Remove, instead ...
3543 (floop-unroll-and-jam): ... reuse this option.
3544 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3545 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3546 (-floop-unroll-and-jam): ... this option.
3548 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3550 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3551 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3552 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3554 2017-12-08 Richard Biener <rguenther@suse.de>
3557 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3558 handle accesses outside of zero-sized vars.
3560 2017-12-08 Martin Jambor <mjambor@suse.cz>
3562 PR tree-optimization/83141
3563 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3564 test for MEM_REFs implicitely changing types with padding. Remove
3566 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3568 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3570 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3572 * config/avr/avr.c (avr_attribute_table): Likewise.
3573 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3574 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3575 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3576 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3577 of struct attribute_spec.
3578 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3580 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3583 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3585 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3588 * profile-count.c (profile_count::from_gcov_type): Move from
3589 profile-count.h; handle overflow.
3590 * profile-count.h (profile_count::from_gcov_type): Move offline.
3592 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3594 PR rtl-optimization/83304
3595 * combine.c (move_deaths): If we do not know where a register died,
3598 2017-12-08 Richard Biener <rguenther@suse.de>
3600 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3601 Provide -fopt-info-loop feedback when we interchange in a nest.
3603 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3605 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3606 for armv6 ARM CPU IDs.
3608 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3610 * common/config/arm/arm-common.c: Include <algorithm>.
3611 (INCLUDE_VECTOR): Define.
3612 (compare_opt_names): New function.
3613 (arm_rewrite_selected_arch): Only strip out extensions that can be
3614 expressed through -mfpu. Sort the remaining extensions
3617 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3619 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3620 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3621 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3622 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3624 2017-06-08 Tristan Gingold <gindold@adacore.com>
3627 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3628 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3630 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3632 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3633 of struct attribute_spec.
3635 2017-12-08 Julia Koval <julia.koval@intel.com>
3637 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3638 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3639 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3640 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3641 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3642 _mm_maskz_dpwssds_epi32): Ditto.
3644 2017-12-08 Richard Biener <rguenther@suse.de>
3646 PR tree-optimization/81303
3647 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3648 conditions try to create a comparison vector type matching
3649 the data vector type.
3650 (vectorizable_condition): Adjust.
3651 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3652 Leave invariant conditions alone in case we can vectorize those.
3654 2017-12-08 Julia Koval <julia.koval@intel.com>
3656 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3657 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3658 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3659 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3660 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3662 2017-12-08 Julia Koval <julia.koval@intel.com>
3664 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3665 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3666 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3667 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3668 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3669 _mm_maskz_dpbusds_epi32): New intrinsics.
3671 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3673 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3675 (ld<bh>io_signed): Likewise.
3676 (st<bhw>io): Likewise.
3677 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3678 SMALL_INT12 constant integer operand.
3680 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3682 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3684 (nios2_large_constant_allowed): ...this. Adjust uses.
3685 (nios2_plus_symbolic_constant_p): Rename to...
3686 (nios2_plus_large_constant_p): ...this. Adjust uses.
3687 (nios2_legitimate_address_p): Correct CONST_INT handling.
3688 (nios2_symbolic_memory_operand_p): Rename to...
3689 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3690 (nios2_large_constant_p): Check for large constant integers too.
3691 (nios2_split_large_constant): Handle constant integers.
3692 (nios2_split_symbolic_memory_operand): Rename to...
3693 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3694 (nios2_legitimize_constant_address): Handle constant integers.
3695 (r0rel_constant_p): Handle small constant integers.
3696 (nios2_print_operand_address): Handle r0-relative integer addresses.
3697 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3698 * config/nios2/nios2.md: Adjust for renamed functions.
3700 2017-12-07 Andrew Waterman <andrew@sifive.com>
3702 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3703 (TARGET_HAVE_SRODATA_SECTION): New define.
3704 (riscv_select_section): New function.
3706 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3707 Alexander Monakov <amonakov@ispras.ru>
3708 Jakub Jelinek <jakub@redhat.com>
3711 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3713 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3715 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3716 a bool success value. Don't call output_operand_lossage here.
3717 (aarch64_print_ldpstp_address): Return a bool success value.
3718 (aarch64_print_operand_address): Call output_addr_const if
3719 aarch64_print_address_internal fails.
3720 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3721 'y'; call output_operand_lossage instead. Call output_operand_lossage
3722 if aarch64_print_ldpstp_address fails.
3724 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3726 * tree-vector-builder.h
3727 (tree_vector_builder::binary_encoded_nelts): Declare.
3728 * tree-vector-builder.c
3729 (tree_vector_builder::binary_encoded_nelts): New function.
3730 * fold-const.c (negate_expr_p): Likewise.
3731 (operand_equal_p, fold_checksum_tree): Likewise.
3732 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3733 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3734 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3735 (uniform_vector_p): Likewise.
3736 * varasm.c (const_hash_1, compare_constant): Likewise.
3737 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3738 (valid_lattice_transition): Operate directly on the VECTOR_CST
3740 * ipa-icf.c: Include tree-vector-builder.h.
3741 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3742 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3744 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3746 * tree.c (build_vector): Delete.
3747 * tree.h (build_vector): Make static and move into the self-testing
3750 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3752 * vector-builder.h (vector_builder::derived): New const overload.
3753 (vector_builder::elt): New function.
3754 * tree-vector-builder.h (tree_vector_builder::type): New function.
3755 (tree_vector_builder::apply_step): Declare.
3756 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3758 * gimple-fold.h (tree_vector_builder): Declare.
3759 (gimple_build_vector): Take a tree_vector_builder instead of a
3760 type and vector of elements.
3761 * gimple-fold.c (gimple_build_vector): Likewise.
3762 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3764 (get_initial_defs_for_reduction): Likewise.
3765 (vectorizable_induction): Likewise.
3767 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3769 * tree-vector-builder.h
3770 (tree_vector_builder::new_binary_operation): Declare.
3771 * tree-vector-builder.c
3772 (tree_vector_builder::new_binary_operation): New function.
3773 * fold-const.c (fold_relational_const): Use it.
3774 (const_binop): Likewise. Check that both input vectors have
3775 the same number of elements, thus excluding things like WIDEN_SUM.
3776 Check whether it is possible to operate directly on the encodings
3779 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3781 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3782 new_unary_operation, operating only on the encoded elements.
3783 (const_unop): Likewise.
3784 (exact_inverse): Likewise.
3785 (distributes_over_addition_p): New function.
3786 (const_binop): Use tree_vector_builder and new_unary_operation
3787 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3788 on the encoded elements unless the encoding is strided and the
3789 operation does not distribute over addition.
3790 (fold_convert_const): Use tree_vector_builder and
3791 new_unary_operation. Operate only on the encoded elements
3792 for truncating integer conversions, or for non-stepped encodings.
3794 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3796 * config/sparc/sparc.c: Include tree-vector-builder.h.
3797 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3798 * expmed.c: Include tree-vector-builder.h.
3799 (make_tree): Use tree_vector_builder instead of build_vector.
3800 * fold-const.c: Include tree-vector-builder.h.
3801 (const_binop): Use tree_vector_builder instead of build_vector.
3802 (const_unop): Likewise.
3803 (native_interpret_vector): Likewise.
3804 (fold_vec_perm): Likewise.
3805 (fold_ternary_loc): Likewise.
3806 * gimple-fold.c: Include tree-vector-builder.h.
3807 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3809 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3810 (simplify_vector_constructor): Use tree_vector_builder instead
3812 * tree-vect-generic.c: Include tree-vector-builder.h.
3813 (add_rshift): Use tree_vector_builder instead of build_vector.
3814 (expand_vector_divmod): Likewise.
3815 (optimize_vector_constructor): Likewise.
3816 * tree-vect-loop.c: Include tree-vector-builder.h.
3817 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3818 of build_vector. Explicitly use a stepped encoding for
3820 * tree-vect-slp.c: Include tree-vector-builder.h.
3821 (vect_get_constant_vectors): Use tree_vector_builder instead
3823 (vect_transform_slp_perm_load): Likewise.
3824 (vect_schedule_slp_instance): Likewise.
3825 * tree-vect-stmts.c: Include tree-vector-builder.h.
3826 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3827 (vect_gen_perm_mask_any): Likewise.
3828 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3829 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3831 (build_vector_from_val): Likewise. Explicitly use a duplicate
3834 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3836 * doc/generic.texi (VECTOR_CST): Describe new representation of
3838 * vector-builder.h: New file.
3839 * tree-vector-builder.h: Likewise.
3840 * tree-vector-builder.c: Likewise.
3841 * Makefile.in (OBJS): Add tree-vector-builder.o.
3842 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3843 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3844 (tree_vector): Change the number of elements to
3845 vector_cst_encoded_nelts.
3846 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3847 (VECTOR_CST_ELTS): Delete.
3848 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3849 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3850 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3851 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3852 (VECTOR_CST_ENCODED_ELT): Likewise.
3853 (vector_cst_encoded_nelts): New function.
3854 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3855 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3856 (vector_cst_int_elt, vector_cst_elt): Declare.
3857 * tree.c: Include tree-vector-builder.h.
3858 (tree_code_size): Abort if passed VECTOR_CST.
3859 (tree_size): Update for new VECTOR_CST layout.
3860 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3861 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3862 (build_vector): Use tree_vector_builder.
3863 (vector_cst_int_elt, vector_cst_elt): New functions.
3864 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3865 encoded elements and then create the vector in the canonical form.
3866 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3867 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3868 (tree_c_tests): Call test_vector_cst_patterns.
3869 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3871 (hash_tree): Likewise.
3872 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3873 (streamer_write_tree_header): Likewise.
3874 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3875 (streamer_alloc_tree): Likewise. Update call to make_vector.
3876 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3878 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3880 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3881 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3882 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3884 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3885 Richard Biener <rguenther@suse.de>
3887 PR tree-optimization/81303
3888 * Makefile.in (gimple-loop-interchange.o): New object file.
3889 * common.opt (floop-interchange): Reuse the option from graphite.
3890 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3891 -floop-interchange and mention it for -O3.
3892 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3893 * gimple-loop-interchange.cc: New file.
3894 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3895 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3896 * passes.def (pass_linterchange): New pass.
3897 * timevar.def (TV_LINTERCHANGE): New time var.
3898 * tree-pass.h (make_pass_linterchange): New declaration.
3899 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3900 interchange. Record IV before/after increment in new parameters.
3901 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3902 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3904 (check_reduction_path): ...New function here.
3905 * tree-vectorizer.h (check_reduction_path): New declaration.
3907 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3910 PR rtl-optimization/80818
3911 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3912 always early clobbered.
3913 * lra-lives.c (process_bb_lives): Check input hard regs for early
3914 clobbered non-operand hard reg.
3916 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3919 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3920 types_compatible_p, just that TYPE_MODE is the same.
3922 2017-12-07 Martin Sebor <msebor@redhat.com>
3925 * attribs.c (empty_attribute_table): Initialize new member of
3926 struct attribute_spec.
3927 (decl_attributes): Add argument. Handle mutually exclusive
3928 combinations of attributes.
3929 (selftests::test_attribute_exclusions): New function.
3930 (selftests::attribute_c_tests): Ditto.
3931 * attribs.h (decl_attributes): Add default argument.
3932 * selftest.h (attribute_c_tests): Declare.
3933 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3934 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3936 * doc/extend.texi (Common Function Attributes): Update const and pure.
3937 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3938 of struct attribute_spec.
3939 * config/arc/arc.c (arc_attribute_table): Same.
3940 * config/arm/arm.c (arm_attribute_table): Same.
3941 * config/avr/avr.c ( avr_attribute_table): Same.
3942 * config/bfin/bfin.c (bfin_attribute_table): Same.
3943 * config/cr16/cr16.c (cr16_attribute_table): Same.
3944 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3945 * config/h8300/h8300.c (h8300_attribute_table): Same.
3946 * config/i386/i386.c (ix86_attribute_table): Same.
3947 * config/ia64/ia64.c (ia64_attribute_table): Same.
3948 * config/m32c/m32c.c (m32c_attribute_table): Same.
3949 * config/m32r/m32r.c (m32r_attribute_table): Same.
3950 * config/m68k/m68k.c (m68k_attribute_table): Same.
3951 * config/mcore/mcore.c (mcore_attribute_table): Same.
3952 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3953 * config/mips/mips.c (mips_attribute_table): Same.
3954 * config/msp430/msp430.c (msp430_attribute_table): Same.
3955 * config/nds32/nds32.c (nds32_attribute_table): Same.
3956 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3957 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3958 * config/rl78/rl78.c (rl78__attribute_table): Same.
3959 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3960 * onfig/rx/rx.c (rx_attribute_table): Same.
3961 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3962 * config/sh/sh.c (sh_attribute_table): Same.
3963 * config/sparc/sparc.c (sparc_attribute_table): Same.
3964 * config/spu/spu.c (spu_attribute_table): Same.
3965 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3966 * config/v850/v850.c (v850_attribute_table): Same.
3967 * config/visium/visium.c (visium_attribute_table): Same.
3969 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3972 * config/arm/arm.c (INCLUDE_STRING): Define.
3973 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3974 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3977 2017-12-07 Michael Matz <matz@suse.de>
3979 Add unroll and jam pass
3981 * gimple-loop-jam.c: New file.
3982 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3983 * common.opt (funroll-and-jam): New option.
3984 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3985 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3986 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3987 * passes.def: Add pass_loop_jam.
3988 * timevar.def (TV_LOOP_JAM): Add.
3989 * tree-pass.h (make_pass_loop_jam): Declare.
3990 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3991 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3992 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3993 to flow_loop_tree_node_add.
3994 (duplicate_subloops, copy_loops_to): Append to sibling list.
3995 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3996 * doc/invoke.texi (-funroll-and-jam): Document new option.
3997 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3999 2017-12-07 Richard Biener <rguenther@suse.de>
4001 PR tree-optimization/83296
4002 PR tree-optimization/67769
4003 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4004 flow sensitive info in an unrelated BB.
4005 (value_replacement): Use reset_flow_sensitive_info.
4006 (minmax_replacement): Reset flow sensitive info on the def
4007 we move. Do not reset flow sensitive info in the whole BB
4008 we move the stmt to.
4009 (abs_replacement): Likewise.
4011 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4014 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4015 rs6000_cpu to the given -mcpu=, or to the default processor.
4017 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4019 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4020 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4021 instead of rs6000_cpu_attr.
4022 (rs6000_adjust_cost): Ditto.
4023 (is_microcoded_insn): Ditto.
4024 (rs6000_adjust_priority): Ditto.
4025 (rs6000_issue_rate): Ditto.
4026 (rs6000_use_sched_lookahead): Ditto.
4027 (rs6000_use_sched_lookahead_guard): Ditto.
4028 (rs6000_sched_reorder): Ditto.
4029 (force_new_group): Ditto.
4030 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4031 (group_ending_nop): Ditto.
4033 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4035 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4036 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4037 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4038 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4040 (rs6000_reassociation_width): Ditto.
4041 (rs6000_emit_epilogue): Ditto.
4042 (rs6000_adjust_cost): Ditto.
4043 (is_microcoded_insn): Ditto.
4044 (is_cracked_insn): Ditto.
4045 (rs6000_adjust_priority): Ditto.
4046 (rs6000_sched_reorder): Ditto.
4047 (rs6000_sched_reorder2): Ditto.
4048 (insn_must_be_first_in_group): Ditto.
4049 (insn_must_be_last_in_group): Ditto.
4050 (rs6000_register_move_cost): Ditto.
4051 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4054 2017-12-07 Julia Koval <julia.koval@intel.com>
4056 * config.gcc: Add vaesintrin.h.
4057 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4059 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4060 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4062 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4063 * config/i386/immintrin.h: Include vaesintrin.h.
4064 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4065 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4066 _mm_aesdec_epi128): New intrinsics.
4068 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4070 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4071 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4072 c-family/c-spellcheck.cc.
4073 (best_macro_match::best_macro_match): Likewise.
4075 (struct edit_distance_traits<cpp_hashnode *>): Move to
4076 c-family/c-spellcheck.h.
4077 (class best_macro_match): Likewise.
4079 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4081 PR tree-optimization/83293
4082 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4083 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4084 might insert into empty bb.
4087 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4088 simplify for plus with :c added, and pointer_plus without that.
4089 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4090 with undefined overflow and the conversion is not widening,
4091 perform negation in utype and only convert to type afterwards.
4092 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4093 simplify for plus with :c added, and pointer_plus without that.
4094 If type is integral with undefined overflow and the conversion is
4095 not widening, perform minus in utype and only convert to type
4096 afterwards. Move the last pointer_diff_expr simplify into the
4099 2017-12-06 Martin Sebor <msebor@redhat.com>
4101 PR tree-optimization/82646
4102 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4103 strncpy, not maxlen.
4105 2017-12-06 Martin Sebor <msebor@redhat.com>
4107 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4110 PR tree-optimization/83075
4111 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4112 strncat/strncpy don't change length of source string.
4114 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4117 2017-11-29 Martin Aberg <maberg@gaisler.com>
4119 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4120 to prevent b2bst errata sequence.
4121 (sqrtdf2_fix): Likewise.
4123 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4125 PR tree-optimization/81945
4126 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4127 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4128 to dest_cfun has orig_loop_num set, either remap it to the new
4129 loop number if the loop got moved too, or clear it.
4131 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4133 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4135 (thunderx2t99_nothing): New insn reservation.
4136 (thunderx2t99_mrs): New insn reservation.
4137 (thunderx2t99_multiple): New insn reservation.
4138 (thunderx2t99_alu_basi): Add bfx to reservation.
4139 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4141 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4144 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4146 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4148 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4149 tree-ssa-pre.c::remove_dead_inserted_code.
4150 * tree-ssa-dce.h: New file.
4151 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4152 (remove_dead_inserted_code): Move and rename to function
4153 tree-ssa-dce.c::simple_dce_from_worklist.
4154 (pass_pre::execute): Update use.
4156 2017-12-05 Richard Biener <rguenther@suse.de>
4158 PR tree-optimization/83277
4159 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4160 to code-gen liveout vars.
4162 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4164 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4165 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4166 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4169 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4171 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4173 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4174 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4177 2017-12-05 Richard Biener <rguenther@suse.de>
4179 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4180 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4181 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4182 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4184 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4186 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4188 2017-12-05 Martin Liska <mliska@suse.cz>
4189 Jakub Jelinek <jakub@redhat.com>
4191 * doc/invoke.texi: Document the options.
4192 * flag-types.h (enum sanitize_code): Add
4193 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4194 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4195 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4196 * opts.c: Define new sanitizer options.
4197 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4198 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4200 2017-12-05 Julia Koval <julia.koval@intel.com>
4202 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4203 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4204 (ix86_handle_option): Handle -mavx512vnni.
4205 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4206 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4207 * config/i386/i386-c (__AVX512VNNI__): New.
4208 * config/i386/i386.c (ix86_target_string): Handle new option.
4209 (ix86_valid_target_attribute_inner_p): Handle new option.
4210 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4211 * config/i386/i386.opt (mavx512vnni): New option.
4213 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4216 * athlon.md: Disable for generic.
4217 * haswell.md: Enable for generic.
4218 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4219 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4221 (ix86_adjust_cost): Move generic to haswell path.
4223 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4225 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4226 instead of MEM_P in a couple more places. Fix formatting issues.
4228 2017-12-04 Jim Wilson <jimw@sifive.com>
4230 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4231 instead of GP_REG_LAST-1.
4232 (riscv_adjust_libcall_cfi_prologue): Likewise.
4233 (riscv_adjust_libcall_cri_epilogue): Likewise.
4234 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4237 2017-12-04 Luis Machado <luis.machado@linaro.org>
4239 * ipa-pure-const.c (check_decl): Add missing newline.
4240 (state_from_flags): Likewise.
4242 2017-12-04 Jeff Law <law@redhat.com>
4244 PR tree-optimizatin/78496
4245 * gimple-ssa-evrp-analyze.h
4246 (evrp_range_analyzer::get_vr_values): Simplify.
4247 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4248 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4249 and gimple-ssa-evrp-analyze.h.
4250 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4251 (simplify_stmt_for_jump_threading): Copy a blob of code from
4252 tree-vrp.c to use ranges to simplify statements.
4253 (dom_opt_dom_walker::before_dom_children): Call
4254 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4255 (dom_opt_dom_walker::after_dom_children): Similarly for
4256 evrp_range_analyzer::leave.
4257 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4260 * gimple-ssa-evrp-analyze.c
4261 (evrp_range_analyzer::extract_range_from_stmt): Always use
4262 vr_values::update_value_range so preexisting range info is
4263 medged with new range info, even if the new range is VR_VARYING.
4265 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4267 * combine.c: Adjust comment.
4268 (use_crosses_set_p): Delete.
4269 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4270 (try_combine): Ditto.
4272 2017-12-04 Richard Biener <rguenther@suse.de>
4274 PR tree-optimization/83255
4275 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4276 Re-add zero-iteration check.
4278 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4280 PR rtl-optimization/83245
4281 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4282 hard registers as earlyclobber, also if not in an asm.
4284 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4288 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4291 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4292 rs6000_cpu based on cpu_index, not tune_index.
4294 2017-12-04 Richard Biener <rguenther@suse.de>
4296 PR tree-optimization/83238
4297 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4298 code match comment, rejecting invalid SESE regions.
4300 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4302 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4303 require base operand is a REG_POINTER prior to reload on targets
4304 with non-equivalent space registers.
4306 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4308 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4309 (update_specialized_profile): Fix updating of counts.
4310 (perhaps_add_new_callers): Likewise.
4312 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4315 * x86-tune.def: Remove obsolette FIXMEs.
4316 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4317 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4318 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4320 (X86_TUNE_PAD_RETURNS): Disable for generic.
4322 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4324 PR tree-optimization/83170
4325 PR tree-optimization/83241
4326 * gimple-ssa-store-merging.c
4327 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4328 gimple_vuse (ins_stmt) in case it has changed.
4329 (imm_store_chain_info::output_merged_store): Likewise.
4331 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4335 * tree-cfg.c (pass_warn_function_return::execute): Handle
4336 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4337 with BUILTINS_LOCATION.
4341 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4342 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4343 instead of DECL_MODE.
4345 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4346 last argument to rtx pointer.
4347 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4348 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4349 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4350 depending on the chosen ISAs.
4351 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4352 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4353 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4355 * config/i386/sse.md (mov<mode>_internal): Likewise.
4356 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4358 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4360 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4362 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4364 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4365 parameters from prototype.
4366 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4367 parameters. Don't print a comment.
4368 (emit_fusion_gpr_load): Adjust.
4369 (emit_fusion_load_store): Adjust.
4370 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4371 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4372 comment on the second line.
4374 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4377 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4378 rs6000_cpu based on cpu_index, not tune_index.
4380 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4382 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4383 which_alternative instead of which_alternative + 1.
4384 (output_asm_insn): Print an extra tab if the template is short.
4386 2017-12-01 Jim Wilson <jimw@sifive.com>
4388 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4390 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4391 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4393 * doc/tm.texi: Regenerate.
4395 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4398 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4399 whether we can allocate pseudos before trying to fix an address.
4400 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4401 memory address is indexed or indirect.
4402 (floatuns_<mode>si2_hw2): Likewise.
4404 2017-12-01 Jason Merrill <jason@redhat.com>
4406 * Makefile.in (TAGS): Add c-family/*.cc.
4408 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4410 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4411 (call_value_insn): Likewise.
4412 (sibcall_insn): Likewise.
4413 (sibcall_value_insn): Likewise.
4414 (movsi_aarch64): Likewise.
4415 (movdi_aarch64): Likewise.
4416 (add_losym_): Likewise.
4417 (ldr_got_small_): Likewise.
4418 (ldr_got_small_sidi): Likewise.
4419 (ldr_got_small_28k_): Likewise.
4420 (ldr_got_small_28k_sidi): Likewise.
4421 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4422 Move output_addr_const to symbolic case. Add error check.
4424 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4426 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4427 * config/s390/s390.md: Disable MVC merging peephole if it would
4428 disable operand forwarding.
4429 (new peephole2): Split MVCs if it would turn them into up to 2
4432 2017-12-01 Richard Biener <rguenther@suse.de>
4434 PR tree-optimization/83232
4435 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4436 detection of same access. Instead of breaking the group here
4437 do not consider the duplicate. Add comment explaining real fix.
4439 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4441 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4443 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4445 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4447 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4449 * function.h (struct function): Remove cilk_frame_decl,
4450 is_cilk_function and calls_cilk_spawn fields.
4451 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4453 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4454 cilk_elemental field.
4455 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4456 * target.def: Adjust comment.
4457 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4458 Don't test cilk_elemental.
4460 PR tree-optimization/83233
4461 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4462 bswap_stat name for the struct.
4465 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4466 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4467 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4468 SWITCH_BREAK_LABEL_P set on the label.
4469 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4470 added for default case if it was missing and not all cases covered.
4471 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4472 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4474 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4475 to avoid -Wimplicit-fallthrough warning.
4476 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4477 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4479 PR tree-optimization/83221
4480 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4482 (init_reassoc): Formatting fix.
4485 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4486 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4490 * tree-cfg.c: Include asan.h.
4491 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4493 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4495 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4497 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4499 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4500 vpcmpeqd instruction.
4502 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4504 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4505 registers type for 128bit mode.
4507 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4509 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4511 * gengtype-state.c (read_state_pair): Likewise.
4512 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4513 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4514 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4515 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4516 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4517 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4518 * builtins.c (fold_builtin_3): Likewise.
4519 * graphite-scop-detection.c
4520 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4521 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4524 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4527 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4528 parameter. Return true if ignore_decl_rtl_set_p is true and other
4529 conditions are satisfied.
4530 * asan.h (asan_protect_global): Add new parameter.
4531 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4532 to asan_protect_global calls.
4534 2017-11-30 Jim Wilson <jimw@sifive.com>
4536 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4537 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4538 -mstrict-align, add info on default value. Delete redundant lines for
4539 -mabi. Add missing -mexplicit-relocs docs.
4541 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4543 * config/arc/arc.md (trap): New pattern.
4545 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4547 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4548 instruction to end into a delay slot.
4549 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4550 can be placed into a delay slot against reg_note.
4552 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4554 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4555 labels number of usages.
4557 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4559 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4561 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4563 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4566 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4567 multiplication by power of 2 constant into two shifts + comparison.
4569 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4572 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4574 2017-11-30 Richard Biener <rguenther@suse.de>
4576 PR tree-optimization/83202
4577 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4578 allow_peel argument and guard peeling.
4579 (canonicalize_loop_induction_variables): Likewise.
4580 (canonicalize_induction_variables): Pass false.
4581 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4582 peeling from cunrolli.
4584 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4586 * combine.c (try_combine): Print a message to dump file whenever
4587 I0, I1, or I2 cannot be combined into I3.
4589 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4591 PR rtl-optimization/83156
4592 PR rtl-optimization/82621
4593 * combine.c (try_combine): Don't split an I2 if one of the dests is
4594 set again before I3. Allow unused dests.
4596 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4598 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4600 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4602 PR rtl-optimization/80818
4603 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4604 recursively. Use insn code for clobber.
4605 (lra_set_insn_recog_data): Pass the new arg to
4606 collect_non_operand_hard_regs.
4607 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4609 (lra_update_insn_regno_info): Pass insn to
4610 add_regs_to_insn_regno_info.
4612 2017-11-29 Jim Wilson <jimw@sifive.com>
4613 Andrew Waterman <andrew@sifive.com>
4615 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4616 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4617 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4618 use SINGLE_SHIFT_COST.
4619 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4620 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4622 2017-11-29 Julia Koval <julia.koval@intel.com>
4624 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4625 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4626 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4627 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4628 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4629 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4630 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4631 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4632 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4633 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4634 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4635 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4636 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4637 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4638 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4639 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4640 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4641 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4642 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4643 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4644 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4645 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4646 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4647 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4648 __builtin_ia32_vpshldv_v2di_mask,
4649 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4650 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4651 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4653 2017-11-29 Julia Koval <julia.koval@intel.com>
4655 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4656 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4657 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4658 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4659 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4660 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4661 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4662 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4663 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4664 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4665 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4666 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4667 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4668 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4669 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4670 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4671 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4672 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4673 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4674 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4675 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4677 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4679 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4680 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4683 2017-11-29 Martin Aberg <maberg@gaisler.com>
4685 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4686 to prevent b2bst errata sequence.
4687 (sqrtdf2_fix): Likewise.
4689 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4691 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4692 (div_sqrt_insn_p): New function.
4693 (sparc_do_work_around_errata): Insert NOP instructions to
4694 prevent sequences that could trigger the TN-0013 errata for
4695 certain LEON3 processors.
4696 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4697 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4698 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4699 (in_branch_delay): Prevent div and sqrt in delay slot if
4701 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4703 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4705 * config/sparc/sparc.c (atomic_insn_p): New function.
4706 (sparc_do_work_around_errata): Insert NOP instructions to
4707 prevent sequences that could trigger the TN-0010 errata for
4709 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4710 instruction referable in atomic_insns_p.
4712 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4714 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4715 (atomic_compare_and_swap_leon3_1): Likewise.
4718 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4720 * config/sparc/sparc.c (fpop_insn_p): New function.
4721 (sparc_do_work_around_errata): Insert NOP instructions to
4722 prevent sequences that could trigger the TN-0012 errata for
4724 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4725 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4726 (in_branch_annul_delay): Prevent floating-point instructions
4727 in delay slot of annulled integer branch.
4729 2017-11-29 Richard Biener <rguenther@suse.de>
4731 PR tree-optimization/83202
4732 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4734 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4735 nodes vectorized to the same stmts multiple times.
4736 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4737 (vect_analyze_slp_instance): Adjust.
4738 (scalar_stmts_to_slp_tree_map_t): New typedef.
4739 (vect_schedule_slp_instance): Add a map recording the SLP node
4740 representing the vectorized stmts for a set of scalar stmts.
4741 Avoid code-generating redundancies.
4742 (vect_schedule_slp): Allocate map and pass it down.
4744 2017-11-29 Nathan Sidwell <nathan@acm.org>
4747 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4750 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4752 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4753 cases for printing LDP/STP memory addresses.
4754 (aarch64_print_address_internal): Renamed from
4755 aarch64_print_operand_address, added parameter, add Pmode check.
4756 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4757 (aarch64_print_operand_address): Indirect to
4758 aarch64_print_address_internal.
4759 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4760 'y' operand output specifier.
4762 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4765 * tree.c (build_simple_mem_ref_loc): Handle
4766 get_addr_base_and_unit_offset returning a MEM_REF.
4769 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4772 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4775 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4777 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4779 (layout::move_to_column): Likewise.
4781 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4784 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4785 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4787 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4788 Martin Jambor <mjambor@suse.cz>
4791 * tree.h (expr_type_first_operand_type_p): Declare
4792 * tree.c (expr_type_first_operand_type_p): New function.
4793 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4794 (ipa_value_from_jfunc): Adjust declaration.
4795 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4796 Use it as result type for arithmetics, unless it is NULL in which case
4797 be more conservative.
4798 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4799 ipa_get_jf_pass_through_result.
4800 (propagate_vals_across_pass_through): Likewise.
4801 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4802 is to propagate_vals_across_pass_through.
4803 (propagate_constants_across_call): Pass PARM_TYPE to
4804 propagate_scalar_across_jump_function.
4805 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4806 ipa_value_from_jfunc.
4807 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4808 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4809 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4810 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4811 target_type, pass it to ipa_value_from_jfunc.
4812 (update_indirect_edges_after_inlining): Pass parameter type to
4813 try_make_edge_direct_simple_call.
4815 2017-11-28 Jeff Law <law@redhat.com>
4817 * gimple-ssa-evrp-analyze.c
4818 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4819 refine ranges if scev_initialized_p returns true.
4820 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4822 2017-11-28 Julia Koval <julia.koval@intel.com>
4824 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4825 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4826 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4827 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4828 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4829 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4830 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4831 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4832 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4833 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4834 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4835 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4836 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4837 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4838 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4839 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4840 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4841 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4842 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4843 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4844 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4845 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4847 2017-11-28 Julia Koval <julia.koval@intel.com>
4849 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4850 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4851 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4852 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4853 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4854 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4855 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4856 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4857 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4858 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4859 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4860 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4861 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4862 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4863 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4864 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4865 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4866 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4867 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4868 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4869 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4870 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4871 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4872 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4873 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4874 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4875 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4876 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4877 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4878 __builtin_ia32_vpshld_v2di_mask): New builtins.
4879 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4880 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4882 2017-11-28 Richard Biener <rguenther@suse.de>
4884 PR tree-optimization/80776
4885 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4887 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4889 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4890 If the incoming edge is an effective fallthru because the other
4891 edge only reaches a __builtin_unreachable () then record ranges
4892 derived from the controlling condition in SSA info.
4893 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4894 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4896 2017-11-28 Olivier Hainque <hainque@adacore.com>
4898 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4899 on mingw build hosts.
4901 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4904 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4906 2017-11-28 Richard Biener <rguenther@suse.de>
4909 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4910 copies generated from memcpy use a character array as reference
4913 2017-11-28 Julia Koval <julia.koval@intel.com>
4914 Sebastian Peryt <sebastian.peryt@intel.com>
4916 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4917 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4918 cilk-common.o, cilk.h, cilk-common.c): Remove.
4920 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4921 * builtins.c (is_builtin_name): Remove cilkplus condition.
4922 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4923 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4924 cilk-builtins.def, cilkplus.def): Remove.
4925 * cif-code.def (CILK_SPAWN): Remove.
4926 * cilk-builtins.def: Delete.
4927 * cilk-common.c: Ditto.
4929 * cilkplus.def: Ditto.
4930 * config/darwin.h (fcilkplus): Delete.
4931 * cppbuiltin.c: Ditto.
4932 * doc/extend.texi: Remove cilkplus doc.
4933 * doc/generic.texi: Ditto.
4934 * doc/invoke.texi: Ditto.
4935 * doc/passes.texi: Ditto.
4936 * gcc.c (fcilkplus): Remove.
4937 * gengtype.c (cilk.h): Remove.
4938 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4940 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4942 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4944 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4945 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4946 cilkplus conditions.
4947 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4948 inline_read_section): Ditto.
4949 * ipa-inline-analysis.c (cilk.h): Remove.
4950 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4951 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4952 append_linker_options): Remove condition for fcilkplus.
4953 * lto/lto-lang.c (cilk.h): Remove.
4954 (lto_init): Remove condition for fcilkplus.
4955 * omp-expand.c (expand_cilk_for_call): Delete.
4956 (expand_omp_taskreg, expand_omp_for_static_chunk,
4957 expand_omp_for): Remove cilkplus
4959 (expand_cilk_for): Delete.
4960 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4961 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4962 execute_lower_omp, diagnose_sb_0): Ditto.
4963 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4964 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4965 * tree-nested.c: Ditto.
4966 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4967 (dump_generic_node): Ditto.
4968 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4969 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4971 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4973 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4975 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4976 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4977 for vector mode and !TARGET_SIMD.
4979 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4981 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4983 * tree.h (SWITCH_LABELS): Remove.
4984 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4985 assert SWITCH_BODY is non-NULL.
4986 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4988 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4990 PR tree-optimization/80788
4991 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4992 has TREE_OVERFLOW set, call drop_tree_overflow.
4994 2017-11-28 Richard Biener <rguenther@suse.de>
4996 PR tree-optimization/83158
4997 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4999 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5002 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5003 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5005 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5007 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5008 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5010 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5013 * varasm.c (compare_constant): Take the mode of the constants into
5014 account when comparing floating point constants.
5016 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5018 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5019 from explicit instantiation of debug_helper.
5020 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5022 2017-11-27 Richard Biener <rguenther@suse.de>
5024 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5027 2017-11-27 Richard Biener <rguenther@suse.de>
5029 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5030 (type_hash_canon): Also clear int_cst_hash_table entry for
5032 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5034 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5036 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5037 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5038 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5040 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5042 * hash-map.h (gt_cleare_cache): Avoid UB.
5044 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5046 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5047 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5048 (decide_unroll_runtime_iterations): Likewise.
5049 (decide_unroll_stupid): Likewise.
5051 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5054 * config/i386/i386.md: Add a loop with incssp.
5056 2017-11-27 Martin Jambor <mjambor@suse.cz>
5058 PR tree-optimization/81248
5059 * tree-sra.c (splice_param_accesses): Remove size check.
5060 (decide_one_param_reduction): Fix size check.
5061 * gimple-pretty-print.c (dump_profile): Silence warning.
5062 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5064 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5067 * dbxout.c (lastlineno): New variable.
5068 (dbx_debug_hooks): Use dbxout_switch_text_section as
5069 switch_text_section debug hook.
5070 (dbxout_function_end): Switch to current_function_section
5071 rather than function_section. If crtl->has_bb_partition,
5072 output just one N_FUN, depending on in_cold_section_p.
5073 (dbxout_source_line): Remember last lineno in lastlineno.
5074 (dbxout_switch_text_section): New function.
5075 (dbxout_function_decl): Adjust dbxout_block caller.
5076 (dbx_block_with_cold_children): New function.
5077 (dbxout_block): Return true if any LBRAC/RBRAC have been
5078 emitted. Use dbx_block_with_cold_children at depth == 0
5079 in second partition. Add PARENT_BLOCKNUM argument, pass
5080 it optionally adjusted to children. Output LBRAC/RBRAC
5081 around recursive call only if the block is in the current
5082 partition, if not and anything was output, emit empty
5084 * final.c (final_scan_insn): Compute cold_function_name
5085 before calling switch_text_section debug hook. Call
5086 that hook even if dwarf2out_do_frame if not emitting
5090 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5091 TREE_READONLY decls.
5093 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5095 PR rtl-optimization/82488
5096 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5098 2017-11-26 Julia Koval <julia.koval@intel.com>
5100 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5102 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5105 2017-11-26 Julia Koval <julia.koval@intel.com>
5107 * config/i386/driver-i386.c (host_detect_local_cpu):
5108 Detect skylake-avx512.
5110 2017-11-26 Julia Koval <julia.koval@intel.com>
5112 * config.gcc: Add -march=cannonlake.
5113 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5114 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5115 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5116 (PTA_CANNONLAKE): New.
5117 (processor_target_table): Add cannonlake.
5118 (ix86_option_override_internal): Ditto.
5119 (fold_builtin_cpu): Ditto.
5120 (get_builtin_code_for_version): Handle cannonlake.
5121 (M_INTEL_COREI7_CANNONLAKE): New.
5122 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5123 * doc/invoke.texi: Add -march=cannonlake.
5125 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5127 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5128 (try_init_one_plugin): Alternative implementation for MinGW.
5129 * Makefile.in (plugin_implib): New.
5130 (gengtype-lex.c): Fix broken AIX workaround.
5131 * configure: Regenerate.
5132 * doc/plugins.texi: Document support for MinGW.
5134 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5136 PR rtl-optimization/81553
5137 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5138 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5139 is a shift where C1 has different mode than the whole shift, use C1's
5140 mode for MULT rather than the shift's mode.
5143 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5144 builtins not enabled in the currently selected ISA.
5146 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5148 PR tree-optimization/71026
5149 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5150 (insert_reciprocals): Change to insert reciprocals before a division
5151 by a square and to insert the square of a reciprocal.
5152 (execute_cse_reciprocals_1): Change to consider division by a square.
5153 (register_division_in): Add importance parameter.
5155 2017-11-24 Richard Biener <rguenther@suse.de>
5157 PR tree-optimization/82402
5158 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5159 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5161 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5163 * match.pd (0-ptr): New transformation.
5165 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5168 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5169 when skipping checking.
5171 2017-11-24 Richard Biener <rguenther@suse.de>
5173 PR tree-optimization/83128
5174 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5175 (vn_reference_lookup_3): Likewise.
5177 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5180 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5181 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5184 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5185 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5186 check for the latter with gimple_call_builtin_p. Do not handle
5187 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5189 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5191 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5194 2017-11-23 Julia Koval <julia.koval@intel.com>
5196 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5197 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5198 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5199 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5200 _mm512_maskz_expandloadu_epi16): New intrinsics.
5201 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5202 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5203 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5204 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5205 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5206 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5207 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5208 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5209 _mm256_maskz_expandloadu_epi8): New intrinsics.
5210 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5211 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5212 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5213 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5214 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5215 * config/i386/sse.md (VI248_VLBW): New iterator.
5216 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5218 2017-11-23 Julia Koval <julia.koval@intel.com>
5220 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5221 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5222 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5223 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5224 _mm512_mask_compressstoreu_epi16): New.
5225 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5226 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5227 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5228 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5229 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5230 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5231 _mm256_mask_compressstoreu_epi8): New.
5232 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5233 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5234 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5235 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5236 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5237 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5238 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5239 __builtin_ia32_compresshi128_mask,
5240 __builtin_ia32_compressstoreuqi512_mask,
5241 __builtin_ia32_compressstoreuhi512_mask,
5242 __builtin_ia32_compressstoreuqi256_mask,
5243 __builtin_ia32_compressstoreuqi128_mask,
5244 __builtin_ia32_compressstoreuhi256_mask,
5245 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5246 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5248 (ix86_expand_special_args_builtin): Handle new types.
5249 (s4fma_expand): Handle new builtin array.
5250 * config/i386/immintrin.h: Include new headers.
5251 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5252 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5254 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5256 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5257 fixes. Declare temp and g variables at the top in order to avoid
5258 {} in most of the cases.
5260 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5262 * match.pd (ptr-0): New transformation.
5264 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5266 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5267 (struct addr_mode_cost_table): New.
5268 (struct tune_params): Add field addr_mode_costs.
5269 * config/arm/arm.c (generic_addr_mode_costs): New.
5270 (arm_slowmul_tune): Initialise addr_mode_costs field.
5271 (arm_fastmul_tune): Likewise.
5272 (arm_strongarm_tune): Likewise.
5273 (arm_xscale_tune): Likewise.
5274 (arm_9e_tune): Likewise.
5275 (arm_marvell_pj4_tune): Likewise.
5276 (arm_v6t2_tune): Likewise.
5277 (arm_cortex_tune): Likewise.
5278 (arm_cortex_a8_tune): Likewise.
5279 (arm_cortex_a7_tune): Likewise.
5280 (arm_cortex_a15_tune): Likewise.
5281 (arm_cortex_a35_tune): Likewise.
5282 (arm_cortex_a53_tune): Likewise.
5283 (arm_cortex_a57_tune): Likewise.
5284 (arm_exynosm1_tune): Likewise.
5285 (arm_xgene1_tune): Likewise.
5286 (arm_cortex_a5_tune): Likewise.
5287 (arm_cortex_a9_tune): Likewise.
5288 (arm_cortex_a12_tune): Likewise.
5289 (arm_cortex_a73_tune): Likewise.
5290 (arm_v7m_tune): Likewise.
5291 (arm_cortex_m7_tune): Likewise.
5292 (arm_v6m_tune): Likewise.
5293 (arm_fa726te_tune): Likewise.
5294 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5296 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5298 * config/arm/arm.c (arm_mem_costs): New function.
5299 (arm_rtx_costs_internal): Use arm_mem_costs.
5301 2017-11-23 Mark Wielaard <mark@klomp.org>
5303 * dwarf2out.c (init_sections_and_labels): Use generation to create
5304 unique ranges_section_label and ranges_base_label. Return generation.
5305 (output_rnglists): Add generation argument. Use generation to create
5306 unique ranges labels.
5307 (dwarf2out_finish): Get generation from init_sections_and_labels
5308 and pass generation to output_rnglists.
5310 2017-11-23 Mike Stump <mikestump@comcast.net>
5311 Eric Botcazou <ebotcazou@adacore.com>
5313 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5314 * cfgloop.h (struct loop): Add unroll field.
5315 * function.h (struct function): Add has_unroll bitfield.
5316 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5317 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5318 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5319 (pass_rtl_unroll_loops::gate): Likewise.
5320 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5321 for which loop->unroll==1.
5322 (decide_unroll_constant_iterations): Use note for consistency and
5323 take loop->unroll into account. Return early if loop->unroll is set.
5324 Fix thinko in existing test.
5325 (decide_unroll_runtime_iterations): Use note for consistency and
5326 take loop->unroll into account.
5327 (decide_unroll_stupid): Likewise.
5328 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5329 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5330 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5332 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5333 (print_loop): Print loop->unroll if set.
5334 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5335 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5336 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5338 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5339 loop->unroll is set and smaller than the trip count. Otherwise bypass
5340 entirely the heuristics if loop->unroll is set. Remove dead note.
5341 Fix off-by-one bug in other note.
5342 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5343 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5345 (tree_unroll_loops_completely): Make static.
5346 (pass_complete_unroll::execute): Use correct type for variable.
5347 (pass_complete_unrolli::execute): Fix formatting.
5348 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5350 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5352 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5353 enable when TARGET_PREFER_AVX128 is set.
5355 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5357 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5359 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5360 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5362 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5364 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5365 frequencies. Fix estimation of aggregate parameters.
5367 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5369 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5372 2017-11-23 Tom de Vries <tom@codesourcery.com>
5374 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5376 2017-11-23 Tom de Vries <tom@codesourcery.com>
5378 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5381 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5384 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5385 sibcall_value_pcrel_fdpic): Use local variable instead of
5387 (calli_tbr_rel): Add missing operand 2.
5388 (call_valuei_tbr_rel): Add missing operand 3.
5390 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5393 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5394 bitpos/bitsize covering the whole destination, use store_expr only if
5395 the complex mode is the same. Otherwise, use expand_normal and if
5396 it returns CONCAT, subreg each part separately instead of trying to
5397 subreg the whole result.
5399 2017-11-23 Richard Biener <rguenther@suse.de>
5401 PR tree-optimization/23094
5402 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5403 come here from walking over backedges in the first iteration.
5404 (vn_reference_lookup_3): Skip clobbers that store the same value.
5406 2017-11-23 Richard Biener <rguenther@suse.de>
5408 PR tree-optimization/81403
5409 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5410 a block we need a leader relative to.
5411 (phi_translate_1): For nary processing require a leader from
5412 get_representative_for given we run expression simplification
5413 using match-and-simplify. Remove previous fix.
5415 2017-11-22 Jeff Law <law@redhat.com>
5417 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5418 Use new method allocate_value_range rather than accessing the
5419 vrp_value_range_pool data member directly.
5420 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5421 to use extract_range_from_stmt method to avoid need for
5422 extract_range_from_assignment method.
5423 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5424 method rather than setting values_propgated data member directly.
5425 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5426 and values propagated data members and extract_range_from_assignment
5427 method. Reorder private data members to conform to standards.
5428 Add new methods set_lattice_propagation_complete and
5429 allocate_value_range.
5431 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5433 PR rtl-optimization/83030
5434 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5435 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5436 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5437 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5438 CROSSING_JUMP_P flag is already set before setting it.
5439 * cfgrtl.c (fixup_partition_crossing): Likewise.
5440 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5443 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5445 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5446 Handle the case where both arguments are using gen_const_vec_series.
5448 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5451 * pretty-print.c (pp_format): Move quoting implementation to
5452 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5453 to pass address of "quote" local.
5454 (pp_begin_quote): New function.
5455 (pp_end_quote): New function.
5456 * pretty-print.h (printer_fn): Convert penultimate param from bool
5458 (pp_begin_quote): New decl.
5459 (pp_end_quote): New decl.
5460 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5461 param from bool to bool *.
5462 * tree-diagnostic.h (default_tree_printer): Likewise.
5464 2017-11-22 Jeff Law <law@redhat.com>
5466 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5467 Initialize vr_values.
5468 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5469 vr_values via vr_values class instance rather than delegators.
5470 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5471 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5472 (evrp_range_analyzer::push_value_range): Likewise.
5473 (evrp_range_analyzer::pop_value_range): Likewise.
5474 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5475 most delegators. Those remaining are exposed as public interfaces.
5476 Make vr_values a pointer and private.
5477 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5479 (evrp_range_analyzer::get_vr_value): New method.
5480 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5481 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5482 (class evrp_dom_walker): Attach evrp_folder class, initialize
5483 it in the ctor. Remove temporary delegators.
5484 (evrp_dom_walker::before_dom_children): Call methods in attached
5485 evrp_range_analyzer class via class instance pointer. Use
5486 free value_range_constant_singleton to remove need for
5487 op_with_constant_singleton_value delegator method. Do not
5488 create a vrp_prop class instance for every call! Narrow
5489 scope of a couple variables.
5490 (evrp_dom_walker::cleanup): Call methods in attached
5491 evrp_range_analyzer class via class instance pointer.
5492 * vr-values.h (class vr_values): Privatize many methods and
5495 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5497 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5498 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5499 * expr.c (expand_expr_real_2): Likewise.
5500 * fold-const.c (const_unop): Likewise.
5501 * optabs-tree.c (optab_for_tree_code): Likewise.
5502 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5503 * tree-inline.c (estimate_operator_cost): Likewise.
5504 * tree-pretty-print.c (dump_generic_node): Likewise.
5505 (op_code_prio): Likewise.
5506 (op_symbol_code): Likewise.
5507 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5508 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5509 * internal-fn.c (direct_internal_fn_optab): New function.
5510 (direct_internal_fn_array, direct_internal_fn_supported_p
5511 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5512 * fold-const-call.c (fold_const_reduction): New function.
5513 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5515 * tree-vect-loop.c: Include internal-fn.h.
5516 (reduction_code_for_scalar_code): Rename to...
5517 (reduction_fn_for_scalar_code): ...this and return an internal
5519 (vect_model_reduction_cost): Take an internal_fn rather than
5521 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5523 (vectorizable_reduction): Use internal functions rather than tree
5524 codes for the reduction operation. Update calls to the functions
5526 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5527 Use calls to internal functions rather than REDUC tree codes.
5528 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5530 2017-11-22 Olivier Hainque <hainque@adacore.com>
5532 * config/vxworks.c (vxworks_override_options): Pick default
5533 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5534 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5535 DWARF_GNAT_ENCODINGS_DEFAULT.
5536 * config/vxworksae.h: Likewise.
5538 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5540 PR tree-optimization/83104
5541 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5544 2017-11-22 Marek Polacek <polacek@redhat.com>
5545 H.J. Lu <hongjiu.lu@intel.com>
5546 Jason Merrill <jason@redhat.com>
5551 * calls.c (initialize_argument_information): Call
5552 warn_parameter_passing_abi target hook.
5553 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5554 argument onto stack.
5555 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5556 * common.opt: Update -fabi-version description.
5557 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5558 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5560 (ix86_is_empty_record): New function.
5561 (ix86_warn_parameter_passing_abi): New function.
5562 (TARGET_EMPTY_RECORD_P): Redefine.
5563 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5564 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5565 * doc/tm.texi: Regenerated.
5566 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5567 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5568 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5569 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5570 instead of int_size_in_bytes.
5571 * expr.c (copy_blkmode_to_reg): Likewise.
5572 * function.c (aggregate_value_p): Return 0 for empty types.
5573 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5575 (locate_and_pad_parm): Call arg size_in_bytes instead
5577 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5578 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5579 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5581 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5582 (std_gimplify_va_arg_expr): Skip empty records. Call
5583 arg_size_in_bytes instead size_in_bytes.
5584 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5585 * tree-core.h (tree_type_common): Add empty_flag.
5586 (tree_decl_common): Update comments.
5587 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5589 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5590 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5592 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5593 * tree.c (default_is_empty_type): New function.
5594 (default_is_empty_record): New function.
5595 (arg_int_size_in_bytes): New function.
5596 (arg_size_in_bytes): New function.
5597 (get_ultimate_context): New function.
5598 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5599 TRANSLATION_UNIT_WARN_EMPTY_P.
5600 (default_is_empty_record, arg_int_size_in_bytes,
5601 arg_size_in_bytes, get_ultimate_context): Declare.
5603 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5605 * config/arm/arm.c (cmse_clear_registers): New function.
5606 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5607 code by call to cmse_clear_registers.
5608 (cmse_nonsecure_entry_clear_before_return): Likewise.
5610 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5612 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5613 (vdot_s32, vdotq_s32): New.
5614 (vdot_lane_u32, vdotq_lane_u32): New.
5615 (vdot_lane_s32, vdotq_lane_s32): New.
5618 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5621 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5622 for unsigned values with fewer HWIs than the precision.
5623 (test_overflow): New function.
5624 (wide_int_cc_tests): Call it.
5626 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5628 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5629 has the same mode class as Pmode.
5631 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5633 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5634 padding_bits_to_clear_ptr.
5635 (cmse_nonsecure_entry_clear_before_return): Likewise.
5637 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5639 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5640 auto_sbitap instead of integer bitfield to control register needing
5643 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5645 PR tree-optimization/83044
5646 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5647 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5648 that need the upper bound. Subtract offset from
5649 get_addr_base_and_unit_offset only if positive and subtract it
5650 before division by eltsize rather than after it.
5653 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5654 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5655 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5659 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5660 expand_binop with *mul_widen_optab, make sure at least one of the
5661 operands doesn't have VOIDmode.
5664 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5666 PR rtl-optimization/82044
5667 PR tree-optimization/82042
5668 * dse.c (record_store): Check for overflow.
5669 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5670 add_wild_read instead of clear_rhs_from_active_local_stores on
5671 overflow and log it into dump_file.
5673 2017-11-22 Richard Biener <rguenther@suse.de>
5675 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5676 fake edges to exit when looking for a place to insert.
5677 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5679 (insert_into_preds_of_block): Commit edge insertion immediately,
5680 assert that doesn't require new BBs.
5681 (fini_pre): Release expressions.
5682 (pass_pre::execute): Shuffle things around a bit, if the fn
5683 is too large do not compute AVAIL either as this is really the
5686 2017-11-22 Richard Biener <rguenther@suse.de>
5688 PR tree-optimization/83089
5689 * tree-if-conv.c (pass_if_conversion::execute): If anything
5690 changed reset SCEV and free the number of iteration estimates.
5692 2017-11-21 Martin Sebor <msebor@redhat.com>
5694 PR tree-optimization/82945
5695 * calls.h (warn_nonstring_bound): Remove unused function.
5697 2017-11-21 Martin Sebor <msebor@redhat.com>
5699 PR tree-optimization/82945
5700 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5701 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5702 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5704 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5705 * calls.h (get_attr_nonstring_decl): Declare new function.
5706 * doc/extend.texi (attribute nonstring): Update.
5707 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5708 get_attr_nonstring_decl and handle it.
5709 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5710 detection of nul-termination.
5711 (strlen_to_stridx): Change to a pointer.
5712 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5713 (pass_strlen::execute): Same.
5715 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5717 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5718 for the new option -mprefer-vector-width=[none|128|256|512].
5719 * config/i386/i386.c (ix86_target_string): remove old style options
5720 -mprefer-avx256 and make -mprefer-avx128 as alias.
5721 (ix86_option_override_internal): Apply defaults for the
5722 -mprefer-vector-width=[128|256] option.
5723 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5724 Implement macros to work with -mprefer-vector-width=.
5725 * config/i386/i386.opt: Implemented option
5726 -mprefer-vector-width=[none|128|256|512].
5727 * doc/invoke.texi: Documentation for
5728 -mprefer-vector-width=[none|128|256|512].
5730 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5732 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5734 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5736 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5738 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5740 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5742 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5743 * expr.c (expand_expr_real_2): Likewise.
5744 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5745 fold_binary_loc): Likewise.
5746 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5747 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5748 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5749 MINUS_EXPR transformations.
5750 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5751 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5752 * tree-inline.c (estimate_operator_cost): Likewise.
5753 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5754 op_symbol_code): Likewise.
5755 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5756 * vr-values.c (extract_range_from_binary_expr): Likewise.
5757 * varasm.c (initializer_constant_valid_p_1): Likewise.
5758 * tree.def: New tree code POINTER_DIFF_EXPR.
5760 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5762 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5763 integer suffix to movbe mnemonic.
5764 (*bswaphi2_movbe): Ditto.
5765 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5767 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5770 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5771 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5772 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5774 2017-11-21 Martin Liska <mliska@suse.cz>
5776 * tree-inline.c (expand_call_inline): Remove not needed
5779 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5780 Jakub Jelinek <jakub@redhat.com>
5783 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5784 Remove static keyword from f variable.
5786 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5788 PR tree-optimization/83086
5789 * gimple-ssa-store-merging.c
5790 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5791 rather than n.base_addr.
5793 2017-11-21 Martin Liska <mliska@suse.cz>
5795 PR rtl-optimization/82044
5796 PR tree-optimization/82042
5797 * dse.c (check_mem_read_rtx): Check for overflow.
5799 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5801 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5804 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5806 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5807 and its corresponding call arg location note.
5809 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5810 Andrew Burgess <andrew.burgess@embecosm.com>
5812 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5814 (arc_return_slot_offset): Likewise.
5815 (arc_eh_return_address_location): New declaration.
5816 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5817 (MUST_SAVE_REGISTER): Add exception handler case.
5818 (MUST_SAVE_RETURN_ADDR): Likewise.
5819 (arc_frame_pointer_required): Likewise.
5820 (arc_frame_pointer_needed): New function.
5821 (arc_compute_frame_size): Changed.
5822 (arc_expand_prologue): Likewise.
5823 (arc_expand_epilogue): Likewise.
5824 (arc_initial_elimination_offset): Likewise.
5825 (arc_return_slot_offset): Delete.
5826 (arc_eh_return_address_location): New function.
5827 (arc_builtin_setjmp_frame_value): Likewise.
5828 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5829 (EH_RETURN_STACKADJ_RTX): Define.
5830 (EH_RETURN_HANDLER_RTX): Likewise.
5831 * config/arc/arc.md (eh_return): Delete.
5833 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5835 * print-tree.h (debug_vec_tree): Remove prototype.
5836 * gdbinit.in (pvt): Remove macro.
5838 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5840 PR tree-optimization/83047
5841 * gimple-ssa-store-merging.c
5842 (imm_store_chain_info::output_merged_store): If the loads with the
5843 same vuse are in different basic blocks, for load_gsi pick a load
5844 location that is dominated by the other loads.
5847 * config/i386/i386.c (ix86_memmodel_check): Start
5848 -Winvalid-memory-model diagnostics with lowercase letter.
5851 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5852 set in_cold_section_p to the partition containing loc_list->first.
5853 When seeing loc_list->last_before_switch node, update secname and
5854 perform range_across_switch second partition handling only after that.
5857 * run-rtl-passes.c: Include debug.h.
5858 (run_rtl_passes): Call debug_hooks->assembly_start.
5859 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5863 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5864 OPTAB_DIRECT in calls to expand_simple_binop.
5866 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5869 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5871 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5874 * spellcheck.h (best_match::blithely_get_best_candidate): New
5877 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5879 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5880 with lower case letter.
5882 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5884 * config/i386/i386.md (bswaphi2): New expander.
5885 (*bswaphi2_movbe): New insn pattern.
5886 (bswaphi -> rorhi pepehole2): New peephole pattern.
5888 2017-11-20 Jeff Law <law@redhat.com>
5890 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5891 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5892 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5893 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5894 gimple-ssa-evrp-analyze.h.
5896 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5897 set BB_VISITED here.
5898 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5900 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5902 * tree-predcom.c: Add general comment on Store-Store chains.
5903 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5905 (get_chain_last_ref_at): Rename into...
5906 (get_chain_last_write_at): ...this.
5907 (get_chain_last_write_before_load): New function.
5908 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5909 CT_STORE_STORE when write reference is added.
5910 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5911 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5912 (initialize_root_vars_store_elim_1): Ditto.
5913 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5914 definition is created.
5915 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5916 chain by replacing it with dominant stored value.
5918 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5920 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5922 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5924 * vr-values.c (extract_range_from_binary_expr): Use a full range
5927 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5929 * config/arm/arm.md (R4_REGNUM): Define constant.
5930 (nonsecure_call_internal): Remove r4 clobber.
5931 (nonsecure_call_value_internal): Likewise.
5932 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5933 clobber and resequence match_operands.
5934 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5935 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5936 (nonsecure_call_value_reg_thumb2): Likewise.
5938 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5940 PR tree-optimization/78821
5941 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5942 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5943 to the address of the base rather than the base itself.
5944 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5945 (find_bswap_or_nop_finalize): New function.
5946 (find_bswap_or_nop): Use it.
5947 (bswap_replace): Return a tree rather than bool, change first
5948 argument from gimple * to gimple_stmt_iterator, allow inserting
5949 into an empty sequence, allow ins_stmt to be NULL - then emit
5950 all stmts into gsi. Fix up MEM_REF address gimplification.
5951 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5952 (struct store_immediate_info): Add N and INS_STMT non-static
5954 (store_immediate_info::store_immediate_info): Initialize them
5955 from newly added ctor args.
5956 (merged_store_group::apply_stores): Formatting fixes. Sort by
5958 (stmts_may_clobber_ref_p): For stores call also
5959 refs_anti_dependent_p.
5960 (gather_bswap_load_refs): New function.
5961 (imm_store_chain_info::try_coalesce_bswap): New method.
5962 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5963 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5964 (imm_store_chain_info::output_merged_store): Fail if number of
5965 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5966 and NOP_EXPR rhs_code.
5967 (pass_store_merging::process_store): Compute n and ins_stmt, if
5968 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5969 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5971 (pass_store_merging::execute): Calculate dominators.
5973 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5974 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5975 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5976 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5977 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5978 class pass_optimize_bswap, bswap_replace,
5979 pass_optimize_bswap::execute): Moved to ...
5980 * gimple-ssa-store-merging.c: ... this file.
5981 Include optabs-tree.h.
5982 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5983 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5984 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5985 anonymous namespace, remove static keywords.
5986 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5987 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5989 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5992 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5994 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5996 * vec.h (debug_helper): New function.
5997 (DEFINE_DEBUG_VEC): New macro.
5998 * hash-set.h (debug_helper): New function.
5999 (DEFINE_DEBUG_HASH_SET): New macro.
6000 * cfg.c (debug_slim (edge)): New function.
6001 Call DEFINE_DEBUG_VEC for edges.
6002 Call DEFINE_DEBUG_HASH_SET for edges.
6003 * cfghooks.c (debug_slim (basic_block)): New function.
6004 Call DEFINE_DEBUG_VEC for basic blocks.
6005 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6006 * print-tree.c (debug_slim): New function to handle trees.
6007 Call DEFINE_DEBUG_VEC for trees.
6008 Call DEFINE_DEBUG_HASH_SET for trees.
6009 (debug (vec<tree, va_gc>) &): Remove.
6010 (debug (<vec<tree, va_gc>) *): Remove.
6011 * print-rtl.c (debug_slim): New function to handle const_rtx.
6012 Call DEFINE_DEBUG_VEC for rtx_def.
6013 Call DEFINE_DEBUG_VEC for rtx_insn.
6014 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6015 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6016 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6017 (debug (vec<rtx_insn *> *ptr): Remove.
6018 (debug_insn_vector): Remove.
6019 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6021 2017-11-20 Tom de Vries <tom@codesourcery.com>
6023 PR rtl-optimization/82020
6024 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6025 IF_THEN_ELSE condition.
6027 2017-11-19 Jeff Law <law@redhat.com>
6029 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6030 of degenerates resulting from ignoring an edge.
6032 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6035 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6037 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6040 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6041 for uninitialied counts.
6043 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6046 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6048 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6051 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6053 2017-11-19 Tom de Vries <tom@codesourcery.com>
6055 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6058 2017-11-19 Tom de Vries <tom@codesourcery.com>
6060 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6063 2017-11-19 Tom de Vries <tom@codesourcery.com>
6065 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6068 2017-11-19 Tom de Vries <tom@codesourcery.com>
6070 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6072 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6074 2017-11-19 Tom de Vries <tom@codesourcery.com>
6076 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6078 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6079 semicolon after MCORE_EXPORT_NAME call.
6081 2017-11-19 Tom de Vries <tom@codesourcery.com>
6084 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6085 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6087 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6089 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6090 (cgraph_edge::clone): Cleanup updating of profile.
6091 * ipa-cp.c (update_profiling_info): Likewise.
6092 * ipa-inline-transform.c (inline_transform): Likewise.
6093 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6094 * ipa-split.c (execute_split_functions): Do not split when function
6096 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6097 * profile-count.c (profile_count::dump): Add global0.
6098 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6100 (profile_count::to_sreal_scale): Likewise.
6101 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6102 (profile_count::combine_with_ipa_count): New function.
6103 * profile-count.h (profile_guessed_global0adjusted): New.
6104 (profile_count::adjusted_zero): New.
6105 (profile_count::global0adjusted): New.
6106 (profile_count::combine_with_ipa_count): New.
6107 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6108 correct profile of return block of split functions.
6109 (copy_cfg_body): Remove unused profile_count.
6110 (copy_body): Likewise.
6111 (expand_call_inline): Update.
6112 (tree_function_versioning): Update.
6114 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6116 * hash-set.h (hash_set::empty): New.
6117 * tree-ssa-threadbackward.h: Delete.
6118 * tree-ssa-threadbackward.c (class thread_jumps): New.
6119 Move max_threaded_paths into class.
6120 (fsm_find_thread_path): Remove arguments that are now in class.
6121 (profitable_jump_thread_path): Rename to...
6122 (thread_jumps::profitable_jump_thread_path): ...this.
6123 (convert_and_register_jump_thread_path): Rename to...
6124 (thread_jumps::convert_and_register_current_path): ...this.
6125 (check_subpath_and_update_thread_path): Rename to...
6126 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6127 (register_jump_thread_path_if_profitable): Rename to...
6128 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6129 (handle_phi): Rename to...
6130 (thread_jumps::handle_phi): ...this.
6131 (handle_assignment): Rename to...
6132 (thread_jumps::handle_assignment): ...this.
6133 (fsm_find_control_statement_thread_paths): Rename to...
6134 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6135 (find_jump_threads_backwards): Rename to...
6136 (thread_jumps::find_jump_threads_backwards): ...this.
6137 Initialize path local data.
6138 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6139 from within thread_jumps class.
6140 (pass_early_thread_jumps::execute): Same.
6142 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6144 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6146 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6148 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6149 to canonical location.
6151 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6154 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6156 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6158 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6160 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6161 rcrt1.o%s/grcrt1.o%s for -static-pie.
6163 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6165 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6166 ix86_shift_rotate_cost): Break out from ...
6167 (ix86_rtx_costs): ... here.
6168 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6171 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6173 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6174 when entry block was promoted unlikely.
6175 (estimate_bb_frequencies): Increase frequency scale.
6176 * profile-count.h (profile_count): Export precision info.
6178 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6180 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6181 disturbin profile of entry block.
6183 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6185 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6188 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6190 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6191 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6193 2017-11-17 Jeff Law <law@redhat.com>
6195 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6196 from evrp_dom_walker class. Various methods moved into new class.
6197 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6198 (evrp_range_analyzer::enter): New method.
6199 (evrp_range_analyzer::leave): New method.
6200 (evrp_dom_walker): Remove delegators no longer needed by this class.
6201 Replace vr_values data member with evrp_range_analyzer
6203 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6204 method extracted from evrp_dom_walker::before_dom_children.
6205 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6206 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6208 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6209 Add private copy constructor and move assignment operators.
6210 Privatize methods and class data where trivially possible.
6211 (evrp_dom_walker::cleanup): New function, extracted from
6212 execute_early_vrp. Simplify access to class data.
6214 * vr-values.h (get_output_for_vrp): Prototype.
6215 * vr-values.c (get_output_for_vrp): New function extracted from
6216 vrp_visit_assignment_or_call and extract_range_from_stmt.
6217 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6219 2017-11-17 Luis Machado <luis.machado@linaro.org>
6221 * config/aarch64/aarch64.c
6222 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6223 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6224 tune_params::AUTOPREFETCHER_WEAK.
6226 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6229 * config/arm/arm.c (arm_valid_target_attribute_rec):
6230 Parse "arch=" and "+<ext>".
6231 (arm_valid_target_attribute_tree): Re-init global options.
6232 (arm_option_override): Make non-static.
6233 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6234 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6235 (__ARM_FEATURE_CRC32): Support undef.
6236 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6237 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6239 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6241 * gdbinit.in (break-on-diagnostic): New command.
6243 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6245 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6246 return to indirect jump for EH return if control-flow
6247 protection is enabled. Change explicit 'false' argument in
6248 pro_epilogue_adjust_stack with a value of
6250 * config/i386/i386.md (simple_return_indirect_internal):
6251 Remove SImode restriction to support 64-bit.
6253 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6255 * combine.c (added_notes_insn): New.
6256 (try_combine): Handle added_notes_insn like added_links_insn.
6257 Rewrite return value code.
6258 (distribute_notes): Set added_notes_insn to the earliest insn we added
6261 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6263 PR rtl-optimization/82621
6264 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6265 dest of one of those SETs is unused.
6267 2017-11-17 Richard Biener <rguenther@suse.de>
6270 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6271 * tree-pretty-print.c (dump_generic_node): Handle
6272 annot_expr_parallel_kind.
6273 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6274 * gimplify.c (gimple_boolify): Likewise.
6276 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6278 * config.gcc (extra_headers): Add cet.h for x86 targets.
6279 * config/i386/cet.h: New file.
6280 * doc/install.texi: Add --enable-cet/--disable-cet.
6282 2017-11-17 Richard Biener <rguenther@suse.de>
6284 PR tree-optimization/83017
6285 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6286 (gen_parallel_loop): Properly count iterations.
6287 (parallelize_loops): Handle loop->can_be_parallel independent
6288 of flag_loop_parallelize_all. Make static profitability test match
6290 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6291 * invoke.texi (parloops-min-per-thread): Document.
6293 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6295 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6296 upstreaming review comments.
6298 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6300 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6301 with fast unaligned access.
6302 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6304 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6306 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6307 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6309 2017-11-17 Richard Biener <rguenther@suse.de>
6311 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6312 folding of references.
6314 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6317 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6318 of replacing call to strncmp with corresponding call to strcmp when
6321 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6323 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6324 option prefer-avx256 for skylake-avx512 configuration.
6325 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6326 (get_builtin_code_for_version): Ditto.
6328 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6329 Monk Chiang <sh.chiang04@gmail.com>
6331 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6332 (FIXED_REGISTERS): Reserve more register numbers.
6333 (CALL_USED_REGISTERS): Likewise.
6334 (REG_ALLOC_ORDER): Likewise.
6335 (REG_CLASS_CONTENTS): Likewise.
6336 (REGISTER_NAMES): Likewise.
6338 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6339 Kito Cheng <kito.cheng@gmail.com>
6341 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6343 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6345 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6347 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6348 parameter list for vec_splats.
6350 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6352 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6354 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6356 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6359 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6360 with the required version of automake 1.11.6.
6362 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6364 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6366 (DU_C2_3_power9): Correct reservation combinations.
6367 (FP_DIV_power9, VEC_DIV_power9): New.
6368 (power9-alu): Split out rotate/shift...
6369 (power9-rot): ...to here, correct dispatch resource.
6370 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6372 (power9-fp): Correct latency.
6373 (power9-sdiv): Add div/sqrt resource.
6374 (power9-ddiv): Correct latency, add div/sqrt resource.
6375 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6376 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6378 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6380 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6382 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6383 switch statement mapping KF built-ins to TF built-ins if we don't
6384 have the proper ISA 3.0 assembler support.
6386 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6388 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6389 (lower_emutls_function_body): Do not compute it.
6391 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6393 * ipa-split.c (split_bb_info): Turn time to sreal.
6394 (split_point): Likewise.
6395 (dump_split_point): Likewise.
6396 (fine_split_points): Likewise.
6397 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6399 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6401 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6404 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6406 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6408 (expensive_function_p): Remove useless assert.
6409 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6411 2017-11-16 Martin Sebor <msebor@redhat.com>
6413 PR tree-optimization/82588
6414 PR tree-optimization/82583
6415 * tree-vrp.c (check_array_ref): Handle flexible array members,
6416 string literals, and inner indices.
6417 (search_for_addr_array): Add detail to diagnostics.
6419 2017-11-16 Nathan Sidwell <nathan@acm.org>
6423 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6424 (SET_DECL_ASSEMBLER_NAME): Forward to
6425 overwrite_decl_assembler_name.
6426 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6427 (overwrite_decl_assembler_name): Declare.
6428 * tree.c (overwrite_decl_assembler_name): New.
6429 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6430 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6431 (LANG_HOOKS_INITIALIZER): Add it.
6432 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6433 * langhooks.c (lhd_set_decl_assembler_name): Use
6434 SET_DECL_ASSEMBLER_NAME.
6435 (lhd_overwrite_decl_assembler_name): Default implementation.
6437 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6438 Jackson Woodruff <jackson.woodruff@arm.com>
6440 PR tree-optimization/71026
6441 * match.pd: Canonicalize constant multiplies in division.
6443 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6445 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6446 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6447 Remove OPT_fomit_frame_pointer entry.
6448 * common/config/arc/arc-common.c: Likewise.
6449 * common/config/arm/arm-common.c: Likewise.
6450 * common/config/avr/avr-common.c: Likewise.
6451 * common/config/c6x/c6x-common.c: Likewise.
6452 * common/config/cr16/cr16-common.c: Likewise.
6453 * common/config/cris/cris-common.c: Likewise.
6454 * common/config/epiphany/epiphany-common.c: Likewise.
6455 * common/config/fr30/fr30-common.c: Likewise.
6456 * common/config/frv/frv-common.c: Likewise.
6457 * common/config/ia64/ia64-common.c: Likewise.
6458 * common/config/iq2000/iq2000-common.c: Likewise.
6459 * common/config/lm32/lm32-common.c: Likewise.
6460 * common/config/m32r/m32r-common.c: Likewise.
6461 * common/config/mcore/mcore-common.c: Likewise.
6462 * common/config/microblaze/microblaze-common.c: Likewise.
6463 * common/config/mips/mips-common.c: Likewise.
6464 * common/config/mmix/mmix-common.c: Likewise.
6465 * common/config/mn10300/mn10300-common.c: Likewise.
6466 * common/config/nios2/nios2-common.c: Likewise.
6467 * common/config/pa/pa-common.c: Likewise.
6468 * common/config/pdp11/pdp11-common.c: Likewise.
6469 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6470 * common/config/riscv/riscv-common.c: Likewise.
6471 * common/config/rs6000/rs6000-common.c: Likewise.
6472 * common/config/rx/rx-common.c: Likewise.
6473 * common/config/s390/s390-common.c: Likewise.
6474 * common/config/sh/sh-common.c: Likewise.
6475 * common/config/sparc/sparc-common.c: Likewise.
6476 * common/config/tilegx/tilegx-common.c: Likewise.
6477 * common/config/tilepro/tilepro-common.c: Likewise.
6478 * common/config/v850/v850-common.c: Likewise.
6479 * common/config/visium/visium-common.c: Likewise.
6480 * common/config/xstormy16/xstormy16-common.c: Likewise.
6481 * common/config/xtensa/xtensa-common.c: Likewise.
6482 * invoke.texi (-fomit-frame-pointer): Update documentation.
6484 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6486 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6488 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6490 * cfg.c (scale_bbs_frequencies_int,
6491 cale_bbs_frequencies_gcov_type): Remove.
6492 * cfg.h (scale_bbs_frequencies_int,
6493 cale_bbs_frequencies_gcov_type): Remove.
6495 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6497 * tree-ssa-loop-manip.c
6498 (scale_dominated_blocks_in_loop): Update to profile counts.
6499 (tree_transform_and_unroll_loop): Likewise.
6501 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6503 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6504 scale_bbs_frequencies_int.
6506 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6508 * final.c (compute_alignments): Use counts rather than frequencies.
6510 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6512 * cfgloopanal.c: Include sreal.h
6513 (average_num_loop_insns): Use counts and sreal for accounting.
6515 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6517 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6520 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6521 Kito Cheng <kito.cheng@gmail.com>
6523 * config/nds32/constraints.md: Provide more constraints.
6524 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6525 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6526 support constraints usage.
6528 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6530 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6532 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6533 Kito Cheng <kito.cheng@gmail.com>
6535 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6536 * config/nds32/nds32.opt: Refine the layout.
6537 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6538 TARGET_EXT_STRING): Support new options.
6539 * config/nds32/nds32.h: Likewise.
6540 * config/nds32/nds32.md: Likewise.
6541 * config/nds32/nds32-predicates.c: Likewise.
6542 * config/nds32/constraints.md: Likewise.
6543 * common/config/nds32/nds32-common.c: Likewise.
6545 2017-11-16 Julia Koval <julia.koval@intel.com>
6548 * config/i386/gfniintrin.h: Add sse check.
6549 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6551 2017-11-16 Julia Koval <julia.koval@intel.com>
6553 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6554 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6555 (ix86_handle_option): Handle -mavx512vbmi2.
6556 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6557 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6558 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6559 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6560 (ix86_valid_target_attribute_inner_p): Ditto.
6561 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6562 * config/i386/i386.opt (mavx512vbmi2): New option.
6563 * doc/invoke.texi: Add new option.
6565 2017-11-16 Julia Koval <julia.koval@intel.com>
6567 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6568 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6569 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6570 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6571 _mm512_gf2p8mul_epi8): New intrinsics.
6572 * config/i386/i386-builtin-types.def
6573 (V64QI_FTYPE_V64QI_V64QI): New type.
6574 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6575 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6576 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6577 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6578 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6579 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6581 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6583 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6584 explicitly as a stream of bytes.
6586 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6588 * config/rs6000/altivec.h (vec_xst_be): New #define.
6589 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6590 and externalize from *altivec_vperm_<mode>_internal.
6591 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6593 (XL_BE_V8HI): Likewise.
6594 (XL_BE_V4SI): Likewise.
6595 (XL_BE_V4SI): Likewise.
6596 (XL_BE_V2DI): Likewise.
6597 (XL_BE_V4SF): Likewise.
6598 (XL_BE_V2DF): Likewise.
6599 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6600 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6601 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6602 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6603 VSX_BUILTIN_VEC_XST_BE.
6604 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6605 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6607 (altivec_init_builtins): Replace conditional calls to def_builtin
6608 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6609 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6610 with unconditional calls. Remove calls to def_builtin for
6611 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6612 __builtin_vec_xst_be.
6613 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6614 to define_expand, and add alternate RTL generation for P8.
6615 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6616 vsx_ld_elemrev_v8hi.
6617 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6618 add alternate RTL generation for P8.
6619 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6620 vsx_ld_elemrev_v16qi.
6621 (vsx_st_elemrev_v8hi): Convert define_insn
6622 to define_expand, and add alternate RTL generation for P8.
6623 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6624 vsx_st_elemrev_v8hi.
6625 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6626 add alternate RTL generation for P8.
6627 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6628 vsx_st_elemrev_v16qi.
6630 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6633 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6634 TARGET_AVX512ER check.
6635 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6636 neither -mzeroupper nor -mno-zeroupper is used and
6637 TARGET_EMIT_VZEROUPPER is set.
6638 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6639 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6641 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6643 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6644 folding of vector compares.
6645 (fold_build_vec_cmp): New helper function.
6646 (fold_compare_helper): New helper function.
6647 (builtin_function_type): Add compare builtins to the list of functions
6648 having unsigned arguments. Cosmetic updates to comment indentation.
6649 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6650 the not+eq combination.
6652 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6654 PR tree-optimization/82726
6655 PR tree-optimization/70754
6656 * tree-predcom.c (order_drefs_by_pos): New function.
6657 (combine_chains): Move code setting has_max_use_after to...
6658 (try_combine_chains): ...here. New parameter. Sort combined chains
6659 according to position information.
6660 (tree_predictive_commoning_loop): Update call to above function.
6661 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6663 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6665 PR tree-optimization/82726
6667 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6669 PR tree-optimization/70754
6670 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6671 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6672 combined stmt before it if not NULL.
6673 (combine_chains): Process refs reversely and compute dominance point
6677 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6679 PR tree-optimization/79663
6680 * tree-predcom.c (combine_chains): Process refs in reverse order
6681 only for ZERO length chains, and add explaining comment.
6683 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6685 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6686 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6687 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6689 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6691 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6693 2017-11-15 Martin Liska <mliska@suse.cz>
6695 * tree-cfg.c (pass_warn_function_return::execute):
6696 Compare warn_return_type for greater than zero.
6698 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6702 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6703 to return true on Xeon and not on Xeon Phi.
6704 (ix86_check_avx256_register): Changed to ...
6705 (ix86_check_avx_upper_register): ... this. Add extra check for
6706 VALID_AVX512F_REG_OR_XI_MODE.
6707 (ix86_avx_u128_mode_needed): Changed
6708 ix86_check_avx256_register to ix86_check_avx_upper_register.
6709 (ix86_check_avx256_stores): Changed to ...
6710 (ix86_check_avx_upper_stores): ... this. Changed
6711 ix86_check_avx256_register to ix86_check_avx_upper_register.
6712 (ix86_avx_u128_mode_after): Changed
6713 avx_reg256_found to avx_upper_reg_found. Changed
6714 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6715 (ix86_avx_u128_mode_entry): Changed
6716 ix86_check_avx256_register to ix86_check_avx_upper_register.
6717 (ix86_avx_u128_mode_exit): Ditto.
6718 * config/i386/i386.h: (host_detect_local_cpu): New define.
6720 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6722 * config/arm/xgene1.md (xgene1): Split into automatons
6723 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6724 (xgene1_f_load): Adjust reservations and/or types.
6725 (xgene1_f_store): Likewise.
6726 (xgene1_load_pair): Likewise.
6727 (xgene1_store_pair): Likewise.
6728 (xgene1_fp_load1): Likewise.
6729 (xgene1_load1): Likewise.
6730 (xgene1_store1): Likewise.
6731 (xgene1_move): Likewise.
6732 (xgene1_alu): Likewise.
6733 (xgene1_simd): Likewise.
6734 (xgene1_bfm): Likewise.
6735 (xgene1_neon_load1): Likewise.
6736 (xgene1_neon_store1): Likewise.
6737 (xgene1_neon_logic): Likewise.
6738 (xgene1_neon_st1): Likewise.
6739 (xgene1_neon_ld1r): Likewise.
6740 (xgene1_alu_cond): Added.
6741 (xgene1_shift_reg): Likwise.
6742 (xgene1_bfx): Likewise.
6743 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6745 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6748 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6750 (can_widen_mult_without_libcall): New function.
6751 (expand_mul_overflow): If only checking unsigned mul overflow,
6752 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6753 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6754 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6755 (expand_DIVMOD): Formatting fix.
6756 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6757 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6758 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6760 PR tree-optimization/82977
6761 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6762 constructed temporary to strlen_to_stridx.put.
6764 2017-11-15 Martin Liska <mliska@suse.cz>
6766 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6767 * configure: Regenerate.
6769 2017-11-15 Martin Liska <mliska@suse.cz>
6772 * config/sh/sh-mem.cc: Use proper probability for
6775 2017-11-14 Jeff Law <law@redhat.com>
6777 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6778 the red zone for stack_clash_protection_final_dynamic_probe targets
6779 when the total dynamic stack size is zero bytes.
6781 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6782 blocks is post order.
6784 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6786 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6787 * final.c (rest_of_clean_state): Set it for the
6788 -fcompare-debug dump.
6789 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6790 class when TDF_COMPARE_DEBUG is set.
6792 * dwarf2out.c (gen_producer_string): Discard
6795 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6798 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6799 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6800 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6801 (__TGMATH_REAL_2_3): Remove macros.
6802 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6803 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6805 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6807 (remquo): Define using __TGMATH_REAL_3.
6809 2017-11-14 Jeff Law <law@redhat.com>
6811 * vr-values.c: New file with contents extracted from tree-vrp.c.
6812 * Makefile.in (OBJS): Add vr-values.o
6813 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6814 (set_value_range, set_and_canonicalize_value_range): Likewise.
6815 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6816 (value_range_constant_singleton, symbolic_range_p): Likewise.
6817 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6818 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6819 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6820 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6821 (find_case_label_range, find_case_label_index): Likewise.
6822 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6823 (range_int_cst_singleton_p, value_inside_range): Likewise.
6824 (get_single_symbol): Likewise.
6825 (switch_update): Move structure definition here.
6826 (to_remove_edges, to_update_switch_stmts): Provide externs.
6827 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6828 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6829 (vrp_val_is_min, set_value_range): Likewise.
6830 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6831 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6832 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6833 (range_is_nonnull, range_int_cst_p): Likewwise.
6834 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6835 (get_single_symbol, operand_less_p): Likewise
6836 (compare_values_warnv, compare_values): Likewise.
6837 (value_inside_range, value_range_constant_singleton): Likewise.
6838 (zero_nonzero_bitgs_from_vr): Likewise.
6839 (extract_range_from_binary_expr_1): Likewise.
6840 (overflow_comparison_p): Likewise.
6841 (to_remove_edges, to_update_switch_stmts): Likewise.
6842 (find_case_label-index, find_case_label_range): Likewise.
6843 (switch_update, set_value_range_to_nonnegative): Remove.
6844 (set_value_range_to_truthvalue): Likewise.
6845 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6846 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6847 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6848 (find_case_label_ranges, test_for_singularity): Likewise.
6849 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6850 (x_vr_values): Move to its remaining use site.
6852 2017-11-10 Jeff Law <law@redhat.com>
6854 * vr-values.h (VR_INITIALIZER): Move #define here.
6855 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6856 * Makefile.in (OBJS): Add tree-evrp.o
6857 * tree-vrp.h (assert_info): Move structure definition here.
6858 (set_value_range_to_varying): Prototype.
6859 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6860 (infer_value_range, register_edge_assert_for): Likewise.
6861 (stmt_interesting_for_vrp): Likewise.
6862 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6863 (set_value_range_to_varying): No longer static.
6864 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6865 (infer_value_range, register_edge_assert_for): Likewise.
6867 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6869 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6870 generate the XXBRD instruction.
6872 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6873 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6874 as being compatible if -mabi=ieeelongdouble.
6875 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6876 to setup float128 built-ins with hardware support.
6877 (BU_FLOAT128_HW_2): Likewise.
6878 (BU_FLOAT128_HW_3): Likewise.
6879 (BU_FLOAT128_HW_VSX_1): Likewise.
6880 (BU_FLOAT128_HW_VSX_2): Likewise.
6881 (scalar_extract_expq): Change float128 built-in functions to
6882 accommodate having both KFmode and TFmode functions. Use the
6883 KFmode variant as the default.
6884 (scalar_extract_sigq): Likewise.
6885 (scalar_test_neg_qp): Likewise.
6886 (scalar_insert_exp_q): Likewise.
6887 (scalar_insert_exp_qp): Likewise.
6888 (scalar_test_data_class_qp): Likewise.
6889 (sqrtf128_round_to_odd): Delete processing the round to odd
6890 built-in functions as special built-in functions, and define them
6891 as float128 built-ins. Use the KFmode variant as the default.
6892 (truncf128_round_to_odd): Likewise.
6893 (addf128_round_to_odd): Likewise.
6894 (subf128_round_to_odd): Likewise.
6895 (mulf128_round_to_odd): Likewise.
6896 (divf128_round_to_odd): Likewise.
6897 (fmaf128_round_to_odd): Likewise.
6898 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6899 support for KFmode and TFmode xststdcqp calls.
6900 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6901 point, switch the built-in handlers for the get/set float128
6902 exponent, get float128 mantissa, float128 test built-ins, and the
6903 float128 round to odd built-in functions. Eliminate creating the
6904 float128 round to odd built-in functions as special built-ins.
6905 (rs6000_init_builtins): Eliminate special creation of the float128
6906 round to odd built-in functions.
6907 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6908 function insns to support both TFmode and KFmode variants.
6909 (xsxsigqp_<mode>): Likewise.
6910 (xsiexpqpf_<mode>): Likewise.
6911 (xsiexpqp_<mode>): Likewise.
6912 (xststdcqp_<mode>): Likewise.
6913 (xststdcnegqp_<mode>): Likewise.
6914 (xststdcqp_<mode>): Likewise.
6916 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6918 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6919 unused path_in_freq_ptr parameter.
6920 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6922 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6924 * ipa-inline.c (edge_badness): Dump sreal frequency.
6925 (compute_inlined_call_time): Match natural implementaiton ...
6926 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6927 forgotten division by CGRAPH_FREQ_BASE.
6929 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6931 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6932 Solaris 11. Update comment.
6933 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6935 * config/sol2.h (STARTFILE_SPEC): Likewise.
6936 * configure: Regenerate.
6938 2017-11-14 Carl Love <cel@us.ibm.com>
6940 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6941 le_ and be_ prefixes to swap* variables. Remove
6942 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6945 2017-11-14 Jason Merrill <jason@redhat.com>
6947 Support GTY((cache)) on hash_map.
6948 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6949 (ggc_cache_remove): Override it instead of ggc_mx.
6950 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6951 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6952 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6953 (simple_cache_map_traits): Override maybe_mx.
6954 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6955 (hash_map): Friend gt_cleare_cache.
6956 (gt_cleare_cache): New.
6957 * tree.h (tree_cache_traits): New hash_map traits class.
6958 (tree_cache_map): New typedef.
6960 2017-11-14 Richard Biener <rguenther@suse.de>
6962 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6963 paramter and handling.
6964 (cleanup_control_flow_bb): Likewise.
6965 (cleanup_control_flow_pre): New helper performing a DFS walk
6966 to call cleanup_control_flow_bb in PRE order.
6967 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6968 via cleanup_control_flow_pre.
6970 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6972 * config/aarch64/aarch64-simd.md
6973 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6974 (*aarch64_simd_bsl<mode>_alt): Likewise.
6975 (aarch64_simd_bsldi_internal): New.
6976 (aarch64_simd_bsldi_alt): Likewise.
6978 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6980 * tracer.c (better_p): Do not compare frequencies.
6981 * reg-stack.c (better_edge): Likewise.
6982 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6985 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6987 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6988 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6989 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6990 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6991 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6992 * predict.c (maybe_hot_frequency_p): Inline to ...
6993 (maybe_hot_count_p): ... here; rewrite to counts.
6994 (counts_to_freqs): Rename to ...
6995 (update_max_bb_count): ... this one.
6996 (expensive_function_p): Use counts.
6997 (estimate_bb_frequencies): Update.
6998 (rebuild_frequencies): Update.
6999 * predict.h (counts_to_freqs): Rename to ...
7000 (update_max_bb_count): ... this one.
7001 * profile.c (compute_branch_probabilities): Add debug info
7002 * tree-inline.c (expand_call_inline): Update debug info.
7003 (optimize_inline_calls): Use update_max_bb_count..
7004 (tree_function_versioning): Use update_max_bb_count..
7005 * value-prof.c (gimple_value_profile_transformations):
7006 Do not use update_max_bb_count.
7008 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7010 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7011 always use frequencies.
7013 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7015 * bb-reorder.c: Remove frequencies from comments.
7016 (better_edge_p): Use profile counts.
7017 (find_traces): Dump profile counts.
7018 (rotate_loop): Use profile counts.
7019 (find_traces_1_round): Likewise.
7020 (connect_better_edge_p): Use counts instead of probabilities for
7022 (copy_bb_p): Drop early check for non-0 frequency.
7023 (sanitize_hot_paths): Update comments.
7025 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7027 * ipa-split.c (struct split_point): Add count.
7028 (consider_split): Do not compute incoming frequency; compute incoming
7029 count and store it to split_point.
7030 (split_function): Set count of the call to split part correctly.
7032 2017-11-13 Carl Love <cel@us.ibm.com>
7034 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7036 2017-11-13 Tom Tromey <tom@tromey.com>
7038 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7040 2017-11-13 Carl Love <cel@us.ibm.com>
7042 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7043 Add support for builtins:
7044 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7045 vector {un,}signed {char,int,short},
7046 vector {un,}signed {char,int,short}) arguments.
7047 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7048 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7049 Add BU_P9V_AV_2 expansions for the builtins.
7050 * config/rs6000/altivec.h (vec_first_match_index,
7051 vec_first_mismatch_index, vec_first_match_or_eos_index,
7052 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7053 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7054 new extern declaration.
7055 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7056 * config/rs6000/vsx.md (first_match_index_<mode>,
7057 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7058 first_mismatch_or_eos_index_<mode>): Add define expand.
7059 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7060 * doc/extend.texi: Update the built-in documenation file for the new
7063 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7065 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7066 into the min/max operations for _Float<N> and _Float<N>X types.
7068 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7071 * dwarf2out.c (do_eh_frame): New static variable.
7072 (dwarf2out_begin_prologue): Set it.
7073 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7075 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7077 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7079 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7081 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7084 * debug.h (dwarf2out_do_eh_frame): Declare.
7085 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7086 (dwarf2out_do_frame): Use it.
7087 (dwarf2out_do_cfi_asm): Likewise.
7088 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7089 (dwarf2out_assembly_start): Likewise.
7090 (dwarf2out_begin_prologue): Fix comment.
7091 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7092 if the target needs either debug or unwind DWARF2 info.
7093 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7094 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7095 -ftrapping-math, -ftrapv and -fwrapv.
7097 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7099 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7100 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7101 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7102 (estimate_edge_size_and_time): Likewise.
7103 (ipa_merge_fn_summary_after_inlining): Likewise.
7104 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7105 (compute_uninlined_call_time): Use sreal_frequency.
7106 (compute_inlined_call_time): Likewise.
7107 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7108 * profile-count.c: Include sreal.h.
7109 (profile_count::to_sreal_scale): New.
7110 * profile-count.h: Forward declare sreal.
7111 (profile_count::to_sreal_scale): Declare.
7113 2017-11-13 Nathan Sidwell <nathan@acm.org>
7115 * diagnostic.c (maybe_line_and_column): New.
7116 (diagnostic_get_location_text): Use it.
7117 (diagnostic_report_current_module): Likewise.
7118 (test_diagnostic_get_location_text): Add tests.
7120 2017-11-13 Luis Machado <luis.machado@linaro.org>
7122 * doc/md.texi (Specifying processor pipeline description): Fix
7123 incorrect latency for the div instruction example.
7125 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7127 PR tree-optimization/78821
7128 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7129 that bit_not_p is the same.
7130 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7131 (split_group): Count precisely bit_not_p bits in each statement.
7132 (invert_op): New function.
7133 (imm_store_chain_info::output_merged_store): Use invert_op to
7134 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7135 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7137 2017-11-13 Martin Liska <mliska@suse.cz>
7139 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7140 (struct source_info): Likewise.
7141 (add_branch_counts): Likewise.
7142 (add_line_counts): Likewise.
7143 (function_summary): Likewise.
7144 (output_intermediate_line): Likewise.
7145 (generate_results): Likewise.
7147 2017-11-13 Martin Liska <mliska@suse.cz>
7149 * gcov.c (struct block_info): Remove typedef for block_t.
7150 (struct line_info): Likewise.
7151 (line_info::has_block): Likewise.
7152 (EXIT_BLOCK): Likewise.
7153 (unblock): Likewise.
7154 (circuit): Likewise.
7155 (get_cycles_count): Likewise.
7156 (process_file): Likewise.
7157 (read_graph_file): Likewise.
7158 (solve_flow_graph): Likewise.
7159 (find_exception_blocks): Likewise.
7160 (add_line_counts): Likewise.
7161 (accumulate_line_info): Likewise.
7162 (output_line_details): Likewise.
7164 2017-11-13 Martin Liska <mliska@suse.cz>
7166 * gcov.c (struct arc_info): Remove typedef for arc_t.
7167 (struct line_info): Likewise.
7168 (add_branch_counts): Likewise.
7169 (output_branch_count): Likewise.
7170 (function_info::~function_info): Likewise.
7171 (circuit): Likewise.
7172 (output_intermediate_line): Likewise.
7173 (read_graph_file): Likewise.
7174 (solve_flow_graph): Likewise.
7175 (find_exception_blocks): Likewise.
7176 (add_line_counts): Likewise.
7177 (accumulate_line_info): Likewise.
7178 (output_line_details): Likewise.
7179 (output_function_details): Likewise.
7181 2017-11-13 Martin Liska <mliska@suse.cz>
7183 * gcov.c (struct function_info): Remove typedef for function_t.
7184 (struct source_info): Likewise.
7185 (source_info::get_functions_at_location): Likewise.
7186 (solve_flow_graph): Likewise.
7187 (find_exception_blocks): Likewise.
7188 (add_line_counts): Likewise.
7189 (output_intermediate_file): Likewise.
7190 (process_file): Likewise.
7191 (generate_results): Likewise.
7192 (release_structures): Likewise.
7193 (read_graph_file): Likewise.
7194 (read_count_file): Likewise.
7195 (accumulate_line_counts): Likewise.
7196 (output_lines): Likewise.
7198 2017-11-13 Martin Liska <mliska@suse.cz>
7200 * gcov.c (function_info::function_info): Remove num_counts
7201 and add vector<gcov_type>.
7202 (function_info::~function_info): Use the vector.
7203 (process_file): Likewise.
7204 (read_graph_file): Likewise.
7205 (read_count_file): Likewise.
7206 (solve_flow_graph): Likewise.
7208 2017-11-13 Martin Liska <mliska@suse.cz>
7210 * gcov.c (function_info::is_artificial): New function.
7211 (process_file): Erase all artificial early.
7212 (generate_results): Skip as all artificial are already
7215 2017-11-13 Martin Liska <mliska@suse.cz>
7217 * gcov.c (read_graph_file): Store to global vector of functions.
7218 (read_count_file): Iterate the vector.
7219 (process_file): Likewise.
7220 (generate_results): Likewise.
7221 (release_structures): Likewise.
7223 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7225 PR tree-optimization/82954
7226 * gimple-ssa-store-merging.c
7227 (imm_store_chain_info::coalesce_immediate_stores): If
7228 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7230 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7232 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7233 Upddate call to ENDIAN_LANE_N.
7234 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7235 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7236 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7237 and use aarch64_endian_lane_rtx.
7238 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7240 2017-11-12 Tom de Vries <tom@codesourcery.com>
7242 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7244 2017-11-12 Tom de Vries <tom@codesourcery.com>
7246 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7248 2017-11-12 Tom de Vries <tom@codesourcery.com>
7250 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7252 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7253 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7254 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7256 2017-11-11 Martin Sebor <msebor@redhat.com>
7259 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7262 * prefic.c (translate_name): Replace strncpy with memcpy to
7263 avoid -Wstringop-truncation.
7265 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7267 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7269 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7271 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7273 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7275 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7278 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7280 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7281 enable generating XXBRH if the value is in a vector register.
7282 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7283 value is in a vector register.
7284 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7285 register to register bswap64's instead of doing the GPR sequence
7286 used on previous machines.
7287 (bswapdi2_xxbrd): New insn.
7288 (bswapdi2_reg): Disallow on ISA 3.0.
7289 (register to register bswap64 splitter): Do not split the insn on
7290 ISA 3.0 systems that use XXBRD.
7292 2017-11-10 Martin Sebor <msebor@redhat.com>
7295 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7297 (find_subframework_file): Same.
7299 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7301 * auto-profile.c (afdo_indirect_call): Drop frequency.
7302 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7303 (cgraph_node::create_edge): Drop frequency argument.
7304 (cgraph_node::create_indirect_edge): Drop frequency argument.
7305 (cgraph_edge::make_speculative): Drop frequency arguments.
7306 (cgraph_edge::resolve_speculation): Do not update frequencies
7307 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7308 (cgraph_node::dump): Check consistency in IPA mode.
7309 (cgraph_edge::maybe_hot_p): Use IPA counter.
7310 (cgraph_edge::verify_count_and_frequency): Rename to ...
7311 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7312 (cgraph_node::verify_node): Update.
7313 * cgraph.h (struct cgraph_edge): Drop frequency.
7314 (cgraph_edge::frequency): New function.
7315 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7317 (cgraph_edge::rebuild_edges): Likewise.
7318 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7319 (duplicate_thunk_for_node): Do not pass frequency.
7320 (cgraph_node::create_clone): Scale only counts.
7321 (cgraph_node::create_virtual_clone): Do not pass frequency.
7322 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7323 (cgraph_node::create_version_clone): Do not pass frequency.
7324 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7325 (cgraph_node::expand_thunk): Do not pass frequency.
7326 (cgraph_node::create_wrapper): Do not pass frequency.
7327 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7329 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7330 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7331 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7332 (gather_caller_stats): Use frequency function.
7333 (ipcp_cloning_candidate_p): Use frequency function.
7334 (ipcp_propagate_stage): Use frequency function.
7335 (get_info_about_necessary_edges): Use frequency function.
7336 (update_profiling_info): Update only IPA profile.
7337 (update_specialized_profile): Use frequency functoin.
7338 (perhaps_add_new_callers): Update only IPA profile.
7339 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7340 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7341 (dump_ipa_call_summary): Use frequency function.
7342 (estimate_edge_size_and_time): Use frequency function.
7343 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7344 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7345 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7346 (update_noncloned_counts): ... ths one; scale counts only.
7347 (clone_inlined_nodes): Do not scale frequency.
7348 (inline_call): Do not pass frequency.
7349 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7350 (compute_inlined_call_time): Use IPA profile.
7351 (want_inline_small_function_p): Use IPA profile.
7352 (want_inline_self_recursive_call_p): Use IPA profile.
7353 (edge_badness): Use IPA profile.
7354 (lookup_recursive_calls): Use IPA profile.
7355 (recursive_inlining): Do not pass frequency.
7356 (resolve_noninline_speculation): Do not update frequency.
7357 (inline_small_functions): Collect max of IPA profile.
7358 (dump_overall_stats): Dump IPA porfile.
7359 (dump_inline_stats): Dump IPA porfile.
7360 (ipa_inline): Collect IPA stats.
7361 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7362 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7363 (ipa_propagate_frequency): Use frequency function.
7364 (ipa_profile): Cleanup.
7365 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7366 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7367 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7368 (input_node): Do not stream frequency.
7369 (input_edge): Do not stream frequency.
7370 (merge_profile_summaries): Scale only IPA profiles.
7371 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7372 * predict.c (drop_profile): Do not recompute frequency.
7373 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7374 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7375 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7376 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7377 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7378 * tree-inline.c (copy_bb): Do not scale frequency.
7379 (expand_call_inline): Do not scale frequency.
7380 (tree_function_versioning): Do not scale frequency.
7381 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7383 2017-11-10 Julia Koval <julia.koval@intel.com>
7385 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7386 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7387 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7388 (_mm256_maskz_gf2p8affine_epi64_epi8)
7389 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7390 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7391 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7392 (__builtin_ia32_vgf2p8affineqb_v32qi)
7393 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7394 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7396 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7400 (arm_option_override): Refactor.
7401 (arm_option_reconfigure_globals): New.
7402 (arm_options_perform_arch_sanity_checks): New.
7403 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7405 (arm_options_perform_arch_sanity_checks): Likewise
7407 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7409 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7410 (power9-qpmul): New.
7411 * rs6000/rs6000.md ("type" attr): Add qmul.
7412 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7413 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7414 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7416 2017-11-10 Martin Sebor <msebor@redhat.com>
7419 * builtins.c (compute_objsize): Handle arrays that
7420 compute_builtin_object_size likes to fail for. Make extern.
7421 * builtins.h (compute_objsize): Declare.
7422 (check_strncpy_sizes): New function.
7423 (expand_builtin_strncpy): Call check_strncpy_sizes.
7424 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7425 -Wstringop-truncation.
7426 (gimple_fold_builtin_strncat): Same.
7427 * gimple.c (gimple_build_call_from_tree): Set call location.
7428 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7429 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7430 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7431 (handle_builtin_strlen): Use strlen_to_stridx.
7432 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7434 Use strlen_to_stridx.
7435 (pass_strlen::execute): Release strlen_to_stridx.
7436 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7437 (-Wstringop-truncation): Document new option.
7439 2017-11-10 Martin Liska <mliska@suse.cz>
7441 PR gcov-profile/82702
7442 * gcov.c (main): Handle intermediate files in a different
7444 (get_gcov_intermediate_filename): New function.
7445 (output_gcov_file): Remove support of intermediate files.
7446 (generate_results): Allocate intermediate file.
7447 (release_structures): Clean-up properly fn_end.
7448 (output_intermediate_file): Start iterating with line 1.
7450 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7452 PR tree-optimization/82929
7453 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7454 ops_swapped_p non-static data member.
7455 (store_immediate_info::store_immediate_info): Clear it.
7456 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7457 ops set ops_swapped_p.
7458 (count_multiple_uses): Handle ops_swapped_p.
7460 2017-11-10 Martin Liska <mliska@suse.cz>
7462 * coverage.c (coverage_init): Stream information about
7463 support of has_unexecuted_blocks.
7464 * doc/gcov.texi: Document that.
7465 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7466 * gcov.c (read_graph_file): Likewise.
7467 (output_line_beginning): Fix a small issue with
7470 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7472 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7473 reference of trivial component.
7475 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7478 * gimple-ssa-store-merging.c
7479 (pass_store_merging::terminate_all_aliasing_chains): For
7480 gimple_store_p stmts also call refs_output_dependent_p.
7482 PR rtl-optimization/82913
7483 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7486 2017-11-09 Jeff Law <law@redhat.com>
7488 * vr-values.h: New file with vr_values class.
7489 * tree-vrp.c: Include vr-values.h
7490 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7491 data objects into the vr_values class.
7492 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7493 (get_value_range): Make it a member function within vr_values class.
7494 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7495 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7496 (op_with_constant_singleton_value_range): Likewise.
7497 (extract_range_for_var_from_comparison_expr): Likewise.
7498 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7499 (extract_range_from_binary_expr): Likewise.
7500 (extract_range_from_unary_expr): Likewise.
7501 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7502 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7503 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7504 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7505 (compare_name_with_value, compare_names): Likewise.
7506 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7507 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7508 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7509 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7510 (extract_range_from_phi_node): Likewise.
7511 (simplify_truth_ops_using_ranges): Likewise.
7512 (simplify_div_or_mod_using_ranges): Likewise.
7513 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7514 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7515 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7516 (simplify_float_conversion_using_ranges): Likewise.
7517 (simplify_internal_call_using_ranges): Likewise.
7518 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7519 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7520 poitner via x_vr_values for calls into gimple folder.
7521 (vrp_initialize_lattice): Make this the vr_values ctor.
7522 (vrp_free_lattice): Make this the vr_values dtor.
7523 (set_vr_value): New function.
7524 (class vrp_prop): Add vr_values data member. Add various member
7525 functions as well as member functions that delegate to vr_values.
7526 (check_array_ref): Make a member function within vrp_prop class.
7527 (search_for_addr_array, vrp_initialize): Likewise.
7528 (vrp_finalize): Likewise. Revamp to avoid direct access to
7529 vr_value, values_propagated, etc.
7530 (check_array_bounds): Extract vrp_prop class instance pointer from
7531 walk info structure. Use it to call member functions.
7532 (check_all_array_refs): Make a member function within vrp_prop class.
7533 Smuggle class instance pointer via walk info structure.
7534 (x_vr_values): New local static.
7535 (vrp_valueize): Use x_vr_values to get class instance.
7536 (vr_valueize_1): Likewise.
7537 (class vrp_folder): Add vr_values data member. Add various member
7538 functions as well as member functions that delegate to vr_values.
7539 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7540 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7541 class instance from vr_values. Use it to call member functions.
7542 (vrp_dom_walker): Add vr_values data member.
7543 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7544 instance via x_vr_values.
7545 (identify_jump_threads): Accept vr_values as argument. Store
7546 it into the walker structure.
7547 (evrp_dom_walker): Add vr_values class data member. Add various
7549 (evrp_dom_walker::try_find_new_range): Use vr_values data
7550 member to access the memory allocator.
7551 (evrp_dom_walker::before_dom_children): Store vr_values class
7552 instance into the vrp_folder class.
7553 (evrp_dom_walker::push_value_range): Rework to avoid direct
7554 access to num_vr_values and vr_value.
7555 (evrp_dom_walker::pop_value_range): Likewise.
7556 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7557 Use vr_values to get to dump_all_value_ranges member function.
7558 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7559 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7560 Pass vr_values class instance down to identify_jump_threads.
7561 Remove call to vrp_free_lattice.
7562 (debug_all_value_ranges): Remove.
7564 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7565 (vrp_folder): Likewise.
7567 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7568 Get it from the existing bitmap instead.
7569 (vrp_intersect_ranges_1): Likewise.
7571 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7573 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7575 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7576 set bit_not_p to it.
7577 (imm_store_chain_info::coalesce_immediate_stores): Break group
7578 if bit_not_p is different.
7579 (count_multiple_uses, split_group,
7580 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7581 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7582 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7583 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7584 to store_immediate_info ctor.
7586 2017-11-09 Jim Wilson <jimw@sifive.com>
7588 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7589 (scan_prog_file): Likewise.
7591 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7593 * bb-reorder.c (max_entry_frequency): Remove.
7594 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7595 connect_traces, push_to_next_round_p): Remove prototypes.
7596 (find_traces_1_round): Use counts only.
7597 (push_to_next_round_p): Likewise.
7598 (find_traces): Likewise.
7599 (rotate_loop): Likewise.
7600 (find_traces_1_round): Likewise.
7601 (connect_traces): Likewise.
7602 (edge_order): Likewise.
7604 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7606 * config/arm/arm.c (output_return_instruction): Add comments to
7607 indicate requirement for cmse_nonsecure_entry return to account
7608 for the size of clearing instruction output here.
7609 (thumb_exit): Likewise.
7610 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7611 return in hardfloat mode.
7613 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7615 * config/rs6000/rs6000.c (machine_function): Add a bool,
7616 "toc_is_wrapped_separately".
7617 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7618 if it wasn't explicitly set or unset, we are optimizing for speed, and
7619 doing separate shrink-wrapping.
7620 (rs6000_get_separate_components): Enable the TOC component if
7621 saving the TOC register in the prologue.
7622 (rs6000_components_for_bb): Handle the TOC component.
7623 (rs6000_emit_prologue_components): Store the TOC register where needed.
7624 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7625 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7627 2017-11-09 Martin Jambor <mjambor@suse.cz>
7629 * ipa-param-manipulation.c: New file.
7630 * ipa-param-manipulation.h: Likewise.
7631 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7632 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7633 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7634 (ipa_parm_adjustment): Likewise.
7635 (ipa_parm_adjustment_vec): Likewise.
7636 (ipa_get_vector_of_formal_parms): Moved declaration to
7637 ipa-param-manipulation.h.
7638 (ipa_get_vector_of_formal_parm_types): Likewise.
7639 (ipa_modify_formal_parameters): Likewise.
7640 (ipa_modify_call_arguments): Likewise.
7641 (ipa_combine_adjustments): Likewise.
7642 (ipa_dump_param_adjustments): Likewise.
7643 (ipa_modify_expr): Likewise.
7644 (ipa_get_adjustment_candidate): Likewise.
7645 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7646 ipa-param-manipulation.c.
7647 (ipa_get_vector_of_formal_parm_types): Likewise.
7648 (ipa_modify_formal_parameters): Likewise.
7649 (ipa_modify_call_arguments): Likewise.
7650 (ipa_modify_expr): Likewise.
7651 (get_ssa_base_param): Likewise.
7652 (ipa_get_adjustment_candidate): Likewise.
7653 (index_in_adjustments_multiple_times_p): Likewise.
7654 (ipa_combine_adjustments): Likewise.
7655 (ipa_dump_param_adjustments): Likewise.
7656 * tree-sra.c: Also include ipa-param-manipulation.h
7657 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7660 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7661 Alan Hayward <alan.hayward@arm.com>
7662 David Sherwood <david.sherwood@arm.com>
7664 * doc/sourcebuild.texi (vect_masked_store): Document.
7666 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7667 Alan Hayward <alan.hayward@arm.com>
7668 David Sherwood <david.sherwood@arm.com>
7670 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7672 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7673 Alan Hayward <alan.hayward@arm.com>
7674 David Sherwood <david.sherwood@arm.com>
7676 * doc/sourcebuild.texi (vect_variable_length): Document.
7678 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7679 Alan Hayward <alan.hayward@arm.com>
7680 David Sherwood <david.sherwood@arm.com>
7682 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7684 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7685 Alan Hayward <alan.hayward@arm.com>
7686 David Sherwood <david.sherwood@arm.com>
7688 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7690 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7691 Alan Hayward <alan.hayward@arm.com>
7692 David Sherwood <david.sherwood@arm.com>
7694 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7695 previously undocumented selectors.
7696 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7698 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7700 * doc/rtl.texi (const_vector): Say that elements can be
7701 const_wide_ints too.
7702 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7703 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7704 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7705 * optabs.c (expand_vector_broadcast): Likewise.
7707 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7708 Alan Hayward <alan.hayward@arm.com>
7709 David Sherwood <david.sherwood@arm.com>
7711 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7712 scaled index even if the unscaled address was invalid.
7713 Don't increase the complexity of using a scale in that case.
7715 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7716 Alan Hayward <alan.hayward@arm.com>
7717 David Sherwood <david.sherwood@arm.com>
7719 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7720 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7721 UNITS_PER_WORD bytes.
7722 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7723 based on REGMODE_NATURAL_SIZE of the inner mode.
7724 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7725 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7726 * expmed.c (lowpart_bit_field_p): Divide the value up into
7727 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7728 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7729 whether something is likely to occupy more than one register.
7731 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7734 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7735 profile_count::adjust_for_ipa_scaling.
7736 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7737 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7739 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7741 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7743 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7744 (split_group): Add total_orig and total_new arguments, estimate the
7745 number of statements related to the store group without store merging
7746 and with store merging.
7747 (imm_store_chain_info::output_merged_store): Adjust split_group
7748 callers, punt if estimated number of statements with store merging
7749 is not smaller than estimated number of statements without it.
7751 (handled_load): Remove has_single_use checks.
7752 (pass_store_merging::process_store): Likewise.
7754 2017-11-09 Richard Biener <rguenther@suse.de>
7756 PR tree-optimization/82902
7757 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7759 2017-11-09 Martin Liska <mliska@suse.cz>
7762 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7765 2017-11-09 Martin Liska <mliska@suse.cz>
7767 PR tree-optimization/82669
7768 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7770 2017-11-09 Martin Liska <mliska@suse.cz>
7772 PR gcov-profile/48463
7773 * coverage.c (coverage_begin_function): Output also end locus
7774 of a function and information whether the function is
7776 * gcov-dump.c (tag_function): Parse and print the information.
7777 * gcov.c (INCLUDE_MAP): Add include.
7778 (INCLUDE_SET): Likewise.
7779 (struct line_info): Move earlier in the source file because
7780 of vector<line_info> in function_info structure.
7781 (line_info::line_info): Likewise.
7782 (line_info::has_block): Likewise.
7783 (struct source_info): Add new member index.
7784 (source_info::get_functions_at_location): New function.
7785 (function_info::group_line_p): New function.
7786 (output_intermediate_line): New function.
7787 (output_intermediate_file): Use the mentioned function.
7788 (struct function_start): New.
7789 (struct function_start_pair_hash): Likewise.
7790 (process_file): Add code that identifies group functions.
7791 Assign lines either to global or function scope.
7792 (generate_results): Skip artificial functions.
7793 (find_source): Assign index for each source file.
7794 (read_graph_file): Read new flag artificial and end_line.
7795 (add_line_counts): Assign it either to global of function scope.
7796 (accumulate_line_counts): Isolate core of the function to
7797 accumulate_line_info and call it for both function and global
7799 (accumulate_line_info): New function.
7800 (output_line_beginning): Fix GNU coding style.
7801 (print_source_line): New function.
7802 (output_line_details): Likewise.
7803 (output_function_details): Likewise.
7804 (output_lines): Iterate both source (global) scope and function
7806 (struct function_line_start_cmp): New class.
7807 * doc/gcov.texi: Reflect changes in documentation.
7809 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7812 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7813 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7814 and similarly for not instead of neg.
7816 2017-11-08 Andi Kleen <ak@linux.intel.com>
7818 * config/i386/i386.opt: Add -mforce-indirect-call.
7819 * config/i386/predicates.md: Check for flag_force_indirect_call.
7820 * doc/invoke.texi: Document -mforce-indirect-call
7822 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7824 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7826 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7827 config/riscv/riscv.c (predict.h): New include.
7828 (riscv_slow_unaligned_access_p): No longer static.
7829 (riscv_block_move_straight): Add require.
7830 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7832 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7835 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7836 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7837 nonimmediate_operand predicate for operand 1 instead of
7840 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7842 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7844 * config/aarch64/constraints.md (Uml): New constraint.
7845 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7848 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7850 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7851 of two vec_duplicates into a vec_concat.
7853 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7855 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7856 Simplify vec_merge of vec_duplicate and vec_concat.
7857 * config/aarch64/constraints.md (Utq): New constraint.
7858 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7861 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7863 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7864 Simplify vec_merge of vec_duplicate and const_vector.
7865 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7867 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7868 mode iterator. Update predicate on operand 1 to
7869 handle non-const_vec constants. Delete constraints.
7870 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7872 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7874 PR tree-optimization/78821
7875 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7877 (store_operand_info::store_operand_info): Initialize it to false.
7878 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7879 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7880 store in the group, and if chain_info is non-NULL, to ignore altogether
7882 (compatible_load_p): Fail if bit_not_p does not match.
7883 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7884 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7885 (pass_store_merging::process_store): Adjust
7886 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7887 call terminate_all_aliasing_chains newly when adding a store into
7888 a chain with non-NULL chain_info.
7890 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7892 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7894 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7896 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7898 (aarch64_layout_frame): Initialise emit_frame_chain.
7899 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7900 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7902 2017-11-08 Martin Liska <mliska@suse.cz>
7904 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7905 of gimple_call_internal_p.
7907 2017-11-07 Tom de Vries <tom@codesourcery.com>
7909 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7911 2017-11-07 Tom de Vries <tom@codesourcery.com>
7913 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7916 2017-11-08 Martin Liska <mliska@suse.cz>
7919 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7921 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7923 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7925 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7928 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7929 (*cmp<mode>_ccz_1): New insn with $k alternative.
7932 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7933 RTX_COMM_COMPARE as commutative as well.
7934 (ix86_binary_operator_ok): Formatting fix.
7935 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7936 *<code><mode>3<mask_name><round_saeonly_name>,
7937 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7938 *<s>mul<mode>3_highpart<mask_name>,
7939 *vec_widen_umult_even_v16si<mask_name>,
7940 *vec_widen_umult_even_v8si<mask_name>,
7941 *vec_widen_umult_even_v4si<mask_name>,
7942 *vec_widen_smult_even_v16si<mask_name>,
7943 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7944 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7945 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7946 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7947 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7948 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7949 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7950 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7951 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7952 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7953 ix86_binary_operator_ok. Formatting fixes.
7954 (*<plusminus_insn><mode>3<mask_name><round_name>,
7955 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7958 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7960 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7961 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7962 eq and ne if TARGET_ISEL.
7963 (cmp): New code_iterator.
7964 (UNS, UNSU_, UNSIK): New code_attrs.
7965 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7966 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7968 ("eq<mode>3"): ... this.
7969 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7971 ("ne<mode>3"): ... this.
7973 2017-11-07 Julia Koval <julia.koval@intel.com>
7976 * common/config/i386/i386-common.c
7977 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7978 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7979 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7980 * config/i386/i386.opt: Ditto.
7981 * config/i386/i386.c (ix86_target_string): Ditto.
7982 (ix86_option_override_internal): Ditto.
7983 (ix86_init_mpx_builtins): Move MPX to args2.
7984 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7985 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7986 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7987 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7988 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7989 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7990 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7992 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7995 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7996 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7997 (zero-extendsidi peephole2): Remove peephole.
7999 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8002 * stor-layout.c: Include attribs.h.
8003 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8004 explicit lookup of "aligned" attribute.
8006 2017-11-07 Andrew Waterman <andrew@sifive.com>
8008 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8009 (riscv_expand_block_move): Likewise.
8010 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8012 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8013 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8014 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8015 (riscv_adjust_block_mem): Likewise.
8016 (riscv_block_move_loop): Likewise.
8017 (riscv_expand_block_move): Likewise.
8018 * config/riscv/riscv.md (movmemsi): New pattern.
8020 2017-11-07 Michael Clark <michaeljclark@mac.com>
8022 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8023 (MUSL_DYNAMIC_LINKER): Likewise.
8025 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8027 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8028 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8031 2017-11-07 Richard Biener <rguenther@suse.de>
8033 * match.pd: Fix build.
8035 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8036 Jackson Woodruff <jackson.woodruff@arm.com>
8038 PR tree-optimization/71026
8039 * match.pd: Canonicalize negate in division.
8041 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8044 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8046 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8048 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8049 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8051 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8053 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8054 non-scalar integral types.
8055 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8056 (-(A-B), -(~A)): New transformations.
8058 2017-11-07 Tom de Vries <tom@codesourcery.com>
8060 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8061 semicolon after "do {} while (0)".
8062 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8063 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8064 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8065 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8066 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8068 2017-11-07 Tom de Vries <tom@codesourcery.com>
8070 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8071 after "do {} while (0)".
8072 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8073 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8074 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8075 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8076 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8078 2017-11-07 Tom de Vries <tom@codesourcery.com>
8081 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8083 (arm_rtx_costs_internal): Add missing semicolon after
8084 HANDLE_NARROW_SHIFT_ARITH call.
8086 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8088 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8089 disable isel if it was not set explicitly.
8091 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8093 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8094 (add_type_attribute) likewise.
8096 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8098 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8099 of struct ix86_frame.
8100 (ix86_initial_elimination_offset): Likewise.
8101 (ix86_expand_split_stack_prologue): Likewise.
8103 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8105 * tree-vrp.h (enum value_range_type): Update stale comment.
8107 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8108 Alan Hayward <alan.hayward@arm.com>
8109 David Sherwood <david.sherwood@arm.com>
8111 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8112 (aarch64_expand_vec_perm_const): Take the number of units too.
8113 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8114 (aarch64_expand_vec_perm_const): Likewise.
8115 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8116 (vec_perm<mode>): Update accordingly.
8118 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8119 Alan Hayward <alan.hayward@arm.com>
8120 David Sherwood <david.sherwood@arm.com>
8122 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8123 Take the number of units too.
8124 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8125 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8126 but check for a vector mode before rather than after the call.
8127 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8128 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8129 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8130 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8131 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8132 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8133 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8134 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8135 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8136 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8137 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8138 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8139 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8140 (aarch64_sqdmull2_n<mode>): Update accordingly.
8142 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8143 Alan Hayward <alan.hayward@arm.com>
8144 David Sherwood <david.sherwood@arm.com>
8146 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8147 the number of units too.
8148 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8149 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8150 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8151 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8152 (vec_store_lanesxi<mode>): Update accordingly.
8154 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8155 Alan Hayward <alan.hayward@arm.com>
8156 David Sherwood <david.sherwood@arm.com>
8158 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8159 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8160 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8161 of units rather than the mode.
8162 * config/aarch64/iterators.md (nunits): New mode attribute.
8163 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8164 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8165 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8166 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8167 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8168 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8169 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8170 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8171 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8172 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8173 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8174 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8175 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8176 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8177 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8178 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8179 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8180 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8181 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8182 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8183 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8184 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8185 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8186 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8187 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8188 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8189 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8190 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8191 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8192 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8193 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8194 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8195 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8196 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8197 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8198 (aarch64_simd_vec_setv2di): Likewise.
8200 2017-11-06 Carl Love <cel@us.ibm.com>
8202 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8204 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8205 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8207 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8209 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8210 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8211 Add power 8 macro expansions.
8212 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8213 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8214 power 8 instructions. (VSX_XXBR): Add iterator.
8216 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8218 * config/arm/arm.md (predicable_short_it): Change default to "no",
8219 improve documentation, remove uses that are identical to the default.
8220 (enabled_for_depr_it): Rename to enabled_for_short_it.
8221 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8222 Remove default uses.
8223 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8224 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8225 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8226 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8228 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8231 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8232 float128 helper macros, which are no longer used after deleting
8233 the old 'q' built-in functions, and moving the round to odd
8234 built-in functions to being special built-in functions.
8235 (BU_FLOAT128_2): Likewise.
8236 (BU_FLOAT128_1_HW): Likewise.
8237 (BU_FLOAT128_2_HW): Likewise.
8238 (BU_FLOAT128_3_HW): Likewise.
8239 (FABSQ): Delete old 'q' built-in functions.
8240 (COPYSIGNQ): Likewise.
8241 (SQRTF128_ODD): Move round to odd built-in functions to be
8242 special built-in functions, so that we can handle
8243 -mabi=ieeelongdouble.
8244 (TRUNCF128_ODD): Likewise.
8245 (ADDF128_ODD): Likewise.
8246 (SUBF128_ODD): Likewise.
8247 (MULF128_ODD): Likewise.
8248 (DIVF128_ODD): Likewise.
8249 (FMAF128_ODD): Likewise.
8250 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8251 built-in names to 'f128'.
8252 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8253 old 'q' built-in functions, as the machine independent code for
8254 'f128' built-in functions handles this.
8255 (rs6000_expand_builtin): Add expansion for float128 round to odd
8256 functions, keying off on -mabi=ieeelongdouble of whether to use
8257 the KFmode or TFmode variant.
8258 (rs6000_init_builtins): Initialize the _Float128 round to odd
8260 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8261 _Float128 'q' built-in functions are now mapped into the new
8262 'f128' built-in functions.
8264 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8266 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8267 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8269 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8272 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8273 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8274 * toplev.c: Include "ipa-fnsummary.h".
8275 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8277 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8279 PR tree-optimization/82838
8280 * gimple-ssa-store-merging.c
8281 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8282 on a separate gimple_seq which is then appended to seq.
8284 2017-11-06 Jeff Law <law@redhat.com>
8287 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8288 (get_probe_interval): New functions.
8289 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8290 (ix86_adjust_stack_and_probe): Likewise.
8291 (output_adjust_stack_and_probe): Likewise.
8292 (ix86_emit_probe_stack_range): Likewise.
8293 (ix86_expand_prologue): Likewise.
8295 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8297 PR tree-optimization/82816
8298 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8299 if the modes of the two types are the same.
8300 (convert_plusminus_to_widen): Likewise.
8302 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8304 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8306 (usadv16qi): New define_expand.
8307 (usadv8hi): New define_expand.
8309 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8312 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8314 (inline_transform): Likewise.
8315 * predict.c (counts_to_freqs): Remove useless conditional.
8316 * profile-count.h (profile_count::apply_scale): Move sanity check.
8317 * tree-inline.c (copy_bb): Always scale.
8318 (copy_cfg_body): Likewise.
8320 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8323 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8325 (*cmp_ite0): Add enabled_for_depr_it attribute.
8326 (*cmp_ite1): Likewise.
8328 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8330 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8333 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8335 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8336 return the value by pointer.
8337 (extract_range_from_multiplicative_op_1): Update accordingly.
8338 Return as soon as an operation fails.
8340 2017-11-05 Tom de Vries <tom@codesourcery.com>
8343 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8344 (DEF_SANITIZER_BUILTIN): ... here.
8345 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8346 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8348 2017-11-05 Tom de Vries <tom@codesourcery.com>
8351 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8353 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8354 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8355 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8357 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8358 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8359 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8361 2017-11-05 Tom de Vries <tom@codesourcery.com>
8364 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8367 2017-11-04 Michael Clark <michaeljclark@mac.com>
8369 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8370 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8372 (*addsi3_extended): Likewise.
8373 (*addsi3_extended2): Likewise.
8374 (<optab>si3): Likewise.
8375 (<optab>di3): Likewise.
8376 (<optab><mode>3): Likewise.
8377 (<*optabe>si3_internal): Likewise.
8378 (zero_extendqi<SUPERQI:mode>2): Likewise.
8379 (*add<mode>hi3): Likewise.
8380 (*xor<mode>hi3): Likewise.
8381 (<optab>di3): Likewise.
8382 (*<optab>si3_extend): Likewise.
8383 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8384 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8385 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8387 2017-11-04 Andrew Waterman <andrew@sifive.com>
8389 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8390 TARGET_STRICT_ALIGN based upon -mtune argument.
8392 2017-11-04 Andrew Waterman <andrew@sifive.com>
8394 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8396 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8398 * config/i386/i386.c (choose_basereg): Use optional scratch
8399 register and add assertion.
8400 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8401 needed, and don't allocate stack.
8402 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8403 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8404 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8406 2017-11-03 Jeff Law <law@redhat.com>
8408 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8409 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8410 to probe at the start of a noreturn function.
8412 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8414 PR tree-optimization/78821
8415 * gimple-ssa-store-merging.c: Update the file comment.
8416 (MAX_STORE_ALIAS_CHECKS): Define.
8417 (struct store_operand_info): New type.
8418 (store_operand_info::store_operand_info): New constructor.
8419 (struct store_immediate_info): Add rhs_code and ops data members.
8420 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8421 and op1r arguments to the ctor, initialize corresponding data members.
8422 (struct merged_store_group): Add load_align_base and load_align
8424 (merged_store_group::merged_store_group): Initialize them.
8425 (merged_store_group::do_merge): Update them.
8426 (merged_store_group::apply_stores): Pick the constant for
8427 encode_tree_to_bitpos from one of the two operands, or skip
8428 encode_tree_to_bitpos if neither operand is a constant.
8429 (class pass_store_merging): Add process_store method decl. Remove
8430 bool argument from terminate_all_aliasing_chains method decl.
8431 (pass_store_merging::terminate_all_aliasing_chains): Remove
8432 var_offset_p argument and corresponding handling.
8433 (stmts_may_clobber_ref_p): New function.
8434 (compatible_load_p): New function.
8435 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8436 if there is overlap and rhs_code is not INTEGER_CST. For
8437 non-overlapping stores terminate group if rhs is not mergeable.
8438 (get_alias_type_for_stmts): Change first argument from
8439 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8440 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8441 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8443 (get_location_for_stmts): Change first argument from
8444 auto_vec<gimple *> & to vec<gimple *> &.
8445 (struct split_store): Remove orig_stmts data member, add orig_stores.
8446 (split_store::split_store): Create orig_stores rather than orig_stmts.
8447 (find_constituent_stmts): Renamed to ...
8448 (find_constituent_stores): ... this. Change second argument from
8449 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8450 to info structures rather than the statements.
8451 (split_group): Rename ALLOW_UNALIGNED argument to
8452 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8453 it. Adjust find_constituent_stores caller.
8454 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8455 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8456 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8457 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8458 (mem_valid_for_store_merging): New function.
8459 (handled_load): New function.
8460 (pass_store_merging::process_store): New method.
8461 (pass_store_merging::execute): Use process_store method. Adjust
8462 terminate_all_aliasing_chains caller.
8464 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8466 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8467 Return true for more constants, symbols and label references.
8468 (aarch64_valid_floating_const): Remove unused function.
8470 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8473 * config/aarch64/aarch64.c (aarch64_layout_frame):
8474 Undo forcing of LR at bottom of frame.
8476 2017-11-03 Jeff Law <law@redhat.com>
8479 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8480 for int_registers_saved.
8482 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8483 extracted from tree-ssa-dom.c.
8484 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8485 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8486 (record_equivalences_from_incoming_edge): Add additional argument
8487 to single_pred_edge_ignoring_loop_edges call.
8488 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8489 (uncprop_dom_walker::before_dom_children): Add additional argument
8490 to single_pred_edge_ignoring_loop_edges call.
8491 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8492 single_pred_edge_ignoring_loop_edges rather than open coding.
8493 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8495 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8497 * match.pd (-(-A)): Rewrite.
8499 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8501 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8502 (rs6000_emit_int_cmove): New declaration.
8503 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8504 (rs6000_emit_sISEL): Delete.
8505 (rs6000_emit_int_cmove): Make non-static.
8506 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8507 instead of rs6000_emit_sISEL.
8509 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8511 * asan.c (create_cond_insert_point): Maintain profile.
8512 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8514 * basic-block.h (struct basic_block_def): Remove frequency.
8515 (EDGE_FREQUENCY): Use to_frequency
8516 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8518 (find_traces): Update to use to_frequency.
8519 (find_traces_1_round): Likewise; use only IPA counts.
8520 (bb_to_key): Likewise.
8521 (connect_traces): Use IPA counts only.
8522 (copy_bb_p): Update to use to_frequency.
8523 (fix_up_crossing_landing_pad): Likewise.
8524 (sanitize_hot_paths): Likewise.
8525 * bt-load.c (basic_block_freq): Likewise.
8526 * cfg.c (init_flow): Set count_max to uninitialized.
8527 (check_bb_profile): Remove frequencies; check counts.
8528 (dump_bb_info): Do not dump frequencies.
8529 (update_bb_profile_for_threading): Update counts only.
8530 (scale_bbs_frequencies_int): Likewise.
8531 (MAX_SAFE_MULTIPLIER): Remove.
8532 (scale_bbs_frequencies_gcov_type): Update counts only.
8533 (scale_bbs_frequencies_profile_count): Update counts only.
8534 (scale_bbs_frequencies): Update counts only.
8535 * cfg.h (struct control_flow_graph): Add count-max.
8536 (update_bb_profile_for_threading): Update prototype.
8537 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8538 (find_many_sub_basic_blocks): Likewise.
8539 * cfgcleanup.c (try_forward_edges): Likewise.
8540 (try_crossjump_to_edge): Likewise.
8541 * cfgexpand.c (expand_gimple_cond): Likewise.
8542 (expand_gimple_tailcall): Likewise.
8543 (construct_init_block): Likewise.
8544 (construct_exit_block): Likewise.
8545 * cfghooks.c (verify_flow_info): Check consistency of counts.
8546 (dump_bb_for_graph): Do not dump frequencies.
8547 (split_block_1): Do not update frequencies.
8548 (split_edge): Do not update frequencies.
8549 (make_forwarder_block): Do not update frequencies.
8550 (duplicate_block): Do not update frequencies.
8551 (account_profile_record): Do not update frequencies.
8552 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8553 for global heuristics.
8554 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8555 (expected_loop_iterations_unbounded): Use counts only.
8556 * cfgloopmanip.c (scale_loop_profile): Simplify.
8557 (create_empty_loop_on_edge): Simplify
8559 (duplicate_loop_to_header_edge): Simplify
8560 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8561 (update_br_prob_note): Take care of removing note when profile
8563 (relink_block_chain): Do not dump frequency.
8564 (rtl_account_profile_record): Use to_frequency.
8565 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8566 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8567 (cgraph_update_edges_for_call_stmt_node): Likewise.
8568 (cgraph_edge::verify_count_and_frequency): Update.
8569 (cgraph_node::verify_node): Temporarily disable frequency verification.
8570 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8571 to_cgraph_frequency.
8572 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8573 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8575 (cgraph_node::expand_thunk): Update profile.
8576 * except.c (dw2_build_landing_pads): Do not update frequency.
8577 * final.c (compute_alignments): Use to_frequency.
8578 (dump_basic_block_info): Do not dump frequency.
8579 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8580 (dump_gimple_bb_header): Do not dump frequency.
8581 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8583 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8584 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8585 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8586 (init_before_recovery): Do not update frequency.
8587 (sched_create_recovery_edges): Do not update frequency.
8588 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8589 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8590 (ipa_cp_c_finalize): Set max_count to uninitialized.
8591 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8592 (param_change_prob): Use counts.
8593 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8595 * ipa-split.c (consider_split): Use to_frequency.
8596 (split_function): Use to_frequency.
8597 * ira-build.c (loop_compare_func): Likewise.
8598 (mark_loops_for_removal): Likewise.
8599 (mark_all_loops_for_removal): Likewise.
8600 * loop-doloop.c (doloop_modify): Do not update frequency.
8601 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8603 * lto-streamer-in.c (input_function): Update count_max.
8604 * omp-expand.c (expand_omp_taskreg): Update count_max.
8605 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8606 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8607 (maybe_hot_count_p): Use ipa counts only.
8608 (maybe_hot_bb_p): Simplify.
8609 (maybe_hot_edge_p): Simplify.
8610 (probably_never_executed): Do not take frequency argument.
8611 (probably_never_executed_bb_p): Do not pass frequency.
8612 (probably_never_executed_edge_p): Likewise.
8613 (combine_predictions_for_bb): Check that profile is nonzero.
8614 (propagate_freq): Do not set frequency.
8615 (drop_profile): Simplify.
8616 (counts_to_freqs): Simplify.
8617 (expensive_function_p): Use to_frequency.
8618 (propagate_unlikely_bbs_forward): Simplify.
8619 (determine_unlikely_bbs): Simplify.
8620 (estimate_bb_frequencies): Add hack to silence graphite issues.
8621 (compute_function_frequency): Use ipa counts.
8622 (pass_profile::execute): Update.
8623 (rebuild_frequencies): Use counts only.
8624 (force_edge_cold): Use counts only.
8625 * profile-count.c (profile_count::dump): Dump new count types.
8626 (profile_count::differs_from_p): Check compatiblity.
8627 (profile_count::to_frequency): New function.
8628 (profile_count::to_cgraph_frequency): New function.
8629 * profile-count.h (struct function): Declare.
8630 (enum profile_quality): Add profile_guessed_local and
8631 profile_guessed_global0.
8632 (class profile_proability): Decrease number of bits to 29;
8633 update from_reg_br_prob_note and to_reg_br_prob_note.
8634 (class profile_count: Update comment; decrease number of bits
8635 to 61. Check compatibility.
8636 (profile_count::compatible_p): New private member function.
8637 (profile_count::ipa_p): New member function.
8638 (profile_count::operator<): Handle global zero correctly.
8639 (profile_count::operator>): Handle global zero correctly.
8640 (profile_count::operator<=): Handle global zero correctly.
8641 (profile_count::operator>=): Handle global zero correctly.
8642 (profile_count::nonzero_p): New member function.
8643 (profile_count::force_nonzero): New member function.
8644 (profile_count::max): New member function.
8645 (profile_count::apply_scale): Handle IPA scalling.
8646 (profile_count::guessed_local): New member function.
8647 (profile_count::global0): New member function.
8648 (profile_count::ipa): New member function.
8649 (profile_count::to_frequency): Declare.
8650 (profile_count::to_cgraph_frequency): Declare.
8651 * profile.c (OVERLAP_BASE): Delete.
8652 (compute_frequency_overlap): Delete.
8653 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8654 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8655 * sched-ebb.c (rank): Use counts only.
8656 * shrink-wrap.c (handle_simple_exit): Use counts only.
8657 (try_shrink_wrapping): Use counts only.
8658 (place_prologue_for_one_component): Use counts only.
8659 * tracer.c (find_best_predecessor): Use to_frequency.
8660 (find_trace): Use to_frequency.
8661 (tail_duplicate): Use to_frequency.
8662 * trans-mem.c (expand_transaction): Do not update frequency.
8663 * tree-call-cdce.c: Do not update frequency.
8664 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8665 (gimple_merge_blocks): Likewise.
8666 (gimple_split_edge): Likewise.
8667 (gimple_duplicate_sese_region): Likewise.
8668 (gimple_duplicate_sese_tail): Likewise.
8669 (move_sese_region_to_fn): Likewise.
8670 (gimple_account_profile_record): Likewise.
8671 (insert_cond_bb): Likewise.
8672 * tree-complex.c (expand_complex_div_wide): Likewise.
8673 * tree-eh.c (lower_resx): Update profile.
8674 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8676 (initialize_cfun): Do not initialize frequencies
8677 (freqs_to_counts): Delete.
8678 (copy_cfg_body): Ignore count parameter.
8679 (copy_body): Update.
8680 (expand_call_inline): Update count_max.
8681 (optimize_inline_calls): Update count_max.
8682 (tree_function_versioning): Update count_max.
8683 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8684 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8686 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8687 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8688 (try_peel_loop): Likewise.
8689 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8691 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8692 (tree_transform_and_unroll_loop): Do not use frequencies
8693 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8694 Use reliable prediction only.
8695 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8696 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8697 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8698 probability scaling.
8699 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8700 not update frequency
8701 (any_remaining_duplicated_blocks): Likewise.
8702 (update_profile): Likewise.
8703 (estimated_freqs_path): Delete.
8704 (freqs_to_counts_path): Delete.
8705 (clear_counts_path): Delete.
8706 (ssa_fix_duplicate_block_edges): Likewise.
8707 (duplicate_thread_path): Likewise.
8708 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8709 * tree-tailcall.c (decrease_profile): Do not update frequency.
8710 (eliminate_tail_call): Likewise.
8711 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8712 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8713 (optimize_mask_stores): Likewise.
8714 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8715 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8716 (ubsan_expand_ptr_ifn): Update profile.
8717 * value-prof.c (gimple_ic): Simplify.
8718 * value-prof.h (gimple_ic): Update prototype.
8719 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8720 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8722 (want_inline_self_recursive_call_p): Likewise.
8723 (resolve_noninline_speculation): Only cummulate defined counts.
8724 (inline_small_functions): Use nonzero_p.
8725 (ipa_inline): Do not access freed node.
8727 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8729 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8730 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8732 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8734 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8735 non-legitimate address.
8737 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8739 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8740 (*lt0_<mode>di, *lt0_<mode>si): New.
8742 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8744 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8745 TARGET_PAIRED_FLOAT.
8747 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8748 Jim Wilson <jim.wilson@linaro.org>
8750 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8751 * config/aarch64/aarch64-tune.md: Regenerated.
8752 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8753 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8755 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8757 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8758 (arc_expand_prologue): Restore blink for millicode.
8759 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8761 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8764 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8765 gen_vec_duplicate after forcing the scalar into a register.
8767 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8769 * combine (try_combine): Print the insns input to try_combine to the
8772 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8775 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8776 Remove second argument from aarch64_process_target_attr call.
8777 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8779 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8781 (aarch64_handle_attr_arch): Remove second argument.
8782 (aarch64_handle_attr_cpu): Ditto.
8783 (aarch64_handle_attr_tune): Ditto.
8784 (aarch64_handle_attr_isa_flags): Ditto.
8785 (aarch64_process_one_target_attr): Ditto.
8786 (aarch64_process_target_attr): Ditto.
8787 (aarch64_option_valid_attribute_p): Remove second argument.
8788 on aarch64_process_target_attr call.
8790 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8792 * diagnostic.c: Include "selftest-diagnostic.h".
8793 (selftest::assert_location_text): New function.
8794 (selftest::test_diagnostic_get_location_text): New function.
8795 (selftest::diagnostic_c_tests): Call it.
8797 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8799 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8800 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8801 (class selftest::test_diagnostic_context): Move to...
8802 * selftest-diagnostic.c: New file.
8803 * selftest-diagnostic.h: New file.
8805 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8807 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8808 offset range for FT32B.
8809 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8810 * config/ft32/ft32.md: Add TARGET_NOPM.
8811 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8812 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8814 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8816 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8818 2017-11-02 Jeff Law <law@redhat.com>
8820 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8821 virtual keyword on FINAL OVERRIDE members.
8823 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8824 virtuals together. Add virtual destructor.
8825 (substitute_and_fold_engine): Similarly.
8827 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8829 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8831 2017-11-02 Richard Biener <rguenther@suse.de>
8833 PR tree-optimization/82795
8834 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8836 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8838 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8840 (gcc_GAS_CHECK_FEATURE): Remove.
8841 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8842 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8843 * configure: Regenerate.
8845 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8847 * config/arc/arc.c (hwloop_optimize): Account for empty
8850 2017-11-02 Richard Biener <rguenther@suse.de>
8853 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8854 Truncate ARRAY_REF index and element size.
8856 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8858 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8860 2017-11-01 Jeff Law <law@redhat.com>
8862 * tree-ssa-ccp.c (ccp_folder): New class derived from
8863 substitute_and_fold_engine.
8864 (ccp_folder::get_value): New member function.
8865 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8866 (ccp_fold_stmt): Remove prototype.
8867 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8868 * tree-ssa-copy.c (copy_folder): New class derived from
8869 substitute_and_fold_engine.
8870 (copy_folder::get_value): Renamed from get_value.
8871 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8872 * tree-vrp.c (vrp_folder): New class derived from
8873 substitute_and_fold_engine.
8874 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8875 (vrp_folder::get_value): New member function.
8876 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8877 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8878 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8879 provide a class interface to folder/substitute routines.
8880 (ssa_prop_fold_stmt_fn): Remove typedef.
8881 (ssa_prop_get_value_fn): Likewise.
8882 (subsitute_and_fold): Remove prototype.
8883 (replace_uses_in): Likewise.
8884 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8885 Renamed from replace_uses_in. Call the virtual member function
8886 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8887 (substitute_and_fold_dom_walker): Remove initialization of
8888 data member entries for calbacks. Add substitute_and_fold_engine
8889 member and initialize it.
8890 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8891 member functions for get_value, replace_phi_args_in c
8892 replace_uses_in, and fold_stmt calls.
8893 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8894 substitute_and_fold. Remove assert. Update ctor call.
8896 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8897 (ssa_prop_visit_phi_fn): Likewise.
8898 (class ssa_propagation_engine): New class to provide an interface
8900 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8902 (ssa_prop_visit_phi): Likewise.
8903 (ssa_propagation_engine::simulate_stmt): Moved into class.
8904 Call visit_phi/visit_stmt from the class rather than via
8905 file scoped static variables.
8906 (ssa_propagation_engine::simulate_block): Moved into class.
8907 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8908 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8909 set file scoped statics for the visit_stmt/visit_phi callbacks.
8910 * tree-complex.c (complex_propagate): New class derived from
8911 ssa_propagation_engine.
8912 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8913 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8914 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8916 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8917 ssa_propagation_engine.
8918 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8919 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8920 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8921 * tree-ssa-copy.c (copy_prop): New class derived from
8922 ssa_propagation_engine.
8923 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8924 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8925 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8926 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8927 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8928 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8929 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8931 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8933 PR rtl-optimization/82778
8934 PR rtl-optimization/82597
8935 * compare-elim.c (struct comparison): Add in_a_setter field.
8936 (find_comparison_dom_walker::before_dom_children): Remove killed
8937 bitmap and df_simulate_find_defs call, instead walk the defs.
8938 Compute last_setter and initialize in_a_setter. Merge definitions
8939 with first initialization for a few variables.
8940 (try_validate_parallel): Use insn_invalid_p instead of
8941 recog_memoized. Return insn rather than just the pattern.
8942 (try_merge_compare): Fix up comment. Don't uselessly test if
8943 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8945 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8948 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8949 Alan Hayward <alan.hayward@arm.com>
8950 David Sherwood <david.sherwood@arm.com>
8952 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8953 aarch64_hard_regno_nregs to get the number of registers
8956 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8957 Alan Hayward <alan.hayward@arm.com>
8958 David Sherwood <david.sherwood@arm.com>
8960 * config/aarch64/constraints.md (Upl): Rename to...
8962 * config/aarch64/aarch64.md
8963 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8966 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8967 Alan Hayward <alan.hayward@arm.com>
8968 David Sherwood <david.sherwood@arm.com>
8970 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8971 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8974 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8975 Alan Hayward <alan.hayward@arm.com>
8976 David Sherwood <david.sherwood@arm.com>
8978 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8979 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8980 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8982 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8983 of a single element.
8984 * config/aarch64/iterators.md: Add a comment above the permute
8985 unspecs to say that they are generated directly by
8986 aarch64_expand_vec_perm_const.
8987 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8989 2017-11-01 Nathan Sidwell <nathan@acm.org>
8991 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8993 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8995 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8996 and medany code models, and describe what they do.
8998 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9000 Revert accidental duplicate:
9002 * combine.c (can_change_dest_mode): Reject changes in
9003 REGMODE_NATURAL_SIZE.
9005 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9007 PR rtl-optimization/64682
9008 PR rtl-optimization/69567
9009 PR rtl-optimization/69737
9010 PR rtl-optimization/82683
9011 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9012 register mentioned in the note, drop the note, unless it came from I3,
9013 in which case it should go to I3 again.
9015 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9017 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9018 and return false if not.
9019 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9021 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9023 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9024 range is known to be empty.
9026 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9027 Alan Hayward <alan.hayward@arm.com>
9028 David Sherwood <david.sherwood@arm.com>
9030 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9031 and CONST_VECTOR_NUNITS instead of computing the number of units from
9032 the byte sizes of the vector and element.
9033 (simplify_binary_operation_1): Likewise.
9034 (simplify_const_binary_operation): Likewise.
9035 (simplify_ternary_operation): Likewise.
9037 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9038 Alan Hayward <alan.hayward@arm.com>
9039 David Sherwood <david.sherwood@arm.com>
9041 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9042 (int_mem_offset): ...this new function.
9043 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9044 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9045 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9048 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9049 Alan Hayward <alan.hayward@arm.com>
9050 David Sherwood <david.sherwood@arm.com>
9052 * lower-subreg.c (interesting_mode_p): New function.
9053 (compute_costs, find_decomposable_subregs, decompose_register)
9054 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9055 (resolve_clobber, dump_choices): Use it.
9057 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9058 Alan Hayward <alan.hayward@arm.com>
9059 David Sherwood <david.sherwood@arm.com>
9061 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
9068 truncations. Check the precision rather than the size.
9070 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9071 Alan Hayward <alan.hayward@arm.com>
9072 David Sherwood <david.sherwood@arm.com>
9074 * machmode.h (is_narrower_int_mode): New function
9075 * optabs.c (expand_float, expand_fix): Use it.
9076 * dwarf2out.c (rotate_loc_descriptor): Likewise.
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 * rtl.h (narrower_subreg_mode): New function.
9083 * ira-color.c (update_costs_from_allocno): Use it.
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 * optabs-query.h (convert_optab_p): New function, split out from...
9090 (convert_optab_handler): ...here.
9091 (widening_optab_handler): Delete.
9092 (find_widening_optab_handler): Remove permit_non_widening parameter.
9093 (find_widening_optab_handler_and_mode): Likewise. Provide an
9094 override that operates on mode class wrappers.
9095 * optabs-query.c (widening_optab_handler): Delete.
9096 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9097 parameter. Assert that the two modes are the same class and that
9098 the "from" mode is narrower than the "to" mode. Use
9099 convert_optab_handler instead of widening_optab_handler.
9100 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9101 instead of widening_optab_handler.
9102 * expr.c (expand_expr_real_2): Update calls to
9103 find_widening_optab_handler.
9104 * optabs.c (expand_widen_pattern_expr): Likewise.
9105 (expand_binop_directly): Take the insn_code as a parameter.
9106 (expand_binop): Only call find_widening_optab_handler for
9107 conversion optabs; use optab_handler otherwise. Update calls
9108 to find_widening_optab_handler and expand_binop_directly.
9109 Use convert_optab_handler instead of widening_optab_handler.
9110 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9111 find_widening_optab_handler and use scalar_mode rather than
9113 (convert_plusminus_to_widen): Likewise.
9115 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9116 Alan Hayward <alan.hayward@arm.com>
9117 David Sherwood <david.sherwood@arm.com>
9119 * machmode.h (fixed_size_mode): New class.
9120 * rtl.h (get_pool_mode): Return fixed_size_mode.
9121 * gengtype.c (main): Add fixed_size_mode.
9122 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9123 (get_raw_arg_mode): Likewise.
9124 * doc/tm.texi: Regenerate.
9125 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9126 * targhooks.c (default_get_reg_raw_mode): Likewise.
9127 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9128 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9129 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9130 (msp430_get_raw_result_mode): Likewise.
9131 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9132 * dbxout.c (dbxout_parms): Require fixed-size modes.
9133 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9134 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9135 * omp-low.c (lower_oacc_reductions): Likewise.
9136 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9137 (simplify_subreg): Update accordingly.
9138 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9139 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9140 that aren't fixed-size.
9141 (get_pool_mode): Return a fixed_size_mode.
9142 (output_constant_pool_2): Take a fixed_size_mode.
9144 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9145 Alan Hayward <alan.hayward@arm.com>
9146 David Sherwood <david.sherwood@arm.com>
9148 * doc/rtl.texi (vec_series): Document.
9149 (const): Say that the operand can be a vec_series.
9150 * rtl.def (VEC_SERIES): New rtx code.
9151 * rtl.h (const_vec_series_p_1): Declare.
9152 (const_vec_series_p): New function.
9153 * emit-rtl.h (gen_const_vec_series): Declare.
9154 (gen_vec_series): Likewise.
9155 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9156 (gen_vec_series): Likewise.
9157 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9158 * simplify-rtx.c (simplify_unary_operation): Handle negations
9160 (simplify_binary_operation_series): New function.
9161 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9162 (test_vector_ops_series): New function.
9163 (test_vector_ops): Call it.
9164 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9165 gen_const_vec_series.
9166 (altivec_lvsr): Likewise.
9167 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9169 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9170 Alan Hayward <alan.hayward@arm.com>
9171 David Sherwood <david.sherwood@arm.com>
9173 * doc/rtl.texi (const): Update description of address constants.
9174 Say that vector constants are allowed too.
9175 * common.md (E, F): Use CONSTANT_P instead of checking for
9177 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9178 checking for CONST_VECTOR.
9179 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9181 * expr.c (expand_expr_real_2): Check for vector modes instead
9182 of checking for CONST_VECTOR.
9183 * rtl.h (const_vec_p): New function.
9184 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9185 (unwrap_const_vec_duplicate): Handle them here too.
9187 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9188 David Malcolm <dmalcolm@redhat.com>
9189 Alan Hayward <alan.hayward@arm.com>
9190 David Sherwood <david.sherwood@arm.com>
9192 * rtl.h (vec_duplicate_p): New function.
9193 * selftest-rtl.c (assert_rtx_eq_at): New function.
9194 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9195 (assert_rtx_eq_at): Declare.
9196 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9197 * selftest-run-tests.c (selftest::run_tests): Call it.
9198 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9199 (simplify_unary_operation_1): Recursively handle vector duplicates.
9200 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9202 (simplify_subreg): Handle subregs of vector duplicates.
9203 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9204 (selftest::simplify_rtx_c_tests): New functions.
9206 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9207 Alan Hayward <alan.hayward@arm.com>
9208 David Sherwood <david.sherwood@arm.com>
9210 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9211 (gen_vec_duplicate): Likewise.
9212 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9214 (gen_const_vector): ...here.
9215 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9216 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9217 whose elements are all equal.
9218 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9219 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9220 (simplify_relational_operation): Likewise.
9221 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9223 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9224 (aarch64_expand_vector_init): Likewise.
9225 * config/arm/arm.c (neon_vdup_constant): Likewise.
9226 (neon_expand_vector_init): Likewise.
9227 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9228 (arm_block_set_unaligned_vect): Likewise.
9229 (arm_block_set_aligned_vect): Likewise.
9230 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9231 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9232 (expand_vec_perm_even_odd_pack): Likewise.
9233 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9234 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9235 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9236 gen_const_vec_duplicate.
9237 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9238 * config/mips/mips.c (mips_gen_const_int_vector): Use
9239 gen_const_vec_duplicate.
9240 (mips_expand_vector_init): Use CONST0_RTX.
9241 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9242 (define_split): Use gen_const_vec_duplicate.
9243 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9244 (define_split): Use gen_const_vec_duplicate.
9245 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9246 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9247 * config/spu/spu.c (spu_const): Likewise.
9249 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9250 Alan Hayward <alan.hayward@arm.com>
9251 David Sherwood <david.sherwood@arm.com>
9253 * combine.c (can_change_dest_mode): Reject changes in
9254 REGMODE_NATURAL_SIZE.
9256 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9258 * configure.ac (--enable-libssp): New.
9259 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9260 trying to determine target-specific default. Adjust indentation.
9261 * configure: Regenerated.
9262 * doc/install.texi (Configuration): Expand --disable-libssp
9265 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9267 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9270 2017-10-31 Martin Jambor <mjambor@suse.cz>
9273 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9275 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9277 * auto-profile.c (autofdo_source_profile::read): Use
9278 UNKNOWN_LOCATION rather than 0.
9279 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9280 (warning_at): ...this overload.
9281 (warning_at_rich_loc_n): Rename to...
9282 (warning_n): ...this overload.
9283 (error_at_rich_loc): Rename to...
9284 (error_at): ...this overload.
9285 (pedwarn_at_rich_loc): Rename to...
9286 (pedwarn): ...this overload.
9287 (permerror_at_rich_loc): Rename to...
9288 (permerror): ...this overload.
9289 (inform_at_rich_loc): Rename to...
9290 (inform): ...this overload.
9291 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9292 (diagnostic_n_impl_richloc): Rename to...
9293 (diagnostic_n_impl): ...this rich_location *-based decl.
9294 (inform_at_rich_loc): Rename to...
9295 (inform): ...this, and add an assertion.
9296 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9297 (warning_at_rich_loc): Rename to...
9298 (warning_at): ...this, and add an assertion.
9299 (warning_at_rich_loc_n): Rename to...
9300 (warning_n): ...this, and add an assertion.
9301 (warning_n): Update location_t-based implementation for removal of
9302 location_t-based diagnostic_n_impl.
9303 (pedwarn_at_rich_loc): Rename to...
9304 (pedwarn): ...this, and add an assertion.
9305 (permerror_at_rich_loc): Rename to...
9306 (permerror): ...this, and add an assertion.
9307 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9308 (error_at_rich_loc): Rename to...
9309 (error_at): ...this, and add an assertion.
9310 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9311 (driver::do_spec_on_infiles): Likewise.
9312 * substring-locations.c (format_warning_va): Update for renaming
9313 of inform_at_rich_loc.
9315 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9317 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9318 _Float<N>X built-in functions so that the variant without the
9319 "__builtin_" prefix is only enabled for the GNU C and Objective C
9320 languages when they are in non-strict ANSI/ISO mode.
9321 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9322 * target.def (floatn_builtin_p): Add a target hook to control
9323 whether _Float<N> and _Float<N>X built-in functions without the
9324 "__builtin_" prefix are enabled, and return true for C and
9325 Objective C in the default hook. Include langhooks.h in
9327 * targhooks.h (default_floatn_builtin_p): Likewise.
9328 * targhooks.c (default_floatn_builtin_p): Likewise.
9329 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9330 floatn_builtin_p target hook.
9331 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9333 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9334 Eric Botcazou <ebotcazou@adacore.com>
9336 PR rtl-optimization/81803
9337 * lra-constraints.c (curr_insn_transform): Also reload the whole
9338 register for a strict subreg no wider than a word if this is for
9339 a WORD_REGISTER_OPERATIONS target.
9341 2017-10-31 Jason Merrill <jason@redhat.com>
9343 * gdbinit.in: Skip over inlines from timevar.h.
9345 2017-10-31 Martin Liska <mliska@suse.cz>
9347 * doc/gcov.texi: Document new option.
9348 * gcov.c (print_usage): Likewise print it.
9349 (process_args): Support the argument.
9350 (format_count): New function.
9351 (format_gcov): Use the function.
9353 2017-10-31 Martin Liska <mliska@suse.cz>
9355 * gcov.c (struct name_map): do not use typedef.
9356 Define operator== and operator<.
9357 (name_search): Remove.
9358 (name_sort): Remove.
9359 (main): Do not allocate names.
9360 (process_file): Add vertical space.
9361 (generate_results): Use std::find.
9362 (release_structures): Do not release memory.
9363 (find_source): Use std::find.
9365 2017-10-31 Martin Liska <mliska@suse.cz>
9367 * gcov.c (struct line_info): Remove it's typedef.
9368 (line_info::line_info): Add proper ctor.
9369 (line_info::has_block): Do not use a typedef.
9370 (struct source_info): Do not use typedef.
9371 (circuit): Likewise.
9372 (get_cycles_count): Likewise.
9373 (output_intermediate_file): Iterate via vector iterator.
9374 (add_line_counts): Use std::vector methods.
9375 (accumulate_line_counts): Likewise.
9376 (output_lines): Likewise.
9378 2017-10-31 Martin Liska <mliska@suse.cz>
9380 * gcov.c (struct source_info): Remove typedef.
9381 (source_info::source_info): Add proper ctor.
9382 (accumulate_line_counts): Use struct, not it's typedef.
9383 (output_gcov_file): Likewise.
9384 (output_lines): Likewise.
9385 (main): Do not allocate an array.
9386 (output_intermediate_file): Use size of vector container.
9387 (process_file): Resize the vector.
9388 (generate_results): Do not preallocate, use newly added vector
9390 (release_structures): Do not release sources.
9391 (find_source): Use vector methods.
9392 (add_line_counts): Do not use typedef.
9394 2017-10-31 Martin Liska <mliska@suse.cz>
9396 * doc/gcov.texi: Document that.
9397 * gcov.c (add_line_counts): Mark lines with a non-executed
9399 (output_line_beginning): Handle such lines.
9400 (output_lines): Pass new argument.
9401 (output_intermediate_file): Print it in intermediate format.
9403 2017-10-31 Martin Liska <mliska@suse.cz>
9405 * color-macros.h: New file.
9406 * diagnostic-color.c: Factor out color related to macros to
9408 * doc/gcov.texi: Document -k option.
9409 * gcov.c (INCLUDE_STRING): Include string.h.
9410 (print_usage): Add -k option.
9411 (process_args): Parse it.
9412 (pad_count_string): New function.
9413 (output_line_beginning): Likewise.
9414 (DEFAULT_LINE_START): New macro.
9415 (output_lines): Support color output.
9417 2017-10-31 Martin Liska <mliska@suse.cz>
9419 PR gcov-profile/82633
9420 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9421 their interaction with GCOV infrastructure.
9422 * configure.ac: Add -fkeep-{inline,static}-functions to
9424 * configure: Regenerate.
9426 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9429 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9431 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9434 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9435 into a register if it does not fit into an immediate offset field.
9437 2017-10-31 Olivier Hainque <hainque@adacore.com>
9439 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9441 2017-10-31 Julia Koval <julia.koval@intel.com>
9443 * config.gcc: Add gfniintrin.h.
9444 * config/i386/gfniintrin.h: New.
9445 * config/i386/i386-builtin-types.def
9446 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9447 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9448 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9449 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9450 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9451 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9452 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9453 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9454 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9455 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9456 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9457 * config/i386/immintrin.h: Include gfniintrin.h.
9458 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9460 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9462 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9464 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9466 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9467 (arm_ashldi3_1bit): Remove pattern.
9468 (ashrdi3): Remove shift by 1 expansion.
9469 (arm_ashrdi3_1bit): Remove pattern.
9470 (lshrdi3): Remove shift by 1 expansion.
9471 (arm_lshrdi3_1bit): Remove pattern.
9472 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9473 cost of ashldi3 by 1.
9474 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9475 (<shift>di3_neon): Likewise.
9477 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9479 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9480 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9481 and (*aarch64_simd_mov<VQ:mode>).
9482 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9483 pattern alternative.
9484 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9485 attributes to match pattern alternative.
9487 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9489 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9490 * config/rs6000/emmintrin.h: New file.
9491 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9493 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9495 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9496 * (movdi_vfp_cortexa8): Remove pattern.
9498 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9500 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9502 (Specific, alpha*-dec-osf5.1): Remove.
9503 (Specific, mips-sgi-irix5): Remove.
9504 (Specific, mips-sgi-irix6): Remove.
9506 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9509 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9510 arguments to clear_bit_region_be.
9512 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9514 * gimplify.c: Include memmodel.h.
9516 2017-10-30 Martin Jambor <mjambor@suse.cz>
9518 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9519 condition whether loop should be executed at all.
9521 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9523 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9524 gimple folding of vec_madd() intrinsics.
9525 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9526 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9527 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9529 2017-10-30 Richard Biener <rguenther@suse.de>
9531 PR tree-optimization/82762
9533 2017-10-23 Richard Biener <rguenther@suse.de>
9535 PR tree-optimization/82129
9537 2017-08-01 Richard Biener <rguenther@suse.de>
9539 PR tree-optimization/81181
9540 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9541 (compute_antic): ... end of iteration here.
9543 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9545 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9546 -std=iso9899:2017 and -std=gnu17.
9547 * doc/standards.texi (C Language): Document C17 support.
9548 * doc/cpp.texi (Overview): Mention -std=c17.
9549 (Standard Predefined Macros): Document C11 and C17 values of
9550 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9551 * doc/extend.texi (Inline): Do not list individual options for
9552 standards newer than C99.
9553 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9555 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9558 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9560 * asan.c (asan_finish_file): Align asan globals array by shadow
9563 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9566 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9567 (struct store_immediate_info): Add bitregion_start and bitregion_end
9569 (store_immediate_info::store_immediate_info): Add brs and bre
9570 arguments and initialize bitregion_{start,end} from those.
9571 (struct merged_store_group): Add bitregion_start, bitregion_end,
9572 align_base and mask fields. Drop unnecessary struct keyword from
9573 struct store_immediate_info. Add do_merge method.
9574 (clear_bit_region_be): Use memset instead of loop storing zeros.
9575 (merged_store_group::do_merge): New method.
9576 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9577 stores as long as the surrounding bitregions have no gaps.
9578 (merged_store_group::merge_overlapping): Use do_merge.
9579 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9580 is byte aligned, rather than requiring that start and width are
9581 byte aligned. Drop unnecessary struct keyword from
9582 struct store_immediate_info. Allocate and populate also mask array.
9583 Make start of the arrays relative to bitregion_start rather than
9584 start and size them according to bitregion_{end,start} difference.
9585 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9586 struct store_immediate_info.
9587 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9588 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9589 struct keyword from struct store_immediate_info.
9590 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9591 between stores as long as the surrounding bitregions have no gaps.
9593 (struct split_store): Add orig non-static data member.
9594 (split_store::split_store): Initialize orig to false.
9595 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9596 if there is exactly a single original stmt. Change stmts argument
9597 to pointer from reference, if NULL, don't push anything to it. Add
9598 first argument, use it to optimize skipping over orig stmts that
9599 are known to be before bitpos already. Simplify.
9600 (split_group): Return unsigned int count how many stores are or
9601 would be needed rather than a bool. Add allow_unaligned argument.
9602 Change split_stores argument from reference to pointer, if NULL,
9603 only do a dry run computing how many stores would be produced.
9604 Rewritten algorithm to use both alignment and misalign if
9605 !allow_unaligned and handle bitfield stores with gaps.
9606 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9607 from bitregion_start instead of start. Compute allow_unaligned
9608 here, if true, do 2 split_group dry runs to compute which one
9609 produces fewer stores and prefer aligned if equal. Punt if
9610 new count is bigger or equal than original before emitting any
9611 statements, rather than during that. Remove no longer needed
9612 new_ssa_names tracking. Replace num_stmts with
9613 split_stores.length (). Use 32-bit stack allocated entries
9614 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9615 if possible. Handle bitfields with gaps.
9616 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9617 Compute bitregion_{start,end} for the stores and construct
9618 store_immediate_info with that. Formatting fixes.
9620 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9623 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9624 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9626 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9628 * gimplify.c: Include tm_p.h.
9630 * common.opt (gcoff): Re-add as ignored option.
9631 (gcoff1, gcoff2, gcoff3): Likewise.
9633 * Makefile.in (OBJS): Delete sdbout.o.
9634 (GTFILES): Delete $(srcdir)/sdbout.c.
9635 * debug.h: Delete sdb_debug_hooks.
9636 * final.c: Delete sdbout.h include.
9637 (final_scan_insn): Delete SDB_DEBUG check.
9638 (rest_of_clean_state): Likewise.
9639 * output.h: Delete sdb_begin_function_line.
9642 * toplev.c: Delete sdbout.h include.
9643 (process_options): Delete SDB_DEBUG check.
9644 * tree-core.h (tree_type_common): Delete pointer field of
9646 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9647 TYPE_SYMTAB_POINTER.
9648 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9649 (TYPE_SYMTAB_IS_POINTER): Delete.
9650 (TYPE_SYMTAB_IS_DIE): Renumber.
9651 * xcoffout.c: Refer to former sdbout.c file.
9652 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9654 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9655 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9656 (Debugging Options): Delete -gcoff.
9657 (-gstabs): Delete SDB reference.
9659 (-gcoff@var{level}): Delete.
9660 * doc/passes.texi (Debugging information output): Delete SDB and
9661 sdbout.c references.
9662 * doc/tm.texi: Regenerate.
9663 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9664 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9666 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9667 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9668 (SDB_DEBUGGING_INFO): Delete.
9669 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9670 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9671 * target.def (output_source_filename): Delete COFF reference.
9673 * common.opt (gcoff): Delete.
9674 (gxcoff+): Update Negative chain.
9675 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9677 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9678 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9679 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9681 (expand_function_start): Change sdb reference to past tense.
9682 (expand_function_end): Change sdb reference to past tense.
9683 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9684 * opts.c (debug_type_names): Delete coff entry.
9685 (common_handle_option): Delete OPT_gcoff case.
9686 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9688 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9689 * config/cris/cris.h: Delete SDB reference in comment.
9690 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9691 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9692 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9693 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9695 (ix86_expand_prologue): Likewise.
9696 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9697 * config/ia64/ia64.h: Likewise.
9698 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9699 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9701 * config/mmix/mmix.h: Likewise.
9702 * config/nds32/nds32.c: Likewise.
9703 * config/stormy/storym16.h: Likewise.
9704 * config/visium/visium.h: Likewise.
9705 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9707 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9709 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9710 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9711 FRAME_POINTER_REGNUM point at high end of local var area.
9713 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9715 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9716 Move comment around. Do not reset best_edge for a copiable
9717 destination if the copy would cause a partition change.
9718 (better_edge_p): Remove redundant check.
9720 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9722 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9724 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9726 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9727 for math functions that have _Float<N> and _Float<N>X variants.
9728 (mathfn_built_in_2): Add support for math functions that have
9729 _Float<N> and _Float<N>X variants.
9730 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9731 (expand_builtin_mathfn_ternary): Add support for fma with
9732 _Float<N> and _Float<N>X variants.
9733 (expand_builtin): Likewise.
9734 (fold_builtin_3): Likewise.
9735 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9736 create math function _Float<N> and _Float<N>X variants as external
9738 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9739 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9740 the __builtin_ prefix and if not strict ansi, without the prefix.
9741 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9742 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9743 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9744 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9745 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9746 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9747 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9748 function signatures for fma _Float<N> and _Float<N>X variants.
9749 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9750 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9751 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9752 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9753 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9754 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9755 * gencfn-macros.c (print_case_cfn): Add support for math functions
9756 that have _Float<N> and _Float<N>X variants.
9757 (print_define_operator_list): Likewise.
9758 (fltfn_suffixes): Likewise.
9760 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9761 for math functions that have _Float<N> and _Float<N>X variants.
9762 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9763 and _Float<N>X variants.
9764 (COPYSIGN): Likewise.
9767 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9768 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9770 (integer_valued_read_call_p): Likewise.
9771 * fold-const-call.c (fold_const_call_ss): Likewise.
9772 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9773 _Float<N> and _Float<N>X variants.
9774 (fold_const_call_ssss): Add support for fma _Float<N> and
9775 _Float<N>X variants.
9776 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9777 support for copysign and fma _Float<N> and _Float<N>X variants.
9778 (backprop::process_builtin_call_use): Likewise.
9779 * tree-call-cdce.c (can_test_argument_range); Add support for
9780 sqrt _Float<N> and _Float<N>X variants.
9781 (edom_only_function): Likewise.
9782 (get_no_error_domain): Likewise.
9783 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9784 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9785 copysign _Float<N> and _Float<N>X variants.
9786 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9787 handled by machine independent code.
9788 (FMAF128): Likewise.
9789 * doc/cpp.texi (Common Predefined Macros): Document defining
9790 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9791 fma _Float<N> and _Float<N>X variants.
9793 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9796 * config/i386/i386-modes.def (CCFPU): Remove definition.
9797 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9798 (ix86_cc_modes_compatible): Ditto.
9799 (ix86_expand_carry_flag_compare): Ditto.
9800 (ix86_expand_int_movcc): Ditto.
9801 (ix86_expand_int_addcc): Ditto.
9802 (ix86_reverse_condition): Ditto.
9803 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9804 Return true/false for unordered/ordered fp comparisons.
9805 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9806 (ix86_prepare_fp_compare_args): Update for rename.
9807 (ix86_expand_fp_compare): Update for rename. Generate unordered
9808 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9809 (ix86_expand_sse_compare_and_jump): Ditto.
9810 * config/i386/predicates.md (fcmov_comparison_operator):
9811 Remove CCFPU mode handling.
9812 (ix86_comparison_operator): Ditto.
9813 (ix86_carry_flag_operator): Ditto.
9814 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9815 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9816 (*cmpu<mode>_cc_i387): Ditto.
9817 (FPCMP): Remove mode iterator.
9818 (unord): Remove mode attribute.
9819 (unord_subst): New define_subst transformation
9820 (unord): New define_subst attribute.
9822 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9823 (*cmpi<unord>xf_i387): Ditto.
9824 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9825 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9826 using unord_subst transformation.
9827 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9828 (round_saeonly): Also handle CCFP mode.
9829 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9830 Remove UNSPEC_SAHF unspec handling.
9832 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9834 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9836 2017-10-27 Jeff Law <law@redhat.com>
9838 * gimple-ssa-sprintf.c: Include domwalk.h.
9839 (class sprintf_dom_walker): New class, derived from dom_walker.
9840 (sprintf_dom_walker::before_dom_children): New function.
9841 (struct call_info): Moved into sprintf_dom_walker class
9842 (compute_formath_length, handle_gimple_call): Likewise.
9843 (sprintf_length::execute): Call the dominator walker rather
9844 than walking the statements.
9846 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9847 gimple statement locations.
9848 (check_array_bounds): Corresponding changes. Get the statement's
9849 location directly from wi->stmt.
9851 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9854 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9856 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9858 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9859 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9861 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9864 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9865 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9866 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9867 maybe_get_pool_constant.
9868 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9871 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9873 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9874 2.26 caveat. Update gas and gld versions.
9875 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9878 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9880 * cgraph.h (set_malloc_flag): Declare.
9881 * cgraph.c (set_malloc_flag_1): New function.
9882 (set_malloc_flag): Likewise.
9883 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9884 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9886 (read_ipa_call_summary): Add support for reading is_return_callee.
9887 (write_ipa_call_summary): Stream is_return_callee.
9888 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9889 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9890 ipa-prop.h, ipa-fnsummary.h.
9891 (pure_const_names): Change to static.
9892 (malloc_state_e): Define.
9893 (malloc_state_names): Define.
9894 (funct_state_d): Add field malloc_state.
9895 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9896 (check_retval_uses): New function.
9897 (malloc_candidate_p): Likewise.
9898 (analyze_function): Add support for malloc attribute.
9899 (pure_const_write_summary): Stream malloc_state.
9900 (pure_const_read_summary): Add support for reading malloc_state.
9901 (dump_malloc_lattice): New function.
9902 (propagate_malloc): New function.
9903 (warn_function_malloc): New function.
9904 (ipa_pure_const::execute): Call propagate_malloc and
9905 ipa_free_fn_summary.
9906 (pass_local_pure_const::execute): Add support for malloc attribute.
9907 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9908 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9910 2017-10-27 Martin Liska <mliska@suse.cz>
9912 PR gcov-profile/82457
9913 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9914 for fork-like functions to be properly instrumented.
9916 2017-10-27 Richard Biener <rguenther@suse.de>
9919 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9920 info when we redirected EH.
9922 2017-10-26 Michael Collison <michael.collison@arm.com>
9924 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9926 (<optab>_trunchf<GPI:mode>2: New pattern.
9927 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9928 * config/aarch64/iterators.md (wv): New mode attribute.
9929 (vf, VF): New mode attributes.
9930 (vgp, VGP): New mode attributes.
9931 (s): Update attribute with SImode and DImode prefixes.
9933 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9935 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9936 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9937 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9938 (nios2_option_overide): Initialize it. Don't allow R0-relative
9939 addressing with PIC.
9940 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9941 (nios2_symbolic_constant_p): Likewise.
9942 (nios2_legitimate_address_p): Likewise.
9943 (nios2_r0rel_section_name_p): New.
9944 (nios2_symbol_ref_in_r0rel_data_p): New.
9945 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9946 (r0rel_constant_p): New.
9947 (nios2_print_operand_address): Handle r0rel_constant_p.
9948 (nios2_cdx_narrow_form_p): Likewise.
9949 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9950 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9951 (Nios II Options): Document -mr0rel-sec.
9953 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9955 * config/nios2/nios2.c: Include xregex.h.
9956 (nios2_gprel_sec_regex): New.
9957 (nios2_option_overide): Initialize it. Don't allow GP-relative
9958 addressing with PIC.
9959 (nios2_small_section_name_p): Check for regex match.
9960 * config/nios2/nios2.opt (mgprel-sec=): New option.
9961 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9962 (Nios II Options): Document -mgprel-sec.
9964 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9966 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9968 2017-10-26 Tom de Vries <tom@codesourcery.com>
9970 PR tree-optimization/82707
9971 * gimple.c (gimple_copy): Fix unsharing of
9972 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9974 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9976 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9977 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9978 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9979 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9980 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9981 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9982 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9983 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9984 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9985 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9986 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9987 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9988 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9989 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9990 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9991 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9992 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9994 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9996 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9998 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9999 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10000 warning to rs6000.c. Remove the Undocumented flag, since it has
10002 (-mabi=ibmlongdouble): Likewise.
10003 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10004 already set, set the default format for long double.
10005 (rs6000_debug_reg_global): Print whether long double is IBM or
10007 (rs6000_option_override_internal): Rework setting long double
10008 format. Only warn if the user is changing the long double default
10009 and they did not use -Wno-psabi.
10010 * doc/invoke.texi (PowerPC options): Update the documentation for
10011 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10013 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10014 Alan Hayward <alan.hayward@arm.com>
10015 David Sherwood <david.sherwood@arm.com>
10017 * rtl.h (wider_subreg_mode): New function.
10018 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10019 rather than an unsigned int *.
10020 * ira-color.c (regno_max_ref_width): Replace with...
10021 (regno_max_ref_mode): ...this new variable.
10022 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10023 Use wider_subreg_mode.
10024 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10025 rather than an unsigned int *.
10026 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10027 (process_alt_operands): Likewise.
10028 (invariant_p): Likewise.
10029 * lra-spills.c (assign_mem_slot): Likewise.
10030 (add_pseudo_to_slot): Likewise.
10031 * lra.c (collect_non_operand_hard_regs): Likewise.
10032 (add_regs_to_insn_regno_info): Likewise.
10033 * reload1.c (regno_max_ref_width): Replace with...
10034 (regno_max_ref_mode): ...this new variable.
10035 (reload): Update accordingly. Update call to
10036 ira_sort_regnos_for_alter_reg.
10037 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10038 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10039 (scan_paradoxical_subregs): Likewise.
10041 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10043 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10044 (aarch64_frame): Add emit_frame_chain boolean.
10045 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10046 Move eh_return case to aarch64_layout_frame.
10047 (aarch64_layout_frame): Initialize emit_frame_chain.
10048 (aarch64_expand_prologue): Use emit_frame_chain.
10050 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10052 * config/aarch64/aarch64.c (aarch64_layout_frame):
10053 Ensure LR is always stored at the bottom of the callee-saves.
10054 Remove rarely used frame layout which saves callee-saves at top of
10055 frame, so the store of LR can be used as a valid probe in all cases.
10057 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10059 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10060 Improve unaligned TImode/TFmode base/offset split.
10062 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10063 Alan Hayward <alan.hayward@arm.com>
10064 David Sherwood <david.sherwood@arm.com>
10066 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10067 * combine.c (find_single_use_1): Likewise.
10068 (expand_field_assignment): Likewise.
10069 (move_deaths): Likewise.
10070 * lra-constraints.c (simplify_operand_subreg): Likewise.
10071 (curr_insn_transform): Likewise.
10072 * lra.c (collect_non_operand_hard_regs): Likewise.
10073 (add_regs_to_insn_regno_info): Likewise.
10074 * rtlanal.c (reg_referenced_p): Likewise.
10075 (covers_regno_no_parallel_p): Likewise.
10077 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10079 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10080 Don't print any bits outside the precision of the value.
10081 * wide-int.cc (test_printing): Add some new tests.
10083 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10085 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10086 supports -xbrace_comment option.
10087 * configure: Regenerate.
10088 * config.in: Regenerate.
10089 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10090 (ASM_CPU_SPEC): Use it.
10092 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10094 * target.def (static_rtx_alignment): New hook.
10095 * targhooks.h (default_static_rtx_alignment): Declare.
10096 * targhooks.c (default_static_rtx_alignment): New function.
10097 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10098 * doc/tm.texi: Regenerate.
10099 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10100 instead of targetm.constant_alignment. Remove call to
10101 set_mem_attributes.
10102 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10103 (cris_preferred_mininum_alignment): New function, split out from...
10104 (cris_constant_alignment): ...here.
10105 (cris_static_rtx_alignment): New function.
10106 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10108 (ix86_constant_alignment): ...here.
10109 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10110 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10111 (mmix_static_rtx_alignment): New function.
10112 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10113 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10115 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10118 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10119 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10121 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10123 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10124 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10125 mark as a sign-extending load.
10126 (local_pic_load_u): Define.
10128 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10130 PR middle-end/82062
10131 * fold-const.c (operand_equal_for_comparison_p): Also return true
10132 if ARG0 is a simple variant of ARG1 with narrower precision.
10133 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10135 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10137 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10139 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10140 scatter_static, scatter_per_elt.
10141 * x86-tune-costs.h: Add new cost entries.
10143 2017-10-25 Richard Biener <rguenther@suse.de>
10145 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10146 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10147 class pass_fre): Move to ...
10148 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10149 class pass_fre): ... here and adjust for statistics.
10151 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10154 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10155 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10156 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10157 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10160 2017-10-25 Richard Biener <rguenther@suse.de>
10162 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10163 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10164 eliminate_push_avail, eliminate_insert): Move inside...
10165 (class eliminate_dom_walker): ... this class in preparation
10167 (fini_eliminate): Remove by merging with ...
10168 (eliminate): ... this function. Adjust for class changes.
10169 (pass_pre::execute): Remove fini_eliminate call.
10170 (pass_fre::execute): Likewise.
10172 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10175 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10176 (VPERMI2, VPERMI2I): New mode iterators.
10177 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10178 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10180 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10181 mode iterator. Remove 3 old define_insn patterns.
10182 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10183 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10184 VPERMI2 mode iterator, remove the other two expanders.
10185 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10186 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10187 instructions, remove the other two patterns.
10188 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10189 mode iterator, remove the other two patterns.
10190 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10191 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10192 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10193 and adjust argument order accordingly.
10194 (ix86_expand_vec_perm): Adjust caller.
10195 (expand_vec_perm_1): Likewise.
10196 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10197 (expand_vec_perm_vpermt2_vpshub2): ... this.
10198 (ix86_expand_vec_perm_const_1): Adjust caller.
10199 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10202 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10203 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10204 (vec_shl_<mode>): Remove unused expander.
10205 (avx512bw_<shift_insn><mode>3): New define_insn.
10206 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10207 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10209 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10212 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10215 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10217 PR rtl-optimization/82396
10218 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10219 (autopref_multipass_init): Simplify initialization.
10220 (autopref_rank_data): Simplify sort order.
10221 * gcc/sched-int.h (autopref_multipass_data_): Remove
10222 multi_mem_insn_p, min_offset and max_offset.
10224 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10226 PR middle-end/60580
10227 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10228 Check special value of flag_omit_frame_pointer.
10229 (aarch64_can_eliminate): Likewise.
10230 (aarch64_override_options_after_change_1): Simplify handling of
10231 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10233 2017-10-24 Richard Biener <rguenther@suse.de>
10235 PR tree-optimization/82697
10236 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10237 zero for conditional load and unconditional store.
10239 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10241 * doc/install.texi: Document bootstrap-cet.
10243 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10246 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10247 ENDBR instruction at function entrance if function is only
10250 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10253 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10254 patterns to better describe from which operation the CF is computed.
10255 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10256 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10257 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10258 is 0, use _0 suffixed expanders instead of emitting a comparison
10261 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10263 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10264 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10266 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10268 PR middle-end/82569
10269 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10270 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10271 * loop-iv.c (iv_get_reaching_def): Likewise.
10272 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10273 variable is promoted and the partition contains undefined values.
10275 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10277 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10279 (nios2_address_cost): Define.
10280 (nios2_legitimize_address): Recognize (exp + constant) directly.
10281 (TARGET_ADDRESS_COST): Define.
10283 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10285 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10286 (nios2_symbolic_memory_operand_p): Declare.
10287 (nios2_split_large_constant): Declare.
10288 (nios2_split_symbolic_memory_operand): Declare.
10289 * config/nios2/nios2.c: Adjust includes.
10290 (nios2_symbolic_constant_allowed): New.
10291 (nios2_symbolic_constant_p): New.
10292 (nios2_plus_symbolic_constant_p): New.
10293 (nios2_valid_addr_expr_p): Recognize addresses involving
10294 symbolic constants.
10295 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10296 (nios2_symbolic_memory_operand_p): New.
10297 (nios2_large_constant_p): New.
10298 (nios2_split_large_constant): New.
10299 (nios2_split_plus_large_constant): New.
10300 (nios2_split_symbolic_memory_operand): New.
10301 (nios2_legitimize_address): Code refactoring. Handle addresses
10302 involving symbolic constants.
10303 (nios2_emit_move_sequence): Likewise.
10304 (nios2_print_operand): Improve error output.
10305 (nios2_print_operand_address): Handle LO_SUM.
10306 (nios2_cdx_narrow_form_p): Likewise.
10307 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10308 operands involving symbolic constants.
10309 (movhi_internal, movsi_internal): Likewise.
10310 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10311 (extendhisi2, extendqi<mode>2): Likewise.
10313 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10315 * tree-pass.h (PROP_rtl_split_insns): Define.
10316 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10318 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10320 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10322 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10325 * target.def (const_not_ok_for_debug_p): Default to
10326 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10327 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10328 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10329 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10330 which targetm.const_not_ok_for_debug_p returned true.
10331 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10333 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10335 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10336 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10337 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10339 (ix86_const_not_ok_for_debug_p): New function.
10340 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10341 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10343 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10346 * system.h: Conditionally include "unique-ptr.h" if
10347 INCLUDE_UNIQUE_PTR is defined.
10348 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10349 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10351 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10353 * config/rl78/rl78.md: New define_expand "subdi3".
10355 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10358 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10359 DF_REF_INSN if DF_REF_INSN_INFO is false.
10361 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10363 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10364 xmm_move instead of sse_move.
10365 (sse_store_index): New function.
10366 (ix86_register_move_cost): Be more sensible about mismatch stall;
10367 model AVX moves correctly; make difference between sse->integer and
10369 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10370 moves; make difference between SSE and AVX.
10371 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10372 and zmm_move. Increase size of sse load and store tables;
10373 add unaligned load and store tables; add ssemmx_to_integer.
10374 * x86-tune-costs.h: Update all entries according to real
10375 move latencies from Agner Fog's manual and chip documentation.
10377 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10380 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10381 * config/i386/constraints.md (Wf): New constraint.
10382 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10383 (cmp<dwi>_doubleword): Removed.
10384 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10385 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10386 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10387 expand with cmp<dwi>_doubleword. For LTU and GEU use
10388 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10390 * common.opt (gcolumn-info): Enable by default.
10391 * doc/invoke.texi (gcolumn-info): Document new default.
10393 2017-10-23 Richard Biener <rguenther@suse.de>
10395 PR tree-optimization/82672
10396 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10397 Fold the stmt if we propagated into it.
10399 2017-10-23 Richard Biener <rguenther@suse.de>
10401 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10402 (bitmap_remove_expr_from_set): ... this. All callers call this
10403 for non-constant values.
10404 (bitmap_set_subtract): Rename to...
10405 (bitmap_set_subtract_expressions): ... this. Adjust and
10407 (bitmap_set_contains_value): Remove superfluous check.
10408 (bitmap_set_replace_value): Inline into single caller ...
10409 (bitmap_value_replace_in_set): ... here and simplify.
10410 (dependent_clean): Merge into ...
10411 (clean): ... this using an overload. Adjust.
10412 (prune_clobbered_mems): Adjust.
10413 (compute_antic_aux): Likewise.
10414 (compute_partial_antic_aux): Likewise.
10416 2017-10-23 Richard Biener <rguenther@suse.de>
10418 PR tree-optimization/82129
10420 2017-08-01 Richard Biener <rguenther@suse.de>
10422 PR tree-optimization/81181
10423 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10424 (compute_antic): ... end of iteration here.
10426 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10428 * target.def (starting_frame_offset): New hook.
10429 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10430 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10431 * doc/tm.texi.in: Regenerate.
10432 * hooks.h (hook_hwi_void_0): Declare.
10433 * hooks.c (hook_hwi_void_0): New function.
10434 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10435 STARTING_FRAME_OFFSET.
10436 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10437 * reload1.c (reload): Likewise.
10438 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10439 instead of STARTING_FRAME_OFFSET.
10440 * function.c (try_fit_stack_local): Likewise.
10441 (assign_stack_local_1): Likewise
10442 (instantiate_virtual_regs): Likewise.
10443 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10444 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10445 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10446 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10447 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10448 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10449 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10450 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10451 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10452 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10453 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10454 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10455 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10456 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10457 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10458 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10459 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10460 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10461 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10462 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10463 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10464 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10465 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10466 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10467 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10468 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10469 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10470 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10471 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10472 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10473 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10474 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10475 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10476 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10477 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10478 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10479 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10480 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10481 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10482 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10483 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10484 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10485 return a HOST_WIDE_INT.
10486 (avr_builtin_setjmp_frame_value): Use it instead of
10487 STARTING_FRAME_OFFSET.
10488 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10489 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10490 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10492 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10493 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10494 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10495 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10496 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10497 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10498 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10499 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10500 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10501 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10502 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10503 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10505 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10506 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10507 * config/mips/mips.c (mips_compute_frame_info): Refer to
10508 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10509 (mips_starting_frame_offset): New function.
10510 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10511 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10512 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10513 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10514 and return a HOST_WIDE_INT.
10515 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10516 (mmix_initial_elimination_offset): Refer to
10517 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10518 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10519 * config/pa/pa.c (pa_starting_frame_offset): New function.
10520 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10521 (pa_expand_prologue): Likewise.
10522 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10523 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10524 !FRAME_GROWS_DOWNWARD handling to...
10525 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10526 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10527 !FRAME_GROWS_DOWNWARD handling to...
10528 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10529 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10530 !FRAME_GROWS_DOWNWARD handling to...
10531 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10532 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10534 (rs6000_starting_frame_offset): New function.
10535 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10536 !FRAME_GROWS_DOWNWARD handling to...
10537 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10538 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10539 !FRAME_GROWS_DOWNWARD handling to...
10540 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10541 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10542 !FRAME_GROWS_DOWNWARD handling to...
10543 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10544 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10545 (rs6000_starting_frame_offset): New function.
10546 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10547 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10548 * config/vax/vax.c (vax_starting_frame_offset): New function.
10549 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10550 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10551 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10552 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10553 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10554 * system.h (STARTING_FRAME_OFFSET): Poison.
10556 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10558 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10559 SCALAR_TYPE_MODE instead of TYPE_MODE.
10561 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10562 Alan Hayward <alan.hayward@arm.com>
10563 David Sherwood <david.sherwood@arm.com>
10565 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10567 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10568 Alan Hayward <alan.hayward@arm.com>
10569 David Sherwood <david.sherwood@arm.com>
10571 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10573 2017-10-23 Richard Biener <rguenther@suse.de>
10575 PR tree-optimization/82129
10576 * tree-ssa-pre.c (bitmap_set_and): Remove.
10577 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10578 canonicalizing expressions in the set to those with lowest
10579 ID rather than taking that from the first edge.
10581 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10583 * combine.c (rtx_equal_for_field_assignment_p): Use
10584 byte_lowpart_offset.
10586 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10587 Alan Hayward <alan.hayward@arm.com>
10588 David Sherwood <david.sherwood@arm.com>
10590 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10591 to a SUBREG_PROMOTED_VAR.
10593 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10594 Alan Hayward <alan.hayward@arm.com>
10595 David Sherwood <david.sherwood@arm.com>
10597 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10598 (expand_debug_source_expr): Likewise.
10599 * combine.c (combine_simplify_rtx): Likewise.
10600 * cse.c (fold_rtx): Likewise.
10601 * optabs.c (expand_float): Likewise.
10602 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10603 (simplify_binary_operation_1): Likewise.
10605 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10606 Alan Hayward <alan.hayward@arm.com>
10607 David Sherwood <david.sherwood@arm.com>
10609 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10610 (record_promoted_value): Likewise.
10611 * expr.c (expand_expr_real_2): Likewise.
10612 * ree.c (update_reg_equal_equiv_notes): Likewise.
10613 (combine_set_extension): Likewise.
10614 * rtlanal.c (low_bitmask_len): Likewise.
10615 * simplify-rtx.c (neg_const_int): Likewise.
10616 (simplify_binary_operation_1): Likewise.
10618 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10619 Alan Hayward <alan.hayward@arm.com>
10620 David Sherwood <david.sherwood@arm.com>
10622 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10623 * regcprop.c (maybe_mode_change): Likewise.
10624 * reload1.c (alter_reg): Likewise.
10626 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10628 * inchash.h (inchash::hash::add_wide_int): New function.
10629 * lto-streamer-out.c (hash_tree): Use it.
10631 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10633 * inchash.h (inchash::hash::add_wide_int): Rename to...
10634 (inchash::hash::add_hwi): ...this.
10635 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10636 (polymorphic_call_target_hasher::hash): Likewise.
10637 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10638 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10639 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10640 * lto-streamer-out.c (hash_tree): Likewise.
10641 * optc-save-gen.awk: Likewise.
10642 * tree.c (add_expr): Likewise.
10644 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10647 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10648 for ordered inequality comparisons even with TARGET_IEEE_FP.
10650 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10653 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10654 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10655 Expand with cmp<dwi>_doubleword.
10657 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10659 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10660 List CET intrinsics.
10661 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10662 specific to -fcf-protection option.
10664 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10666 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10667 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10668 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10669 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10670 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10671 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10672 (extra_objs): Add cet.o for Linux/x86 targets.
10673 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10674 * config/i386/cet.c: New file.
10675 * config/i386/cetintrin.h: Likewise.
10676 * config/i386/t-cet: Likewise.
10677 * config/i386/cpuid.h (bit_SHSTK): New.
10678 (bit_IBT): Likewise.
10679 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10680 pass IBT and SHSTK bits.
10681 * config/i386/i386-builtin-types.def
10682 (VOID_FTYPE_UNSIGNED_PVOID): New.
10683 (VOID_FTYPE_UINT64_PVOID): Likewise.
10684 * config/i386/i386-builtin.def: Add CET intrinsics.
10685 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10686 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10687 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10688 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10690 * config/i386/i386.c (rest_of_insert_endbranch): New.
10691 (pass_data_insert_endbranch): Likewise.
10692 (pass_insert_endbranch): Likewise.
10693 (make_pass_insert_endbranch): Likewise.
10694 (ix86_notrack_prefixed_insn_p): Likewise.
10695 (ix86_target_string): Add -mibt, -mshstk flags.
10696 (ix86_option_override_internal): Add flag_cf_protection
10698 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10699 (ix86_print_operand): Add 'notrack' prefix output.
10700 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10701 (ix86_expand_builtin): Expand CET intrinsics.
10702 (x86_output_mi_thunk): Add 'endbranch' instruction.
10703 * config/i386/i386.h (TARGET_IBT): New.
10704 (TARGET_IBT_P): Likewise.
10705 (TARGET_SHSTK): Likewise.
10706 (TARGET_SHSTK_P): Likewise.
10707 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10708 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10709 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10710 (builtin_setjmp_setup): New pattern.
10711 (builtin_longjmp): Likewise.
10712 (rdssp<mode>): Likewise.
10713 (incssp<mode>): Likewise.
10714 (saveprevssp): Likewise.
10715 (rstorssp): Likewise.
10716 (wrss<mode>): Likewise.
10717 (wruss<mode>): Likewise.
10718 (setssbsy): Likewise.
10719 (clrssbsy): Likewise.
10720 (nop_endbr): Likewise.
10721 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10723 * config/i386/immintrin.h: Include <cetintrin.h>.
10724 * config/i386/linux-common.h
10725 (file_end_indicate_exec_stack_and_cet): New prototype.
10726 (TARGET_ASM_FILE_END): New.
10728 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10730 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10731 latencies instead of having separate table; make difference between
10732 integer and float costs.
10733 * i386.h (processor_costs): Remove scalar_stmt_cost,
10734 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10735 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10737 * x86-tune-costs.h: Remove entries which has been removed in
10738 procesor_costs from all tables; make cond_taken_branch_cost
10739 and cond_not_taken_branch_cost COST_N_INSNS based.
10741 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10743 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10745 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10747 * config/i386/i386.md (isa): Remove fma_avx512f.
10748 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10749 <avx512>_fmadd_<mode>_mask3<round_name>,
10750 <avx512>_fmsub_<mode>_mask<round_name>,
10751 <avx512>_fmsub_<mode>_mask3<round_name>,
10752 <avx512>_fnmadd_<mode>_mask<round_name>,
10753 <avx512>_fnmadd_<mode>_mask3<round_name>,
10754 <avx512>_fnmsub_<mode>_mask<round_name>,
10755 <avx512>_fnmsub_<mode>_mask3<round_name>,
10756 <avx512>_fmaddsub_<mode>_mask<round_name>,
10757 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10758 <avx512>_fmsubadd_<mode>_mask<round_name>,
10759 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10760 (*vec_widen_umult_even_v16si<mask_name>,
10761 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10762 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10764 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10766 * extend.texi: Add 'nocf_check' documentation.
10767 * gimple.texi: Add second parameter to
10768 gimple_build_call_from_tree.
10769 * invoke.texi: Add -fcf-protection documentation.
10770 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10772 2017-10-20 Richard Biener <rguenther@suse.de>
10774 PR tree-optimization/82473
10775 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10776 the largest input type.
10778 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10780 * c-attribs.c (handle_nocf_check_attribute): New function.
10781 (c_common_attribute_table): Add 'nocf_check' handling.
10782 * gimple-parser.c: Add second argument NULL to
10783 gimple_build_call_from_tree.
10784 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10785 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10787 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10788 * common.opt: Add fcf-protection flag.
10789 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10790 * flag-types.h: Add enum cf_protection_level.
10791 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10792 Add 'nocf_check' attribute propagation to gimple call.
10793 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10794 (gimple_build_call_from_tree): Update prototype.
10795 (gimple_call_nocf_check_p): New function.
10796 (gimple_call_set_nocf_check): Likewise.
10797 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10798 * ipa-icf.c: Add nocf_check attribute in statement hash.
10799 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10800 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10801 * toplev.c (process_options): Add flag_cf_protection handling.
10803 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10805 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10807 2017-10-20 Richard Biener <rguenther@suse.de>
10809 PR tree-optimization/82603
10810 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10811 remove false predicated stores.
10813 2017-10-20 Richard Biener <rguenther@suse.de>
10815 * graphite-isl-ast-to-gimple.c
10816 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10817 Remove return value and simplify, dump copied stmt after lhs
10819 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10820 Reduce dump verbosity.
10821 (gsi_insert_earliest): Likewise.
10822 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10823 * graphite.c (print_global_statistics): Adjust dumping.
10824 (print_graphite_scop_statistics): Likewise.
10825 (print_graphite_statistics): Do not dump loops here.
10826 (graphite_transform_loops): But here.
10828 2017-10-20 Nicolas Roche <roche@adacore.com>
10830 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10831 * configure: Regenerate.
10833 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10836 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10837 functions when optimizing replace GIMPLE_RETURN stmts with
10838 calls to __builtin_unreachable ().
10841 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10842 for -fsanitize=thread link of executables.
10843 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10844 link of executables.
10847 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10848 New mode iterators.
10849 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10850 define_insns for logical vector shifts to use VI248_AVX512BW
10851 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10852 condition, useless isa and prefix attributes. Change the first
10853 2 of these define_insns to ...
10854 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10855 define_insn for avx512vl.
10856 (<shift_insn><mode>3): ... and this, new define_insn without
10857 masking for non-avx512vl.
10860 * config/i386/sse.md (*andnot<mode>3,
10861 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10862 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10863 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10864 not applied use empty suffix even for TARGET_AVX512VL.
10865 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10866 is applied, supply evex,evex or evex,evex,evex instead of just
10869 2017-10-20 Julia Koval <julia.koval@intel.com>
10871 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10872 (OPTION_MASK_ISA_GFNI_UNSET): New.
10873 (ix86_handle_option): Handle OPT_mgfni.
10874 * config/i386/cpuid.h (bit_GFNI): New.
10875 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10876 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10877 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10878 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10879 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10880 * config/i386/i386.opt: Add mgfni.
10882 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10884 * config/msp430/msp430.c (msp430_option_override): Disable
10885 -fdelete-null-pointer-checks.
10886 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10888 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10890 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10891 of x87 and SSE instructions.
10893 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10895 * asan.c (create_cond_insert_point): Do not update edge count.
10896 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10897 (afdo_propagate_circuit): Likewise.
10898 (afdo_calculate_branch_prob): Likewise.
10899 (afdo_annotate_cfg): Likewise.
10900 * basic-block.h (struct edge_def): Remove count.
10901 (edge_def::count): New accessor.
10902 * bb-reorder.c (rotate_loop): Update.
10903 (find_traces_1_round): Update.
10904 (connect_traces): Update.
10905 (sanitize_hot_paths): Update.
10906 * cfg.c (unchecked_make_edge): Update.
10907 (make_single_succ_edge): Update.
10908 (check_bb_profile): Update.
10909 (dump_edge_info): Update.
10910 (update_bb_profile_for_threading): Update.
10911 (scale_bbs_frequencies_int): Update.
10912 (scale_bbs_frequencies_gcov_type): Update.
10913 (scale_bbs_frequencies_profile_count): Update.
10914 (scale_bbs_frequencies): Update.
10915 * cfganal.c (connect_infinite_loops_to_exit): Update.
10916 * cfgbuild.c (compute_outgoing_frequencies): Update.
10917 (find_many_sub_basic_blocks): Update.
10918 * cfgcleanup.c (try_forward_edges): Update.
10919 (try_crossjump_to_edge): Update
10920 * cfgexpand.c (expand_gimple_cond): Update
10921 (expand_gimple_tailcall): Update
10922 (construct_exit_block): Update
10923 * cfghooks.c (verify_flow_info): Update
10924 (redirect_edge_succ_nodup): Update
10925 (split_edge): Update
10926 (make_forwarder_block): Update
10927 (duplicate_block): Update
10928 (account_profile_record): Update
10929 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10930 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10931 * cfgloopmanip.c (scale_loop_profile): Update.
10933 (lv_adjust_loop_entry_edge): Update.
10934 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10935 (force_nonfallthru_and_redirect): Update.
10936 (purge_dead_edges): Update.
10937 (rtl_flow_call_edges_add): Update.
10938 * cgraphunit.c (init_lowered_empty_function): Update.
10939 (cgraph_node::expand_thunk): Update.
10940 * gimple-pretty-print.c (dump_probability): Update.
10941 (dump_edge_probability): Update.
10942 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10943 * haifa-sched.c (sched_create_recovery_edges): Update.
10944 * hsa-gen.c (convert_switch_statements): Update.
10945 * ifcvt.c (dead_or_predicable): Update.
10946 * ipa-inline-transform.c (inline_transform): Update.
10947 * ipa-split.c (split_function): Update.
10948 * ipa-utils.c (ipa_merge_profiles): Update.
10949 * loop-doloop.c (add_test): Update.
10950 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10951 * lto-streamer-in.c (input_cfg): Update.
10952 (input_function): Update.
10953 * lto-streamer-out.c (output_cfg): Update.
10954 * modulo-sched.c (sms_schedule): Update.
10955 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10956 * predict.c (maybe_hot_edge_p): Update.
10957 (unlikely_executed_edge_p): Update.
10958 (probably_never_executed_edge_p): Update.
10959 (dump_prediction): Update.
10960 (drop_profile): Update.
10961 (propagate_unlikely_bbs_forward): Update.
10962 (determine_unlikely_bbs): Update.
10963 (force_edge_cold): Update.
10964 * profile.c (compute_branch_probabilities): Update.
10965 * reg-stack.c (better_edge): Update.
10966 * shrink-wrap.c (handle_simple_exit): Update.
10967 * tracer.c (better_p): Update.
10968 * trans-mem.c (expand_transaction): Update.
10969 (split_bb_make_tm_edge): Update.
10970 * tree-call-cdce.c: Update.
10971 * tree-cfg.c (gimple_find_sub_bbs): Update.
10972 (gimple_split_edge): Update.
10973 (gimple_duplicate_sese_region): Update.
10974 (gimple_duplicate_sese_tail): Update.
10975 (gimple_flow_call_edges_add): Update.
10976 (insert_cond_bb): Update.
10977 (execute_fixup_cfg): Update.
10978 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10979 * tree-complex.c (expand_complex_div_wide): Update.
10980 * tree-eh.c (lower_resx): Update.
10981 (unsplit_eh): Update.
10982 (cleanup_empty_eh_move_lp): Update.
10983 * tree-inline.c (copy_edges_for_bb): Update.
10984 (freqs_to_counts): Update.
10985 (copy_cfg_body): Update.
10986 * tree-ssa-dce.c (remove_dead_stmt): Update.
10987 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10988 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10989 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10990 (unloop_loops): Update.
10991 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10992 * tree-ssa-loop-split.c (connect_loops): Update.
10993 (split_loop): Update.
10994 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10995 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10996 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10997 * tree-ssa-reassoc.c (branch_fixup): Update.
10998 * tree-ssa-tail-merge.c (replace_block_by): Update.
10999 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11000 (compute_path_counts): Update.
11001 (update_profile): Update.
11002 (recompute_probabilities): Update.
11003 (update_joiner_offpath_counts): Update.
11004 (estimated_freqs_path): Update.
11005 (freqs_to_counts_path): Update.
11006 (clear_counts_path): Update.
11007 (ssa_fix_duplicate_block_edges): Update.
11008 (duplicate_thread_path): Update.
11009 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11010 (case_bit_test_cmp): Update.
11011 (collect_switch_conv_info): Update.
11012 (gen_inbound_check): Update.
11013 (do_jump_if_equal): Update.
11014 (emit_cmp_and_jump_insns): Update.
11015 * tree-tailcall.c (decrease_profile): Update.
11016 (eliminate_tail_call): Update.
11017 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11018 (vect_do_peeling): Update.
11019 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11020 * ubsan.c (ubsan_expand_null_ifn): Update.
11021 (ubsan_expand_ptr_ifn): Update.
11022 * value-prof.c (gimple_divmod_fixed_value): Update.
11023 (gimple_mod_pow2): Update.
11024 (gimple_mod_subtract): Update.
11025 (gimple_ic): Update.
11026 (gimple_stringop_fixed_value): Update.
11028 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11031 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11033 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11035 PR rtl-optimization/82395
11036 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11037 based on non_spilled_static_chain_regno_p.
11039 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11041 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11042 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11043 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11045 2017-10-19 Martin Sebor <msebor@redhat.com>
11047 PR tree-optimization/82596
11048 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11050 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11052 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11053 (get_mem_refs_of_builtin_call): Likewise.
11054 * builtins.c (expand_builtin_apply): Adjust call to
11055 allocate_dynamic_stack_space.
11056 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11057 the third argument to allocate_dynamic_stack_space, otherwise -1.
11058 (expand_builtin): Deal with all alloca variants.
11059 (is_inexpensive_builtin): Likewise.
11060 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11061 * calls.c (special_function_p): Deal with all alloca variants.
11062 (initialize_argument_information): Adjust call to
11063 allocate_dynamic_stack_space.
11064 (expand_call): Likewise.
11065 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11066 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11067 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11068 use it for the stack usage computation.
11069 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11070 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11071 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11072 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11073 (in_loop_p): Remove first argument and useless check.
11074 (pass_walloca::execute): Remove useless test and adjust call to above.
11075 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11076 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11077 (gimplify_call_expr): Deal with all alloca variants.
11078 * hsa-gen.c (gen_hsa_alloca): Likewise.
11079 (gen_hsa_insns_for_call): Likewise.
11080 * ipa-pure-const.c (special_builtin_state): Likewise.
11081 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11082 * tree-object-size.c (alloc_object_size): Likewise.
11083 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11084 (call_may_clobber_ref_p_1): Likewise.
11085 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11086 (ccp_fold_stmt): Likewise.
11087 (optimize_stack_restore): Likewise.
11088 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11089 (mark_all_reaching_defs_necessary_1): Likewise.
11090 (propagate_necessity): Likewise.
11091 (eliminate_unnecessary_stmts): Likewise.
11092 * tree.c (build_common_builtin_nodes): Build
11093 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11094 (build_alloca_call_expr): New function.
11095 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11096 (CASE_BUILT_IN_ALLOCA): Likewise.
11097 (build_alloca_call_expr): Declare.
11098 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11100 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11103 * dwarf2out.c (new_die_raw): New static inline function.
11104 (new_die): Use it to create the DIE.
11105 (add_AT_external_die_ref): Likewise.
11106 (clone_die): Likewise.
11107 (clone_as_declaration): Likewise.
11108 (dwarf2out_vms_debug_main_pointer): Likewise.
11109 (base_type_die): Likewise. Remove early return for corner cases.
11110 Do not call add_pubtype on the DIE here.
11111 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11112 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11113 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11114 native order exists for base types, attach the DIE manually and call
11115 add_pubtype on it. Do not equate a reverse order DIE to the type.
11117 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11119 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11120 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11122 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11124 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11126 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11127 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11128 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11129 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11130 * opts.c (sanitizer_opts): Add builtin.
11131 * ubsan.c (instrument_builtin): New function.
11132 (pass_ubsan::execute): Call it.
11133 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11134 * doc/invoke.texi: Document -fsanitize=builtin.
11136 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11137 builtins, store max (log2 (align), 0) into uchar field instead of
11138 align into uptr field.
11139 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11140 store uchar 0 field instead of uptr 0 field.
11141 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11142 instead of passing one address of struct with 2 locations pass
11143 two addresses of structs with 1 location each.
11144 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11145 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11146 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11147 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11148 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11149 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11150 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11151 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11153 2017-10-19 Martin Liska <mliska@suse.cz>
11156 * file-find.c (remove_prefix): Remove.
11157 * file-find.h (remove_prefix): Likewise.
11158 * gcc-ar.c: Remove smartness of lookup.
11160 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11162 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11163 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11164 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11166 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11169 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11170 (setcc + and to xor + setcc): New peephole2.
11172 2017-10-19 Tom de Vries <tom@codesourcery.com>
11174 * doc/sourcebuild.texi (Test Directives, Variants of
11175 dg-require-support): Add dg-require-stack-size.
11177 2017-10-19 Martin Liska <mliska@suse.cz>
11180 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11181 that have a large alignment.
11182 (gimplify_target_expr): Likewise.
11184 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11186 PR rtl-optimization/82602
11187 * ira.c (rtx_moveable_p): Return false for volatile asm.
11189 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11192 * config/i386/i386-modes.def (CCGZ): New CC mode.
11193 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11194 * config/i386/predicates.md (ix86_comparison_operator):
11196 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11197 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11198 with double-word subtraction.
11199 (put_condition_code): Handle CCGZmode.
11201 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11203 * wide-int.cc (debug (const wide_int &)): New.
11204 (debug (const wide_int *)): New.
11205 (debug (const widest_int &)): New.
11206 (debug (const widest_int *)): New.
11208 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11210 PR middle-end/82556
11211 * lra-constraints.c (curr_insn_transform): Use non-input operand
11212 instead of output one for matched reload.
11214 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11216 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11217 (tree-ssa-loop-ivopts.h): New header file.
11218 (struct builtin_info): New fields.
11219 (classify_builtin_1): Compute and record base and offset parts for
11220 memset builtin partition by calling strip_offset.
11221 (offset_cmp, fuse_memset_builtins): New functions.
11222 (finalize_partitions): Fuse adjacent memset partitions by calling
11224 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11225 Expose the interface.
11226 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11228 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11230 PR tree-optimization/82574
11231 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11232 that data reference must be executed exactly once per iteration
11233 against the outermost loop in nest.
11234 (classify_partition): Update call to above function.
11236 2017-10-18 Richard Biener <rguenther@suse.de>
11238 PR tree-optimization/82591
11239 * graphite.c (graphite_transform_loops): Move code gen message
11241 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11242 Here. Handle scop_to_isl_ast failing.
11243 (scop_to_isl_ast): Limit the number of ISL operations.
11245 2017-10-18 Richard Biener <rguenther@suse.de>
11247 * graphite-isl-ast-to-gimple.c
11248 (translate_isl_ast_to_gimple::set_rename): Simplify.
11249 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11250 (graphite_copy_stmts_from_block): ... here.
11251 (copy_bb_and_scalar_dependences): Simplify.
11252 (add_parameters_to_ivs_params): Canonicalize.
11253 (generate_entry_out_of_ssa_copies): Simplify.
11254 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11255 by passing in ISL dimension.
11256 (parameter_index_in_region_1): Rename to ...
11257 (parameter_index_in_region): ... this.
11258 (extract_affine): Adjust assert, pass down parameter index.
11259 (add_param_constraints): Use range-info when available.
11260 (build_scop_context): Adjust.
11261 * sese.c (new_sese_info): Adjust.
11262 (free_sese_info): Likewise.
11263 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11264 Remove unused typedefs.
11265 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11267 2017-10-18 Martin Liska <mliska@suse.cz>
11269 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11271 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11273 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11274 (s390_sched_init): Do not reset s390_sched_state if we entered the
11275 current basic block via a fallthru edge and all others are unlikely.
11277 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11279 * config/s390/s390.c (NUM_SIDES): New variable.
11280 (LONGRUNNING_THRESHOLD): New variable.
11281 (LATENCY_FACTOR): New variable.
11282 (s390_sched_score): Decrease score for long-running instructions on
11284 (s390_sched_variable_issue): Perform bookkeeping for long-running
11287 2017-10-18 Richard Biener <rguenther@suse.de>
11289 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11290 Simplify with removal of the parameter rename map.
11291 (set_rename): Likewise.
11292 (should_copy_to_new_region): Likewise.
11293 (graphite_copy_stmts_from_block): Likewise.
11294 (copy_bb_and_scalar_dependences): Remove initialization of
11295 unused copied_bb_map.
11296 (copy_def): Remove.
11297 (copy_internal_parameters): Likewise.
11298 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11299 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11300 Use INTEGRAL_TYPE_P.
11301 (parameter_index_in_region_1): Rename to ...
11302 (assign_parameter_index_in_region): ... this. Assert we have
11303 a parameter we handle.
11304 (scan_tree_for_params): Adjust.
11305 * sese.h (parameter_rename_map_t): Remove.
11306 (struct sese_info_t): Remove unused parameter_rename_map and
11307 copied_bb_map members.
11308 * sese.c (new_sese_info): Adjust.
11309 (free_sese_info): Likewise.
11311 2017-10-18 Martin Liska <mliska@suse.cz>
11314 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11315 on an abnormal edge.
11317 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11319 * doc/invoke.texi (ffunction-sections and fdata-sections):
11322 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11324 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11325 the use statement can throw internally.
11327 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11329 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11330 any RTX present on the RHS of a SET.
11331 * compare-elim.c (try_eliminate_compare): Restore comment.
11333 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11335 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11336 may be also called on tcc_type nodes.
11337 * langhooks.c (lhd_tree_size): Likewise.
11339 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11341 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11342 format_warning_at_substring.
11343 (maybe_warn): Convert source_range * param to a location_t. Pass
11344 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11345 (format_directive): Remove code to extract source_ranges and
11346 source_range * in favor of just a location_t.
11347 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11349 * substring-locations.c (format_warning_va): Convert
11350 source_range * param to a location_t.
11351 (format_warning_at_substring): Likewise.
11352 * substring-locations.h (format_warning_va): Likewise.
11353 (format_warning_at_substring): Likewise.
11355 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11357 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11359 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11360 and scatter/gather ops.
11362 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11363 vec_gather_load and vec_scatter_store.
11364 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11365 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11366 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11367 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11368 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11369 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11371 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11373 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11374 Detect FTST instruction and handle its register pops. Only pop
11375 second operand if can_pop_second_op is true.
11376 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11377 set can_pop_second_op to false in the compare_for_stack_reg call.
11379 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11380 output_fp_compare for stack register operands.
11381 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11382 instructions here. Do not emit stack register pops here. Assert
11383 that FCOMPP pops next to top stack register. Rewrite function.
11385 2017-10-17 Nathan Sidwell <nathan@acm.org>
11387 PR middle-end/82577
11388 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11389 use DECL_ASSEMBLER_NAME_RAW.
11391 PR middle-end/82546
11392 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11395 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11396 Wilco Dijkstra <wilco.dijkstra@arm.com>
11398 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11399 converstion to Pmode from the buf_addr.
11401 2017-10-17 Richard Biener <rguenther@suse.de>
11403 * graphite-dependences.c (scop_get_reads_and_writes): Change
11404 output parameters to references.
11406 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11408 PR 71026/tree-optimization
11409 * fold-const.c (distribute_real_division): Removed.
11410 (fold_binary_loc): Remove calls to distribute_real_divison.
11412 2017-10-17 Richard Biener <rguenther@suse.de>
11414 * graphite-scop-detection.c
11415 (scop_detection::stmt_has_simple_data_refs_p): Always use
11416 the full nest as region.
11417 (try_generate_gimple_bb): Likewise.
11418 * sese.c (scalar_evolution_in_region): Simplify now that
11419 SCEV can handle instantiation in regions.
11420 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11421 in the non-loop part of a function if requested.
11423 2017-10-17 Richard Biener <rguenther@suse.de>
11425 PR tree-optimization/82563
11426 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11428 (graphite_regenerate_ast_isl): Call it.
11429 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11431 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11433 PR tree-optimization/82549
11434 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11435 Formatting fixes. Instead of calling make_bit_field_ref with negative
11438 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11440 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11441 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11442 _mm_maskz_reduce_ss): New.
11443 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11444 __builtin_ia32_reducess_mask): Ditto..
11445 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11446 * config/i386/sse.md (reduces<mode>): Renamed to ...
11447 (reduces<mode><mask_scalar_name>): ... this.
11448 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11450 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11451 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11453 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11455 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11456 * selftest-run-tests.c (selftest::run_tests): Call
11457 selftest::unique_ptr_tests_cc_tests.
11458 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11459 * unique-ptr-tests.cc: New file.
11461 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11464 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11466 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11467 (make_hard_regno_born, make_hard_regno_dead): Update
11468 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11470 2017-10-16 Jeff Law <law@redhat.com>
11472 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11474 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11476 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11478 2017-10-16 Olivier Hainque <hainque@adacore.com>
11480 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11481 with_cpu if we were configured for an e500v2 target cpu name.
11483 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11485 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11486 * doc/invoke.texi: Document +nodsp as a valid extension for
11489 2017-10-16 Martin Liska <mliska@suse.cz>
11491 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11492 (test_set_range): Likewise.
11493 (test_range_functions): Rename to ...
11494 (test_bit_in_range): ... this.
11495 (sbitmap_c_tests): Add new test.
11497 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11499 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11501 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11502 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11504 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11506 * config/aarch64/aarch64-builtins.c
11507 (aarch64_types_quadopu_lane_qualifiers): New.
11508 (TYPES_QUADOPU_LANE): New.
11509 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11510 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11511 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11512 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11513 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11514 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11515 (Vdottype, DOTPROD): New.
11516 (sur): Add SDOT and UDOT.
11518 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11520 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11521 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11522 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11523 Add TARGET_DOTPROD.
11524 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11525 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11526 Enable TARGET_DOTPROD.
11527 (cortex-a75.cortex-a55): Likewise.
11528 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11530 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11532 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11533 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11535 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11537 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11538 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11539 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11540 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11541 * config/arm/types.md (neon_dot, neon_dot_q): New.
11542 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11544 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11546 * config/arm/arm.h (TARGET_DOTPROD): New.
11547 * config/arm/arm.c (arm_arch_dotprod): New.
11548 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11549 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11550 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11551 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11552 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11553 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11554 * doc/invoke.texi (armv8.2-a): Document dotprod
11556 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11558 * i386.c (ix86_vec_cost): New function.
11559 (ix86_rtx_costs): Handle vector operations better.
11560 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11561 * x86-tune-costs.h: Add new costs to all tables.
11563 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11565 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11567 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11568 divsd, sqrtss and sqrtsd
11569 * x86-tune-costs.h: Add new entries to all costs.
11570 (znver1_cost): Fix to match real instruction latencies.
11572 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11573 Michael Collison <michael.collison@arm.com>
11575 * compare-elim.c: Include emit-rtl.h.
11576 (can_merge_compare_into_arith): New function.
11577 (try_validate_parallel): Likewise.
11578 (try_merge_compare): Likewise.
11579 (try_eliminate_compare): Call the above when no previous clobber
11581 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11584 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11586 PR middle-end/62263
11587 PR middle-end/82498
11588 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11589 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11591 PR middle-end/62263
11592 PR middle-end/82498
11593 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11594 to be any operand_equal_p operands. For & (B - 1) require
11595 B to be power of 2. Recognize
11596 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11598 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11601 * optabs.c (expand_memory_blockage): Fix call of
11602 targetm.have_memory_blockage.
11604 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11607 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11608 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11609 objects to extra_objs instead of overwriting it.
11611 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11613 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11614 Use any_fp_register_operand as operand[3] predicate. Simplify
11615 equality test for operands[2] and operands[4] memory location.
11616 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11617 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11618 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11619 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11620 any_fp_register_operand as operand[1] predicate. Simplify
11621 equality test for operands[0] and operands[3] memory location.
11622 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11623 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11624 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11626 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11628 * target-insns.def: Add memory_blockage.
11629 * optabs.c (expand_memory_blockage): New function.
11630 (expand_asm_memory_barrier): Rename ...
11631 (expand_asm_memory_blockage): ... to this.
11632 (expand_mem_thread_fence): Call expand_memory_blockage
11633 instead of expand_asm_memory_barrier.
11634 (expand_mem_singnal_fence): Ditto.
11635 (expand_atomic_load): Ditto.
11636 (expand_atomic_store): Ditto.
11637 * doc/md.texi (Standard Pattern Names For Generation):
11638 Document memory_blockage instruction pattern.
11640 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11642 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11643 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11644 * config/rl78/rl78.md: New define_expand "adddi3".
11646 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11648 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11651 2017-10-13 Jeff Law <law@redhat.com>
11653 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11655 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11658 * internal-fn.c (expand_mul_overflow): If both operands have
11659 the same highpart of -1 or 0 and the topmost bit of lowpart
11660 is different, overflow is if res <= 0 rather than res < 0.
11662 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11664 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11665 TARGET_P9_VECTOR code for unaligned_load case.
11667 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11669 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11671 2017-10-13 Nathan Sidwell <nathan@acm.org>
11673 * tree-core.h (tree_contains_struct): Make bool.
11674 * tree.c (tree_contains_struct): Likewise.
11675 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11676 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11677 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11678 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11679 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11681 2017-10-13 Richard Biener <rguenther@suse.de>
11683 * graphite-isl-ast-to-gimple.c
11684 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11685 parameters and dominance check.
11686 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11687 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11688 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11689 Do not update SSA form here or do intermediate IL verification.
11690 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11691 (graphite_initialize): Remove check on the number of loops in
11692 the function and inline into graphite_transform_loops.
11693 (graphite_finalize): Inline into graphite_transform_loops.
11694 (graphite_transform_loops): Perform SSA update and IL verification
11696 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11698 2017-10-13 Richard Biener <rguenther@suse.de>
11700 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11701 graphite_expression_type_precision): Avoid global constructor
11703 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11704 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11705 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11706 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11707 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11708 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11710 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11713 * config/i386/i386.h (ix86_red_zone_size): New.
11714 * config/i386/i386.md (push peephole2s): Replace
11715 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11717 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11718 Alan Hayward <alan.hayward@arm.com>
11719 David Sherwood <david.sherwood@arm.com>
11721 * combine.c (can_change_dest_mode): Reject changes in
11722 REGMODE_NATURAL_SIZE.
11724 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11725 Alan Hayward <alan.hayward@arm.com>
11726 David Sherwood <david.sherwood@arm.com>
11728 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11729 (expand_debug_source_expr): Likewise.
11730 * combine.c (combine_simplify_rtx): Likewise.
11731 * cse.c (fold_rtx): Likewise.
11732 * fwprop.c (canonicalize_address): Likewise.
11733 * targhooks.c (default_shift_truncation_mask): Likewise.
11735 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11736 Alan Hayward <alan.hayward@arm.com>
11737 David Sherwood <david.sherwood@arm.com>
11739 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11740 (widened_mode): Likewise.
11741 (expand_unop): Likewise.
11742 * ree.c (transform_ifelse): Likewise.
11743 (merge_def_and_ext): Likewise.
11744 (combine_reaching_defs): Likewise.
11745 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11747 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11748 Alan Hayward <alan.hayward@arm.com>
11749 David Sherwood <david.sherwood@arm.com>
11751 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11752 * combine.c (gen_lowpart_for_combine): Likewise.
11753 * dwarf2out.c (rtl_for_decl_location): Likewise.
11754 * final.c (alter_subreg): Likewise.
11755 * rtlhooks.c (gen_lowpart_general): Likewise.
11756 (gen_lowpart_if_possible): Likewise.
11758 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11759 Alan Hayward <alan.hayward@arm.com>
11760 David Sherwood <david.sherwood@arm.com>
11762 * calls.c (expand_call): Use subreg_lowpart_offset.
11763 * cse.c (cse_insn): Likewise.
11764 * regcprop.c (copy_value): Likewise.
11765 (copyprop_hardreg_forward_1): Likewise.
11767 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11770 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11771 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11772 =Q constraints to +Q and into insn condition add check
11773 that operands[0] and operands[1] are equal.
11774 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11775 =Q constraints to +Q and into insn condition add check
11776 that operands[0] is equal to either operands[1] or operands[2].
11779 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11780 instead of handling MINUS_EXPR twice (once for each argument),
11781 canonicalize operand order and handle just once, use rtype where
11782 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11785 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11786 any values of __C while still being pattern recognizable as a simple
11787 rotate instruction.
11789 2017-10-13 Richard Biener <rguenther@suse.de>
11791 PR tree-optimization/82451
11793 2017-10-02 Richard Biener <rguenther@suse.de>
11795 PR tree-optimization/82355
11796 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11797 a mapping for the enclosing loop but avoid generating one for
11798 the loop tree root.
11799 (copy_bb_and_scalar_dependences): Remove premature codegen
11800 error on PHIs in blocks duplicated into multiple places.
11801 * graphite-scop-detection.c
11802 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11803 in the region use it as loop and nest to analyze the DR in.
11804 (try_generate_gimple_bb): Likewise.
11805 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11806 (add_loop_constraints): For blocks in a loop not in the region
11807 create a dimension with a single iteration.
11808 * sese.h (gbb_loop_at_index): Remove assert.
11810 * cfgloop.c (loop_preheader_edge): For the loop tree root
11811 return the single successor of the entry block.
11812 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11813 Reset the SCEV hashtable and niters.
11814 * graphite-scop-detection.c
11815 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11816 assert that we only have POLYNOMIAL_CHREC that vary in loops
11817 contained in the region.
11818 (scop_detection::graphite_can_represent_expr): Adjust.
11819 (scop_detection::stmt_has_simple_data_refs_p): For loops
11820 not in the region set loop to NULL. The nest is now the
11821 entry edge to the region.
11822 (try_generate_gimple_bb): Likewise.
11823 * sese.c (scalar_evolution_in_region): Adjust for
11824 instantiate_scev change.
11825 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11826 Make nest parameter the edge into the region.
11827 (create_data_ref): Likewise.
11828 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11829 entry edge into a region and adjust instantiate_scev calls.
11830 (create_data_ref): Likewise.
11831 (graphite_find_data_references_in_stmt): Likewise.
11832 (find_data_references_in_stmt): Pass the loop preheader edge
11833 from the nest argument.
11834 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11835 parameter the edge into the region.
11836 (instantiate_parameters): Use the loop preheader edge as entry.
11837 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11839 (get_instantiated_value_entry): Make instantiate_below parameter
11840 the edge into the region.
11841 (instantiate_scev_name): Likewise. Adjust dominance checks,
11842 when we cannot use loop-based instantiation instantiate by
11843 walking use-def chains.
11844 (instantiate_scev_poly): Adjust.
11845 (instantiate_scev_binary): Likewise.
11846 (instantiate_scev_convert): Likewise.
11847 (instantiate_scev_not): Likewise.
11848 (instantiate_array_ref): Remove.
11849 (instantiate_scev_3): Likewise.
11850 (instantiate_scev_2): Likewise.
11851 (instantiate_scev_1): Likewise.
11852 (instantiate_scev_r): Do not blindly handle N-operand trees.
11853 Do not instantiate array-refs. Handle all constants and invariants.
11854 (instantiate_scev): Make instantiate_below parameter
11855 the edge into the region.
11856 (resolve_mixers): Use the loop preheader edge for the region
11857 parameter to instantiate_scev_r.
11858 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11860 2017-10-13 Richard Biener <rguenther@suse.de>
11862 PR tree-optimization/82525
11863 * graphite-isl-ast-to-gimple.c
11864 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11866 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11867 Fail code generation when we cannot represent the isl integer.
11868 (binary_op_to_tree): Elide modulo operations that are no-ops
11869 in the type we code generate. Remove now superfluous code
11872 2017-10-13 Richard Biener <rguenther@suse.de>
11874 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11875 (scop_detection::harmful_loop_in_region): Remove premature
11876 IV type restriction.
11877 (scop_detection::graphite_can_represent_scev): We can handle
11878 pointer IVs just fine.
11880 2017-10-13 Alan Modra <amodra@gmail.com>
11882 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11883 "Clobbers and Scratch Registers". Add paragraph on
11884 alternative to clobbers for scratch registers and OpenBLAS
11887 2017-10-13 Alan Modra <amodra@gmail.com>
11889 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11890 example of a memory input for a string of known length. Move
11891 commentary out of table. Add a number of new examples
11892 covering array memory inputs.
11894 2017-10-12 Martin Liska <mliska@suse.cz>
11896 PR tree-optimization/82493
11897 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11898 (test_range_functions): New function.
11899 (sbitmap_c_tests): Likewise.
11900 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11901 * selftest.h (sbitmap_c_tests): New function.
11903 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11905 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11907 * config/rs6000/amo.h: Fix spacing issue.
11909 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11912 * config/i386/i386.md (*ashl<mode>3_mask_1,
11913 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11914 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11917 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11919 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11920 (profile_probability): Set max_probability
11921 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11923 * profile-count.c (profile_probability::differs_from_p): Do not
11924 rely on max_probaiblity == 10000
11926 2017-10-12 Jeff Law <law@redhat.com>
11928 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11931 2017-10-12 Martin Sebor <msebor@redhat.com>
11935 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11936 (handle_alias_pairs): Call it.
11937 * common.opt (-Wattribute-alias): New option.
11938 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11939 * doc/invoke.texi (-Wattribute-alias): Document.
11941 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11944 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11946 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11948 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11949 (make_hard_regno_born, make_hard_regno_dead): Update
11950 bb_killed_pseudos and bb_gen_pseudos.
11952 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11954 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11956 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11958 * config/alpha/alpha.c (alpha_split_conditional_move):
11959 Use std::swap instead of manually swapping.
11960 (alpha_stdarg_optimize_hook): Ditto.
11961 (alpha_canonicalize_comparison): Ditto.
11963 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11965 * tree-loop-distribution.c (struct builtin_info): New struct.
11966 (struct partition): Refactor fields into struct builtin_info.
11967 (partition_free): Free struct builtin_info.
11968 (build_size_arg_loc, build_addr_arg_loc): Delete.
11969 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11970 information from struct builtin_info.
11971 (find_single_drs): New function refactored from classify_partition.
11972 Also moved builtin validity checks to this function.
11973 (compute_access_range, alloc_builtin): New functions.
11974 (classify_builtin_st, classify_builtin_ldst): New functions.
11975 (classify_partition): Refactor code into functions find_single_drs,
11976 classify_builtin_st and classify_builtin_ldst.
11977 (distribute_loop): Don't do runtime alias check when distributing
11979 (find_seed_stmts_for_distribution): New function.
11980 (pass_loop_distribution::execute): Refactor code finding seed
11981 stmts into above function. Support distribution for the innermost
11982 two-level loop nest. Adjust dump information.
11984 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11986 * tree-loop-distribution.c: Adjust the general comment.
11987 (NUM_PARTITION_THRESHOLD): New macro.
11988 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11989 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11990 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11991 in call to build_partition_graph.
11992 (finalize_partitions): New parameter. Make loop distribution more
11993 conservative by fusing more partitions.
11994 (distribute_loop): Don't do runtime alias check in case of loop nest
11996 (find_seed_stmts_for_distribution): New function.
11997 (prepare_perfect_loop_nest): New function.
11998 (pass_loop_distribution::execute): Refactor code finding seed stmts
11999 and loop nest into above functions. Support loop nest distribution.
12000 Adjust dump information accordingly.
12002 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12004 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12005 and set PTYPE_SEQUENTIAL for merged partition.
12007 2017-10-12 Richard Biener <rguenther@suse.de>
12009 PR tree-optimization/69728
12011 2017-09-19 Richard Biener <rguenther@suse.de>
12013 PR tree-optimization/69728
12014 * graphite-sese-to-poly.c (schedule_error): New global.
12015 (add_loop_schedule): Handle empty domain by failing the
12017 (build_original_schedule): Handle schedule_error.
12019 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12020 domain by returning an unchanged schedule.
12022 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12024 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12025 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12027 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12029 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12032 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12035 * expr.c (store_field): Don't optimize away bitsize == 0 store
12036 from CALL_EXPR with addressable return type.
12038 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12040 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12041 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12042 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12043 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12044 TARGET_ISEL instead of TARGET_ISEL<sel>.
12046 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12048 * config/rs6000/rs6000.c
12049 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12051 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12053 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12055 (reg_or_cint_operand): Fix comment.
12056 (reg_or_zero_operand): New predicate.
12057 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12058 * config/rs6000/rs6000.c (output_isel): Delete.
12059 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12060 instead of reg_or_cint_operand. Output instruction directly (not via
12062 (isel_unsigned_<mode>): Ditto.
12063 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12064 gpc_reg_operand. Add an instruction alternative for this. Output
12065 instruction directly.
12066 (*isel_reversed_unsigned_<mode>): Ditto.
12068 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12070 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12071 (TARGET_CANONICALIZE_COMPARISON): Define.
12073 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12076 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12077 Check whether the dest is REG before adding REG_EQUIV note.
12079 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12082 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12084 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12085 (make_hard_regno_born, make_hard_regno_dead): Update
12086 bb_killed_pseudos and bb_gen_pseudos.
12088 2017-10-11 Nathan Sidwell <nathan@acm.org>
12090 * incpath.h (enum incpath_kind): Name enum, prefix values.
12091 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12092 * incpath.c (heads, tails): Use INC_MAX.
12093 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12094 (merge_include_chains, split_quote_chain,
12095 register_include_chains): Update incpath_kind names.
12096 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12097 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12099 (add_framework_path, darwin_register_objc_includes): Likewise.
12100 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12102 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12104 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12105 Do not use float_operator operator predicate.
12106 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12107 * config/i386/predicates.md (float_operator): Remove predicate.
12109 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12111 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12112 (*jccxf_i387): Ditto.
12113 (*jcc<mode>_i387): Ditto.
12114 (*jccu<mode>_i387): Ditto.
12115 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12116 (*jcc_*_i387 splitters): Remove.
12117 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12118 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12119 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12122 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12124 * profile-count.h (slow_safe_scale_64bit): New function.
12125 (safe_scale_64bit): New inline.
12126 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12127 * profile-count.c: Include wide-int.h
12128 (slow_safe_scale_64bit): New.
12130 2017-10-11 Nathan Sidwell <nathan@acm.org>
12132 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12133 HAS_DECL_ASSEMBLER_NAME_P.
12134 * gimple-expr.c (gimple_decl_printable_name: Check
12135 HAS_DECL_ASSEMBLER_NAME_P too.
12136 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12137 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12138 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12139 * passes.c (rest_of_decl_compilation): Check
12140 HAS_DECL_ASSEMBLER_NAME_P too.
12141 * recog.c (verify_changes): Likewise.
12142 * tree-pretty-print.c (dump_decl_name): Likewise.
12143 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12145 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12146 (SET_DECL_ASSEMBLER_NAME): Use it.
12147 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12148 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12149 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12151 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12153 * config.gcc (i386, x86_64): Add extra objects.
12154 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12155 (ix86_min_insn_size): Declare.
12156 (ix86_issue_rate): Declare.
12157 (ix86_adjust_cost): Declare.
12158 (ia32_multipass_dfa_lookahead): Declare.
12159 (ix86_macro_fusion_p): Declare.
12160 (ix86_macro_fusion_pair_p): Declare.
12161 (ix86_bd_has_dispatch): Declare.
12162 (ix86_bd_do_dispatch): Declare.
12163 (ix86_core2i7_init_hooks): Declare.
12164 (ix86_atom_sched_reorder): Declare.
12165 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12166 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12167 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12168 (rip_relative_addr_p): Rename to ...
12169 (ix86_rip_relative_addr_p): ... this one; export.
12170 (memory_address_length): Update.
12171 (ix86_issue_rate): Move to x86-tune-sched.c.
12172 (ix86_flags_dependent): Move to x86-tune-sched.c.
12173 (ix86_agi_dependent): Move to x86-tune-sched.c.
12174 (exact_dependency_1): Move to x86-tune-sched.c.
12175 (exact_store_load_dependency): Move to x86-tune-sched.c.
12176 (ix86_adjust_cost): Move to x86-tune-sched.c.
12177 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12178 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12179 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12180 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12181 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12182 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12183 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12184 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12185 (min_insn_size): Rename to ...
12186 (ix86_min_insn_size): ... this one; export.
12187 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12188 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12189 (core2i7_first_cycle_multipass_backtrack): Move to
12190 x86-tune-sched-core.c.
12191 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12192 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12193 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12194 (ix86_avoid_jump_mispredicts): Update.
12195 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12196 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12197 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12198 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12199 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12200 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12201 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12202 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12203 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12204 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12205 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12206 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12207 (BIG): Move to ix86-tune-sched-bd.c.
12208 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12209 (enum insn_path): Move to ix86-tune-sched-bd.c.
12210 (get_mem_group): Move to ix86-tune-sched-bd.c.
12211 (is_cmp): Move to ix86-tune-sched-bd.c.
12212 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12213 (is_branch): Move to ix86-tune-sched-bd.c.
12214 (is_prefetch): Move to ix86-tune-sched-bd.c.
12215 (init_window): Move to ix86-tune-sched-bd.c.
12216 (allocate_window): Move to ix86-tune-sched-bd.c.
12217 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12218 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12219 (process_end_window): Move to ix86-tune-sched-bd.c.
12220 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12221 (find_constant): Move to ix86-tune-sched-bd.c.
12222 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12223 (has_immediate): Move to ix86-tune-sched-bd.c.
12224 (get_insn_path): Move to ix86-tune-sched-bd.c.
12225 (get_insn_group): Move to ix86-tune-sched-bd.c.
12226 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12227 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12228 (add_insn_window): Move to ix86-tune-sched-bd.c.
12229 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12230 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12231 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12232 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12233 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12234 (do_dispatch): Move to ix86-tune-sched-bd.c.
12235 (has_dispatch): Move to ix86-tune-sched-bd.c.
12236 * config/i386/t-i386: Add new object files.
12237 * config/i386/x86-tune-costs.h: New file.
12238 * config/i386/x86-tune-sched-atom.c: New file.
12239 * config/i386/x86-tune-sched-bd.c: New file.
12240 * config/i386/x86-tune-sched-core.c: New file.
12241 * config/i386/x86-tune-sched.c: New file.
12243 2017-10-11 Liu Hao <lh_mouse@126.com>
12245 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12246 the generic version below instead.
12247 (should_colorize): Recognize Windows consoles as terminals
12249 * pretty-print.c [__MINGW32__] (write_all): New function.
12250 [__MINGW32__] (find_esc_head): Likewise.
12251 [__MINGW32__] (find_esc_terminator): Likewise.
12252 [__MINGW32__] (eat_esc_sequence): Likewise.
12253 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12255 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12258 2017-10-11 Richard Biener <rguenther@suse.de>
12260 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12261 Properly call analyze_scalar_evolution with the loop of the stmt.
12263 2017-10-11 Richard Biener <rguenther@suse.de>
12265 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12266 * tree-core.h (tree_base): Add chrec_var union member.
12267 * tree.h (CHREC_VAR): Remove.
12268 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12269 * tree-chrec.h (build_polynomial_chrec): Adjust.
12270 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12271 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12273 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12275 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12276 * match.pd: ... here.
12277 ((T) X == (T) Y): Relax condition.
12279 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12281 PR tree-optimization/82472
12282 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12284 (break_alias_scc_partitions): Update postorder number.
12286 2017-10-11 Martin Liska <mliska@suse.cz>
12289 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12291 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12293 2017-10-11 Martin Liska <mliska@suse.cz>
12295 * print-rtl.c (print_insn): Move declaration of idbuf
12296 to same scope as name.
12298 2017-10-11 Martin Liska <mliska@suse.cz>
12303 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12304 volatile arguments.
12306 2017-10-11 Martin Liska <mliska@suse.cz>
12309 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12310 volatile arguments.
12312 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12314 * config.gcc (default_gnu_indirect_function): Default to yes for
12315 arm*-*-linux* with glibc.
12317 2017-10-11 Richard Biener <rguenther@suse.de>
12319 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12320 default-defs and types we do not want to analyze.
12321 (interpret_loop_phi): Replace unreachable code with an assert.
12322 (compute_scalar_evolution_in_loop): Remove and inline ...
12323 (analyze_scalar_evolution_1): ... here, replacing condition with
12324 what makes the intent clearer. Remove handling of cases
12325 get_scalar_evolution now handles.
12327 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12329 PR rtl-optimization/81434
12330 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12331 comment for main loop. In sched_group_found if, also add checks for
12332 pass and min_cost_group.
12334 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12336 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12337 (rs6000_insn_cost): New function.
12338 * config/rs6000/rs6000.md (cost): New attribute.
12340 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12341 H.J. Lu <hongjiu.lu@intel.com>
12345 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12346 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12347 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12348 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12349 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12350 OPTION_MASK_ISA_MMX and some other bit set require both
12351 mmx and the ISAs without the mmx bit.
12352 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12353 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12354 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12355 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12356 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12357 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12358 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12359 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12360 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12361 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12362 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12363 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12364 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12365 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12366 Add OPTION_MASK_ISA_MMX.
12368 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12370 * config.gcc (armv7*-*-freebsd*): New target.
12371 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12373 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12375 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12376 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12379 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12381 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12382 a template parameter.
12383 (WIDE_INT_REF_FOR): Update accordingly.
12384 * tree.h (wi::int_traits <const_tree>): Delete.
12385 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12386 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12387 (wi::tree_to_wide_ref): New typedef.
12388 (wi::to_wide): New function.
12389 * calls.c (get_size_range): Use wi::to_wide when operating on
12390 trees as wide_ints.
12391 * cgraph.c (cgraph_node::create_thunk): Likewise.
12392 * config/i386/i386.c (ix86_data_alignment): Likewise.
12393 (ix86_local_alignment): Likewise.
12394 * dbxout.c (stabstr_O): Likewise.
12395 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12396 * expr.c (const_vector_from_tree): Likewise.
12397 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12398 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12399 (fold_negate_expr_1, int_const_binop_1, const_binop)
12400 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12401 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12402 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12403 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12404 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12405 (fold_not_const, round_up_loc): Likewise.
12406 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12407 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12408 (alloca_call_type): Likewise.
12409 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12410 * godump.c (go_output_typedef): Likewise.
12411 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12412 * internal-fn.c (get_min_precision): Likewise.
12413 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12414 * ipa-polymorphic-call.c
12415 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12416 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12417 (ipa_modify_call_arguments): Likewise.
12418 * match.pd: Likewise.
12419 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12420 * print-tree.c (print_node_brief, print_node): Likewise.
12421 * stmt.c (expand_case): Likewise.
12422 * stor-layout.c (layout_type): Likewise.
12423 * tree-affine.c (tree_to_aff_combination): Likewise.
12424 * tree-cfg.c (group_case_labels_stmt): Likewise.
12425 * tree-data-ref.c (dr_analyze_indices): Likewise.
12426 (prune_runtime_alias_test_list): Likewise.
12427 * tree-dump.c (dequeue_and_dump): Likewise.
12428 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12429 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12430 * tree-pretty-print.c (dump_generic_node): Likewise.
12431 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12432 (simple_iv_with_niters): Likewise.
12433 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12434 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12435 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12436 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12437 (refine_value_range_using_guard, number_of_iterations_ne_max)
12438 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12439 (get_cst_init_from_scev, record_nonwrapping_iv)
12440 (scev_var_range_cant_overflow): Likewise.
12441 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12442 * tree-ssa-pre.c (compute_avail): Likewise.
12443 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12444 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12445 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12446 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12447 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12448 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12449 (dump_case_nodes, try_switch_expansion): Likewise.
12450 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12451 (vect_do_peeling): Likewise.
12452 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12453 * tree-vect-stmts.c (vectorizable_load): Likewise.
12454 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12455 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12456 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12457 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12458 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12459 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12460 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12461 (evrp_dom_walker::before_dom_children): Likewise.
12462 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12463 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12464 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12465 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12466 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12467 (get_range_pos_neg): Likewise.
12468 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12469 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12470 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12471 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12472 * config/avr/avr.c (avr_fold_builtin): Likewise.
12473 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12474 * config/msp430/msp430.c (msp430_attr): Likewise.
12475 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12476 * config/powerpcspe/powerpcspe-c.c
12477 (altivec_resolve_overloaded_builtin): Likewise.
12478 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12479 (rs6000_expand_ternop_builtin): Likewise.
12480 * config/rs6000/rs6000-c.c
12481 (altivec_resolve_overloaded_builtin): Likewise.
12482 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12483 (rs6000_expand_ternop_builtin): Likewise.
12484 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12486 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12488 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12489 when copying loop nest with only one inner loop.
12491 2017-10-10 Richard Biener <rguenther@suse.de>
12493 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12494 blocks if SCEV is active.
12495 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12497 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12498 (scev_initialize): Assert we are not yet initialized.
12500 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12502 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12503 inner loop's exit stmt by making it always exit the loop, otherwise
12504 we would generate an infinite empty loop.
12506 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12508 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12509 renaming variables in new preheader if it's deleted.
12511 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12513 * tree-loop-distribution.c (struct partition): Remove unused field
12514 loops of the structure.
12515 (partition_alloc, partition_free): Ditto.
12516 (build_rdg_partition_for_vertex): Ditto.
12518 2017-10-09 Jeff Law <law@redhat.com>
12520 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12521 return type to match prototype and documentation.
12523 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12525 * config/rs6000/rs6000.c (processor_costs): Move to ...
12526 * config/rs6000/rs6000.h: ... here.
12527 (rs6000_cost): Declare.
12529 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12531 * except.c (setjmp_fn): New global variable.
12532 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12533 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12534 if DONT_USE_BUILTIN_SETJMP is defined.
12536 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12538 * target.def (insn_cost): New hook.
12539 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12540 * doc/tm.texi: Regenerate.
12541 * rtlanal.c (insn_cost): Use the new hook.
12543 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12545 * combine.c (combine_validate_cost): Compute the new insn_cost,
12546 not just pattern_cost.
12547 (try_combine): Adjust comment.
12549 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12551 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12553 * combine.c (uid_insn_cost): Adjust comment.
12554 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12556 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12557 * dse.c (find_shift_sequence): Ditto.
12558 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12559 (bb_valid_for_noce_process_p): Use pattern_cost.
12560 * rtl.h (insn_rtx_cost): Delete.
12561 (pattern_cost): New prototype.
12562 (insn_cost): New prototype.
12563 * rtlanal.c (insn_rtx_cost): Rename to...
12564 (pattern_cost): ... this.
12567 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12569 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12570 (*jcc<mode>_0_r_i387): Ditto.
12571 (*jccxf_r_i387): Ditto.
12572 (*jcc<mode>_r_i387): Ditto.
12573 (*jccu<mode>_r_i387): Ditto.
12574 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12575 (*jcc): Rename from *jcc_1.
12577 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12579 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12580 deferred rescans after the lvx/stvx recombination pre-pass.
12582 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12584 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12585 memory operation instruction support.
12586 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12587 (rs6000-ibm-aix[789]*): Likewise.
12588 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12589 Document new functions.
12591 2017-10-09 Richard Biener <rguenther@suse.de>
12593 PR tree-optimization/82397
12594 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12595 equality only for semantically equal trees.
12597 2017-10-09 Richard Biener <rguenther@suse.de>
12599 PR tree-optimization/82449
12600 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12601 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12602 allow constant addresses.
12603 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12606 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12608 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12611 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12614 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12617 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12620 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12622 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12625 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12626 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12628 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12630 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12631 (WI_BINARY_PREDICATE_RESULT): Likewise.
12632 (wi::binary_traits::operator_result): New type.
12633 (wi::binary_traits::predicate_result): Likewise.
12634 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12635 (generic_wide_int::operator==, generic_wide_int::operator!=)
12636 (generic_wide_int::operator&, generic_wide_int::and_not)
12637 (generic_wide_int::operator|, generic_wide_int::or_not)
12638 (generic_wide_int::operator^, generic_wide_int::operator+
12639 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12641 (operator~, unary operator-, operator==, operator!=, operator&)
12642 (operator|, operator^, operator+, binary operator-, operator*): New
12644 * expr.c (get_inner_reference): Use wi::bit_and_not.
12645 * fold-const.c (fold_binary_loc): Likewise.
12646 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12647 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12648 (bit_value_binop): Likewise.
12649 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12650 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12651 (extract_range_from_binary_expr_1): Likewise.
12652 (masked_increment): Likewise.
12653 (simplify_bit_ops_using_ranges): Likewise.
12655 2017-10-09 Martin Jambor <mjambor@suse.cz>
12658 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12659 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12660 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12662 (hsa_fixup_mov_insn_type): New function.
12663 (hsa_op_with_type::get_in_type): Use it.
12664 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12665 immediates in an assert.
12666 (hsa_op_with_type::extend_int_to_32bit): New method.
12667 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12668 types. Convert to dest type if necessary.
12669 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12670 (reg_for_gimple_ssa): Pass false as min32int to
12671 hsa_type_for_scalar_tree_type.
12672 (gen_hsa_addr): Fixup type when creating addresable temporary.
12673 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12674 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12675 necessary. Call hsa_fixup_mov_insn_type.
12676 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12677 extend operands and convert to dest type if necessary.
12678 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12679 to dest type if necessary.
12680 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12681 if conversion nt necessary and size matches.
12682 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12683 to dest type if necessary.
12684 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12685 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12687 (gen_hsa_clrsb): Likewise.
12688 (gen_hsa_ffs): Likewise.
12689 (gen_hsa_divmod): Extend operands and convert to dest type if
12691 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12693 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12695 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12696 Remove empty default arguments. Use a brace block as output
12698 (conditional return): Ditto.
12700 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12701 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12703 (group_ending_nop): Ditto.
12704 (doloop_end): Ditto.
12705 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12706 (splitters for those): Ditto.
12708 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12710 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12711 a conditional jump (and the compare for it) so that pc_rtx is the
12713 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12714 for the deleted and renamed ctr<mode>_internal[234] patterns.
12715 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12716 Delete second conditional return pattern.
12717 (ctr<mode>_internal2): Delete this second bdnz pattern.
12718 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12719 (ctr<mode>_internal4): Delete this second bdz pattern.
12721 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12723 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12724 (always_initialized_rtx_for_ssa_name_p): New predicate.
12725 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12726 (finish_out_of_ssa): Free new field of SA.
12727 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12728 * tree-ssa-coalesce.c: Include tree-ssa.h.
12729 (get_parm_default_def_partitions): Remove extern keyword.
12730 (get_undefined_value_partitions): New function.
12731 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12732 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12733 uninitialized bits.
12734 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12736 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12738 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12740 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12742 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12743 for targets that preffer 128bit.
12745 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12747 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12749 2017-10-08 Olivier Hainque <hainque@adacore.com>
12751 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12752 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12754 (thumb_set_return_address): Likewise.
12756 2017-10-08 Olivier Hainque <hainque@adacore.com>
12758 * common/config/arm/arm-common.c (arm_except_unwind_info):
12759 Handle DWARF2_UNWIND_INFO.
12761 2017-10-07 Michael Collison <michael.collison@arm.com>
12763 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12766 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12768 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12769 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12770 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12771 defined, force the creation of a new block for a dispatch label.
12773 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12775 * invoke.texi (Wsuggest-attribute=cold): Document.
12776 * common.opt (Wsuggest-attribute=cold): New
12777 * ipa-pure-const.c (warn_function_cold): New function.
12778 * predict.c (compute_function_frequency): Use it.
12779 * predict.h (warn_function_cold): Declare.
12781 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12783 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12786 2017-10-06 Martin Liska <mliska@suse.cz>
12788 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12789 keyword for member functions.
12790 (struct sanopt_tree_couple): New struct.
12791 (struct sanopt_tree_couple_hash): New function.
12792 (struct sanopt_ctx): Add new hash_map.
12793 (has_dominating_ubsan_ptr_check): New function.
12794 (record_ubsan_ptr_check_stmt): Likewise.
12795 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12796 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12797 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12799 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12802 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12803 aarch64_simd_valid_immediate on CONST_VECTORs.
12804 (aarch64_reg_or_bic_imm): Likewise.
12806 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12808 PR rtl-optimization/82396
12809 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12811 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12813 * graphite-dependences.c (scop_get_reads): Move code to...
12814 (scop_get_must_writes): Move code to...
12815 (scop_get_may_writes): Move code to...
12816 (scop_get_reads_and_writes): ... here.
12817 (scop_get_dependences): Call scop_get_reads_and_writes.
12819 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12821 PR tree-optimization/82434
12822 * fold-const.h (can_native_encode_type_p,
12823 can_native_encode_string_p): Remove.
12824 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12825 don't encode anything, just return what would be otherwise returned.
12826 (native_encode_fixed, native_encode_complex, native_encode_vector):
12828 (native_encode_string): Likewise. Inline by hand
12829 can_native_encode_string_p.
12830 (can_native_encode_type_p): Remove.
12831 (can_native_encode_string_p): Remove.
12832 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12833 STRING_CSTs using can_native_encode_string_p, test all
12834 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12835 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12836 argument from native_encode_expr.
12837 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12838 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12841 2017-10-06 Richard Biener <rguenther@suse.de>
12843 PR tree-optimization/82397
12844 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12845 operand_equal_p but rely on data_ref_compare_tree for detecting
12847 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12848 to match up with dr_group_sort_cmp.
12850 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12853 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12855 * config/s390/s390-builtin-types.def: Regenerate.
12857 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12860 * config/s390/s390-builtin-types.def: Regenerate.
12861 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12862 Change flag from B_VXE to B_VX.
12863 (s390_vec_min_dbl): Remove B_VXE flag.
12865 2017-10-06 Richard Biener <rguenther@suse.de>
12867 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12868 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12869 translate_isl_ast_to_gimple::is_valid_rename,
12870 translate_isl_ast_to_gimple::get_rename,
12871 translate_isl_ast_to_gimple::get_def_bb_for_const,
12872 translate_isl_ast_to_gimple::get_new_name,
12873 translate_isl_ast_to_gimple::collect_all_ssa_names,
12874 translate_isl_ast_to_gimple::copy_loop_phi_args,
12875 translate_isl_ast_to_gimple::collect_all_ssa_names,
12876 translate_isl_ast_to_gimple::copy_loop_phi_args,
12877 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12878 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12879 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12880 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12881 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12882 translate_isl_ast_to_gimple::copy_cond_phi_args,
12883 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12884 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12885 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12886 translate_isl_ast_to_gimple::rename_uses,
12887 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12888 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12889 (set_rename_for_each_def): Likewise.
12890 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12891 here. Handle rewriting SCEV analyzable uses here.
12892 (copy_bb_and_scalar_dependences): Generate code for PHI
12894 (graphite_regenerate_ast_isl): Adjust.
12895 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12896 (add_write, add_read): New functions.
12897 (build_cross_bb_scalars_def): Use it and simplify.
12898 (build_cross_bb_scalars_use): Likewise.
12899 (graphite_find_cross_bb_scalar_vars): Inline into...
12900 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12901 simulating out-of-SSA. Compute liveout and add dependencies.
12902 (build_scops): Force an empty entry block.
12903 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12905 (sese_build_liveouts): Declare.
12906 (sese_trivially_empty_bb_p): Likewise.
12907 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12908 compute liveout and debug_liveout.
12909 (sese_bad_liveouts_use): Remove.
12910 (sese_reset_debug_liveouts_bb): Likewise.
12911 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12912 (sese_build_liveouts): Build liveout and debug_liveout and store
12914 (new_sese_info): Adjust.
12915 (free_sese_info): Likewise.
12916 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12917 do not build liveout here.
12918 (move_sese_in_condition): Adjust region entry.
12919 (scev_analyzable_p): Match up with chrec_apply requirements.
12920 (sese_trivially_empty_bb_p): New.
12921 * tree-into-ssa.c (get_reaching_def): Properly support generating
12922 default-defs for incremental rewrite of anonymous names.
12924 2017-10-06 Richard Biener <rguenther@suse.de>
12926 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12927 precision do not perform modulo reduction.
12929 2017-10-06 Richard Biener <rguenther@suse.de>
12931 PR tree-optimization/82436
12932 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12933 conservatively choose the vectorization factor when checking
12934 whether we can perform the required load permutation.
12935 (vect_transform_slp_perm_load): Assert when we may not fail.
12937 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12939 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12940 message for incompatible -msdata=* and -mcall-* options.
12942 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12944 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12945 rate for post-reload scheduling.
12947 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12949 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12951 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12953 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12954 to improve monte carlo in scimark.
12956 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12958 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12959 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12960 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12961 pentium4_cost, nocona_cost): Set reassociation width to 1.
12962 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12963 width to 2 for fp operations and 1 otherwise.
12964 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12966 (atom_cost): Set reassociation width to 2.
12967 (slm_cost, generic_cost): Set fp reassociation width
12968 to 2 and 1 otherwise.
12969 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12970 (core_cost): Set fp reassociation width to 4 and vector to 2.
12971 (ix86_reassociation_width): Rewrite using cost table; special case
12972 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12973 and TARGET_AVX128_OPTIMAL.
12974 * config/i386/i386.h (processor_costs): Add
12975 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12976 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12977 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12978 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12979 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12980 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12982 2017-10-05 Nathan Sidwell <nathan@acm.org>
12984 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12986 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12988 * config/arm/arm.c (arm_test_fpu_data): New.
12989 (arm_run_selftests): Call arm_test_fpu_data.
12991 2017-10-04 Nathan Sidwell <nathan@acm.org>
12993 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12995 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12997 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12999 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13000 check type for aarch64_simd_valid_immediate.
13001 (aarch64_output_simd_mov_immediate): Update prototype.
13002 (aarch64_simd_valid_immediate): Update prototype.
13003 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13004 support for ORR-immediate.
13005 (and<mode>3): modified pattern to add support for BIC-immediate.
13006 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13007 now checks for valid immediate for BIC and ORR based on new enum
13009 (aarch64_output_simd_mov_immediate): Function now used to output
13010 BIC/ORR imm as well based on new enum argument.
13011 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13013 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13014 (aarch64_reg_or_bic_imm): Likewise.
13016 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13018 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13019 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13021 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13025 PR rtl-optimization/82396
13026 * haifa-sched.c (autopref_multipass_init): Simplify
13028 (autopref_rank_data): Simplify sort order.
13029 * sched-int.h (autopref_multipass_data_): Remove
13030 multi_mem_insn_p, min_offset and max_offset.
13032 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13034 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13036 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13038 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13039 vect_doubleint_cvt.
13041 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13043 * doc/sourcebuild.texi: Document vect_long_mult.
13045 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13047 PR tree-optimization/82413
13048 * fold-const.c (build_range_check): Use widest_int when comparing
13049 the maximum ETYPE value with HIGH.
13051 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13053 PR rtl-optimization/82396
13054 * haifa-sched.c (autopref_multipass_init): Simplify
13056 (autopref_rank_data): Simplify sort order.
13057 * sched-int.h (autopref_multipass_data_): Remove
13058 multi_mem_insn_p, min_offset and max_offset.
13060 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13062 PR tree-optimization/82381
13063 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13064 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13067 PR tree-optimization/82374
13068 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13069 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13070 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13071 current_function_decl to the new decl.
13073 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13075 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13076 helper macro for IEEE float128 hardware built-in functions.
13077 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13079 (TRUNCF128_ODD): Likewise.
13080 (ADDF128_ODD): Likewise.
13081 (SUBF128_ODD): Likewise.
13082 (MULF128_ODD): Likewise.
13083 (DIVF128_ODD): Likewise.
13084 (FMAF128_ODD): Likewise.
13085 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13086 UNSPEC_TRUNC_ROUND_TO_ODD.
13087 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13088 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13089 floating point round to odd instructions.
13090 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13091 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13092 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13093 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13094 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13095 (trunc<mode>sf2_hw): Change the truncate with round to odd
13096 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13097 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13098 to odd hardware instructions.
13099 (sub<mode>3_odd): Likewise.
13100 (mul<mode>3_odd): Likewise.
13101 (div<mode>3_odd): Likewise.
13102 (sqrt<mode>2_odd): Likewise.
13103 (fma<mode>4_odd): Likewise.
13104 (fms<mode>4_odd): Likewise.
13105 (nfma<mode>4_odd): Likewise.
13106 (nfms<mode>4_odd): Likewise.
13107 (trunc<mode>df2_odd): Change the truncate with round to odd
13108 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13110 * doc/extend.texi (PowerPC built-in functions): Update documentation
13111 for existing IEEE float128-bit built-in functions. Add built-in
13112 functions that generate the IEEE 128-bit floating point round to
13115 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13117 PR rtl-optimization/77729
13118 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13119 to (X&(C1&~C2))|C2 transformations.
13121 2017-10-03 Martin Jambor <mjambor@suse.cz>
13123 PR tree-optimization/82363
13124 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13125 mismatch, mark lacc written regardless of racc.
13127 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13129 PR tree-optimization/82381
13130 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13131 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13132 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13133 fallthrough into reassoc_stmt_dominates_stmt_p.
13136 * combine.c (combine_instructions): Don't combine in unreachable
13139 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13142 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13143 function to not use the have_cpu variable. Do not set cpu_index,
13144 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13145 or the default cpu.
13146 (rs6000_valid_attribute_p): Remove duplicate initializations of
13147 old_optimize and func_optimize.
13148 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13149 (rs6000_activate_target_options): Make global.
13150 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13153 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13155 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13156 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13157 Set *poffset to 0 and *psize and *pmax_size to -1 if
13158 *poffset + *psize overflows in HOST_WIDE_INT.
13160 PR tree-optimization/82387
13161 PR tree-optimization/82388
13162 PR tree-optimization/82389
13163 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13164 instead of live_bytes non-NULL.
13166 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13169 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13172 2017-10-02 Richard Biener <rguenther@suse.de>
13174 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13175 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13176 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13178 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13180 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13181 requested precision matches the type's.
13182 * calls.c (alloc_max_size): Calculate the new candidate size as
13183 a widest_int and use wi::to_widest when comparing it with the
13184 current candidate size.
13185 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13186 zero rather than integer_zero_node.
13187 * match.pd: Check for a no-op conversion before using wi::add
13188 rather than after. Use tree_to_uhwi when summing small shift
13189 counts into an unsigned int.
13191 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13192 Alan Hayward <alan.hayward@arm.com>
13193 David Sherwood <david.sherwood@arm.com>
13196 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13197 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13198 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13199 POINTER_AND_FP_REGS.
13201 2017-10-02 Richard Biener <rguenther@suse.de>
13203 PR tree-optimization/82355
13204 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13205 a mapping for the enclosing loop but avoid generating one for
13206 the loop tree root.
13207 (copy_bb_and_scalar_dependences): Remove premature codegen
13208 error on PHIs in blocks duplicated into multiple places.
13209 * graphite-scop-detection.c
13210 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13211 in the region use it as loop and nest to analyze the DR in.
13212 (try_generate_gimple_bb): Likewise.
13213 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13214 (add_loop_constraints): For blocks in a loop not in the region
13215 create a dimension with a single iteration.
13216 * sese.h (gbb_loop_at_index): Remove assert.
13218 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13220 * omp-expand.c (adjust_context_scope): New function.
13221 (expand_parallel_call): Call adjust_context_scope.
13223 2017-10-01 Jeff Law <law@redhat.com>
13225 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13226 dom_opt_dom_walker class with direct access to private members.
13227 Add comments. Call test_for_singularity.
13228 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13229 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13230 m_dummy_cond anymore.
13231 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13233 (pass_dominator:execute): Build the dummy_cond here and pass it
13234 to the dom_opt_dom_walker ctor.
13235 (test_for_singularity): New function.
13237 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13238 Maya Rashish <coypu@sdf.org>
13240 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13241 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13242 (alpha*-*-netbsd*) Use nbsd_tm_file.
13243 (arm*-*-netbsdelf*) Likewise.
13244 (i[34567]86-*-netbsdelf*) Likewise.
13245 (x86_64-*-netbsd*) Likewise.
13246 (mips*-*-netbsd*) Likewise.
13247 (powerpc-*-netbsd*) Likewise.
13248 (sh*-*-netbsd*) Likewise.
13249 (sparc-*-netbsdelf*) Likewise.
13250 (sparc64-*-netbsd*) Likewise.
13251 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13253 (vax-*-netbsdelf*) Likewise.
13254 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13255 (UINT_FAST8_TYPE) Likewise.
13256 (INT_FAST16_TYPE) Check CHAR_FAST16.
13257 (UINT_FAST16_TYPE) Likewise.
13259 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13262 * config/i386/i386.md
13263 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13264 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13265 *divmodsi4_zext_2): New define_insn_and_split.
13266 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13267 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13268 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13269 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13270 New define_insn_and_split.
13271 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13272 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13273 operands[1] having DImode when mode is SImode.
13275 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13276 always SImode for DIV and MOD in REG_EQUAL notes.
13278 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13280 PR middle-end/82319
13281 * match.pd: Fix handling of NaNs in pattern.
13283 2017-09-29 Jeff Law <law@redhat.com>
13285 * sbitmap.c (bitmap_bit_in_range_p): New function.
13286 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13287 * tree-ssa-dse.c (live_bytes_read): New function.
13288 (dse_classify_store): Ignore reads of dead bytes.
13290 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13291 typos and whitespace errors.
13292 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13293 * config/s390/s390.c (s390_emit_prologue): Likewise.
13295 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13298 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13299 with a symbol for LRA.
13301 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13303 PR rtl-optimization/82338
13304 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13306 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13308 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13309 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13310 (qsort_chk): Declare.
13311 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13312 (qsort_chk): New function.
13314 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13316 PR tree-optimization/82337
13317 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13318 phi definition if the PHI result appears in an abnormal PHI.
13319 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13320 basis appears in an abnormal PHI.
13322 2017-09-29 Richard Biener <rguenther@suse.de>
13324 * graphite-isl-ast-to-gimple.c
13325 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13326 (binary_op_to_tree): Use it.
13327 (get_rename_from_scev): Likewise.
13328 (copy_loop_phi_nodes): Likewise.
13329 (copy_bb_and_scalar_dependences): Likewise.
13330 (translate_pending_phi_nodes): Likewise.
13332 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13335 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13336 for movabsq $(i32 << shift), r64.
13338 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13340 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13341 index when encoding %esp as %rsp to avoid 0x67 prefix.
13343 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13345 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13346 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13348 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13350 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13351 Extensions with more than 16 double VFP registers.
13352 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13353 to_clear_mask and all code related to it. Replace the remaining
13354 entry by a sbitmap and adapt code accordingly.
13356 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13358 * brig-builtins.def: Change pure attributes to const.
13360 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13362 * config.gcc (default_gnu_indirect_function): Default to yes for
13363 sparc*-*-linux* with glibc.
13365 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13367 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13368 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13369 when creating .init_array and .fini_array sections with priority
13372 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13375 * config/aarch64/aarch64.c
13376 (aarch64_builtin_support_vector_misalignment): Always return false
13377 when misalignment is unknown.
13379 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13381 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13382 this function to return false if the definition used by the swap
13383 instruction is artificial, or if the memory address from which the
13384 constant value is loaded is not represented by a base address held
13385 in a register or if the base address register is a frame or stack
13386 pointer. Additionally, return false if the base address of the
13387 loaded constant is a SYMBOL_REF but is not considered to be a
13389 (replace_swapped_load_constant): New function.
13390 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13391 loaded constant vector with a load of a swapped constant vector.
13393 2017-09-27 Carl Love <cel@us.ibm.com>
13395 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13396 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13397 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13400 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13402 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13403 first, always call autopref_rank_data otherwise.
13405 2017-09-27 Richard Biener <rguenther@suse.de>
13407 * graphite-scop-detection.c (find_scop_parameters): Move
13408 loop bound handling ...
13409 (gather_bbs::before_dom_children): ... here, avoiding the need
13410 to build scop_info->loop_nest.
13411 (record_loop_in_sese): Remove.
13412 * sese.h (sese_info_t::loop_nest): Remove.
13413 * sese.c (new_sese_info): Do not allocate loop_nest.
13414 (free_sese_info): Do not free loop_nest.
13416 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13419 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13420 lhs from calls if the lhs has addressable type.
13422 2017-09-27 Richard Biener <rguenther@suse.de>
13424 * graphite.h (scop::max_alias_set): New member.
13425 * graphite-scop-detection.c: Remove references to non-existing
13426 --param in comments.
13427 (build_alias_sets): Record the maximum alias set used for drs.
13428 (build_scops): Support zero as unlimited for
13429 --param graphite-max-arrays-per-scop.
13430 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13431 and inline into ...
13432 (build_poly_sr_1): ... here. Compute alias set based on the
13433 maximum alias set used for drs rather than
13434 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13436 2017-09-27 Richard Biener <rguenther@suse.de>
13438 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13439 --param loop-block-tile-size=0 to disable tiling.
13441 2017-09-27 Richard Biener <rguenther@suse.de>
13443 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13444 (graphite-max-nb-scop-params): Document special value zero.
13445 * domwalk.h (dom_walker::STOP): New symbolical constant.
13446 (dom_walker::dom_walker): Add optional parameter for bb to
13448 (dom_walker::~dom_walker): Declare.
13449 (dom_walker::before_dom_children): Document STOP return value.
13450 (dom_walker::m_user_bb_to_rpo): New member.
13451 (dom_walker::m_bb_to_rpo): Likewise.
13452 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13453 mapping here if not provided by the user.
13454 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13455 provided by the user.
13456 (dom_walker::STOP): Define.
13457 (dom_walker::walk): Do not compute bb to RPO mapping here.
13458 Support STOP return value from before_dom_children to stop
13460 * graphite-optimize-isl.c (optimize_isl): If the schedule
13461 is the same still generate code if -fgraphite-identity
13462 or -floop-parallelize-all are given.
13463 * graphite-scop-detection.c: Include cfganal.h.
13464 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13466 (gather_bbs::before_dom_children): Return STOP for BBs
13468 (build_scops): Compute bb to RPO mapping and pass it to
13469 the domwalk. Treat --param graphite-max-nb-scop-params=0
13470 as not limiting the number of params.
13471 * graphite.c (graphite_initialize): Remove limit on the
13472 number of basic-blocks in a function.
13473 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13474 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13475 default value of 10.
13477 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13479 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13480 Adjust code to eliminate needing to do the shift right 32-bits
13481 operation after XSCVDPSPN.
13483 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13485 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13486 ((X / Y) != 0 -> X >= Y): Likewise.
13488 2017-09-26 Carl Love <cel@us.ibm.com>
13490 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13491 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13492 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13493 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13494 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13495 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13496 definitions and overloading.
13497 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13498 statement for P9V_BUILTIN_XST_LEN_R.
13499 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13500 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13501 define_expand and define_insn for the instructions and builtins.
13502 * doc/extend.texi: Update the built-in documentation file for the new
13503 built-in functions.
13504 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13505 define_insn for the instructions
13507 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13510 * gcc/config/netbsd-protos.h: New file.
13511 * gcc/config/netbsd.c: New file.
13512 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13513 * gcc/config/t-netbsd: New file.
13514 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13515 (tmake_file) Add t-netbsd.
13516 (extra_objs) Add netbsd.o.
13518 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13522 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13524 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13526 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13527 sign extension from a vector register to a GPR by doing a 32-bit
13528 direct move and then an EXTSW.
13529 (extendsi<mode>2 splitter): Likewise.
13530 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13531 right or vector extract after doing XSCVDPSPN. Use
13532 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13534 (movdi_from_sf_zero_ext): Likewise.
13535 (reload_gpr_from_vsxsf): Likewise.
13536 (p8_mfvsrd_4_disf): Delete, no longer used.
13537 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13538 then needing to move the SFmode to a GPR to use the XSCVDPSP
13539 instruction instead of FRSP and XSCVDPSPN.
13540 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13541 it is adjacent to the other XSCVSPDP insns.
13542 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13543 SFmode to be in traditional Altivec registers.
13544 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13545 (vsx_xscvspdpn): Likewise.
13546 (vsx_xscvspdpn_scalar): Likewise.
13548 2017-09-26 Martin Jambor <mjambor@suse.cz>
13550 * tree-sra.c (compare_access_positions): Put integral types first,
13551 stabilize sorting of integral types, remove conditions putting
13552 non-full-precision integers last.
13553 (sort_and_splice_var_accesses): Disable scalarization if a
13554 non-integert would be represented by a non-full-precision integer.
13556 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13558 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13559 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13560 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13561 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13562 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13563 conditionals inside the function instead of around it. Call
13564 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13565 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13567 2017-09-26 Richard Biener <rguenther@suse.de>
13569 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13571 (scop_detection::build_scop_breadth): ... this. Removed.
13572 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13573 (scop_detection::harmful_stmt_in_bb): Likewise.
13574 (scop_detection::graphite_can_represent_stmt): Likewise.
13575 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13576 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13577 (scop_detection::can_represent_loop_1): ... this. Removed.
13578 (scop_detection::harmful_loop_in_region): Simplify after inlining
13579 the above and remove more quadraticness.
13580 (build_scops): Adjust.
13581 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13584 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13587 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13588 REGNO (base) == SP_REG if base is a REG.
13590 PR middle-end/35691
13591 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13592 if it is different SSA_NAME.
13593 (optimize_range_tests_cmp_bitwise): New function.
13594 (optimize_range_tests): Call it.
13596 2017-09-26 Richard Biener <rguenther@suse.de>
13598 PR tree-optimization/82321
13599 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13600 for the def being inside the loop.
13602 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13604 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13606 * config/s390/s390-builtins.def: Fix constraint on op4.
13608 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13610 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13611 independent expanders.
13612 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13613 ("vec_ordered", "vec_unordered"): New expanders.
13615 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13617 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13620 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13622 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13623 vec_unpacks_lo_v16qi.
13624 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13626 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13628 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13629 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13630 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13632 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13634 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13636 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13638 ("*vec_slb<mode>"): New insn pattern.
13639 ("vec_shr_<mode>"): New expander.
13640 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13641 and force the shift count operand to V16QImode.
13642 ("vec_srb<mode>"): Set shift count mode to V16QI.
13644 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13646 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13647 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13648 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13650 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13653 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13655 2017-09-26 Richard Biener <rguenther@suse.de>
13657 PR tree-optimization/82320
13658 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13661 2017-09-25 Jeff Law <law@redhat.com>
13663 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13664 prototype for new argument.
13665 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13666 mostly extracted from rs6000_emit_allocate_stack.
13667 (rs6000_emit_probe_stack_range_stack_clash): New function.
13668 (rs6000_emit_allocate_stack): Call
13669 rs6000_emit_probe_stack_range_stack_clash as needed.
13670 (rs6000_emit_probe_stack_range): Add additional argument
13671 to call to gen_probe_stack_range{si,di}.
13672 (output_probe_stack_range): New.
13673 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13674 (output_probe_stack_range_stack_clash): New.
13675 (rs6000_emit_prologue): Emit notes into dump file as requested.
13676 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13677 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13678 Add additional operand and pass it to output_probe_stack_range.
13680 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13682 PR tree-optimization/82163
13683 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13684 (checking_verify_loop_closed_ssa): New parameter.
13685 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13686 (check_loop_closed_ssa_stmt): Delete.
13687 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13688 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13689 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13692 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13694 * brig-builtins.def: Treat HSAIL barrier builtins as
13695 setjmp/longjump style functions.
13697 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13699 * target.def (constant_alignment): New hook.
13700 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13701 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13702 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13703 * doc/tm.texi: Regenerate.
13704 * targhooks.h (default_constant_alignment): Declare.
13705 (constant_alignment_word_strings): Likewise.
13706 * targhooks.c (default_constant_alignment): New function.
13707 (constant_alignment_word_strings): Likewise.
13708 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13709 instead of CONSTANT_ALIGNMENT.
13710 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13711 (force_const_mem): Likewise.
13712 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13713 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13714 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13715 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13716 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13718 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13719 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13720 constant_alignment_word_strings.
13721 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13722 (CONSTANT_ALIGNMENT): Likewise.
13723 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13724 (arm_constant_alignment): New function.
13725 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13726 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13727 constant_alignment_word_strings.
13728 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13729 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13730 constant_alignment_word_strings.
13731 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13732 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13733 (cris_constant_alignment): New function.
13734 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13735 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13736 (epiphany_constant_alignment): New function.
13737 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13738 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13739 constant_alignment_word_strings.
13740 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13741 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13742 constant_alignment_word_strings.
13743 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13744 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13745 constant_alignment_word_strings.
13746 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13747 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13748 * config/i386/i386.c (ix86_constant_alignment): Make static.
13749 Use the same interface as the target hook.
13750 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13751 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13752 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13753 constant_alignment_word_strings.
13754 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13755 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13756 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13757 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13758 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13759 constant_alignment_word_strings.
13760 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13761 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13762 constant_alignment_word_strings.
13763 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13764 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13765 constant_alignment_word_strings.
13766 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13767 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13769 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13770 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13771 * config/mips/mips.c (mips_constant_alignment): New function.
13772 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13773 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13774 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13775 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13776 (mmix_constant_alignment): Make static. Use the same interface
13777 as the target hook.
13778 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13779 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13780 constant_alignment_word_strings.
13781 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13782 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13783 constant_alignment_word_strings.
13784 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13785 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13786 constant_alignment_word_strings.
13787 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13788 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13789 (rs6000_constant_alignment): New function.
13790 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13791 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13792 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13793 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13794 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13795 (rs6000_constant_alignment): New function.
13796 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13797 * config/s390/s390.c (s390_constant_alignment): New function.
13798 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13799 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13800 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13801 constant_alignment_word_strings.
13802 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13803 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13804 (sparc_constant_alignment): New function.
13805 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13806 * config/spu/spu.c (spu_constant_alignment): New function.
13807 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13808 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13809 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13810 constant_alignment_word_strings.
13811 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13812 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13813 constant_alignment_word_strings.
13814 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13815 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13816 constant_alignment_word_strings.
13817 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13818 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13819 (visium_constant_alignment): New function.
13820 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13821 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13822 (xtensa_constant_alignment): New function.
13823 * system.h (CONSTANT_ALIGNMENT): Poison.
13825 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13827 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13828 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13829 (rs6000_builtin_valid_without_lhs): New helper function.
13830 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13831 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13833 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13835 * target.h (vec_perm_indices): Use unsigned short rather than
13837 (auto_vec_perm_indices): Likewise.
13838 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13839 Use unsigned int rather than unsigned char.
13840 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13842 2017-09-25 Richard Biener <rguenther@suse.de>
13844 * cfgloop.h (sort_sibling_loops): Declare.
13845 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13846 (sort_sibling_loops): New function sorting the sibling loop list
13848 * graphite.c (graphite_transform_loops): Sort sibling loops.
13850 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13852 * target.def (vec_perm_const_ok): Change sel parameter to
13854 * optabs-query.c (can_vec_perm_p): Update accordingly.
13855 * doc/tm.texi: Regenerate.
13856 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13857 auto_vec_perm_indices and remove separate nelt field.
13858 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13859 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13860 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13861 (aarch64_expand_vec_perm_const): Update accordingly.
13862 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13863 to vec_perm_indices.
13864 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13865 auto_vec_perm_indices and remove separate nelt field.
13866 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13867 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13868 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13870 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13871 to vec_perm_indices.
13872 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13873 sel to vec_perm_indices.
13874 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13875 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13876 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13878 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13881 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13884 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13885 on the FUNCTION_DECL function context if it has a DIE that is a
13888 2017-09-25 Richard Biener <rguenther@suse.de>
13890 PR tree-optimization/82285
13891 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13894 2017-09-25 Tom de Vries <tom@codesourcery.com>
13898 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13901 2017-09-25 Richard Biener <rguenther@suse.de>
13903 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13904 ISL errors other than isl_error_quota happen. Dump if the
13905 schedule is the same.
13906 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13907 errors instead of aborting inside ISL.
13909 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13912 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13913 of libgcc_eh for m64.
13914 * config/i386/darwin64.h: Likewise.
13916 2017-09-25 Richard Biener <rguenther@suse.de>
13918 PR middle-end/82144
13919 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13920 attribute for incomplete types nor twice for complete ones.
13922 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13925 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13926 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13929 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13932 * config/i386/i386.opt (mprefer-avx256): Use
13933 ix86_target_flags variable.
13934 * config/i386/i386.c (ix86_target_string): Move
13935 -mprefer-avx256 to flag2_opts.
13937 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13939 PR middle-end/35691
13940 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13941 and x != -1 | y != -1 into (x & y) != -1.
13943 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13945 * config.gcc: Add new case statement to set
13946 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13947 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13948 s390x-*-linux* case statements. Added aarch64 to the list of
13949 supported architectures.
13951 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13953 PR tree-optimization/82289
13954 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13955 STMT_VINFO_RELEVANT_P.
13957 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13958 Alan Hayward <alan.hayward@arm.com>
13959 David Sherwood <david.sherwood@arm.com>
13961 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13962 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13963 (extract_range_from_binary_expr_1): Don't call
13964 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13966 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13967 Alan Hayward <alan.hayward@arm.com>
13968 David Sherwood <david.sherwood@arm.com>
13970 * target.def (preferred_vector_alignment): New hook.
13971 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13973 * doc/tm.texi: Regenerate.
13974 * targhooks.h (default_preferred_vector_alignment): Declare.
13975 * targhooks.c (default_preferred_vector_alignment): New function.
13976 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13978 (DR_TARGET_ALIGNMENT): New macro.
13979 (aligned_access_p): Update commentary.
13980 (vect_known_alignment_in_bytes): New function.
13981 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13983 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13984 Calculate the misalignment based on the target alignment rather than
13986 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13987 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13988 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13989 the target alignment, rather than masking the element misalignment
13990 with the number of elements in a vector. Also use the target
13991 alignment when calculating the maximum number of peels.
13992 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13993 instead of TYPE_ALIGN_UNIT.
13994 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13995 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13996 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13997 (vect_create_data_ref_ptr): Likewise.
13998 (vect_setup_realignment): Realign by ANDing with
13999 -DR_TARGET_MISALIGNMENT.
14000 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14001 the number of peels based on DR_TARGET_ALIGNMENT.
14002 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14003 with the guaranteed alignment boundary when deciding whether
14005 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14006 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14007 (ensure_base_align): Remove stmt_info parameter. Get the
14008 target base alignment from DR_TARGET_ALIGNMENT.
14009 (vectorizable_store): Update call accordingly. Interpret
14010 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14012 (vectorizable_load): Likewise.
14014 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14015 Alan Hayward <alan.hayward@arm.com>
14016 David Sherwood <david.sherwood@arm.com>
14018 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14019 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14020 (vect_enhance_data_refs_alignment): Likewise.
14022 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14024 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14025 error. Only quit immediately if parsing is complete.
14026 (BEGIN): Initialize fatal_err and parse_done.
14027 (begin fpu, end fpu): Check number of arguments.
14028 (begin arch, end arch): Likewise.
14029 (begin cpu, end cpu): Likewise.
14030 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14031 (optalias): Likewise.
14033 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14035 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14036 * config/arm/arm-isa.h: Delete. Move definitions to ...
14037 * arm-cpus.in: ... here. Use new feature and fgroup values.
14038 * config/arm/arm.c (arm_option_override): Use lower case for feature
14040 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14041 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14042 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14044 (print_isa_bits_for): New function.
14045 (gen_isa): New function.
14046 (gen_comm_data): Use print_isa_bits_for.
14047 (define feature): New keyword.
14048 (define fgroup): New keyword.
14049 * config/arm/t-arm (TM_H): Remove.
14050 (GTM_H): Add arm-isa.h.
14051 (arm-isa.h): Add rule to generate file.
14052 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14053 case for feature bit names.
14055 2017-09-22 Richard Biener <rguenther@suse.de>
14057 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14059 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14060 print of no dependency loops ...
14061 * graphite.c (graphite_transform_loops): ... here.
14062 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14064 (same_close_phi_node, remove_duplicate_close_phi,
14065 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14066 (canonicalize_loop_closed_ssa): ... here and simplify.
14067 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14068 (optimize_isl): Use dump_printf_loc to tell when we stopped
14069 optimizing because of an ISL timeout.
14071 2017-09-22 Richard Biener <rguenther@suse.de>
14073 PR tree-optimization/82291
14074 * tree-if-conv.c (predicate_mem_writes): Make sure to
14075 remove writes in blocks predicated with false.
14077 2017-09-22 Richard Biener <rguenther@suse.de>
14079 * sese.c: Include cfganal.h.
14080 (if_region_set_false_region): Remove.
14081 (create_if_region_on_edge): Likewise.
14082 (move_sese_in_condition): Re-implement without destroying
14085 2017-09-22 Richard Biener <rguenther@suse.de>
14087 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14088 Verify both BBs contain loop PHI nodes before dispatching to
14089 copy_loop_phi_args.
14090 (graphite_regenerate_ast_isl): Do not recompute dominators,
14091 do not verify three times. Restructure for clarity.
14092 * graphite-scop-detection.c (same_close_phi_node,
14093 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14094 defined_in_loop_p, canonicalize_loop_closed_ssa,
14095 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14096 checking and SSA rewrite, move to ...
14097 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14098 (graphite_initialize): Do not pass in ctx, do not reset the
14099 SCEV cache, compute only dominators.
14100 (graphite_transform_loops): Allocate ISL ctx after
14101 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14102 Maintain post-dominators only around build_scops.
14103 * sese.c (if_region_set_false_region): Make static. Free
14104 and recompute dominators.
14105 (move_sese_in_condition): Assert we don't get called with
14106 post-dominators computed.
14107 * sese.h (if_region_set_false_region): Remove.
14109 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14111 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14112 mode attribute for TARGET_AVX512VL.
14114 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14116 * config/i386/i386.opt (mprefer-avx256): New option.
14117 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14119 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14120 for TARGET_PREFER_AVX256.
14121 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14123 2017-09-21 Jeff Law <law@redhat.com>
14125 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14126 Fix dump output if the only stack space is for pushed registers.
14128 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14130 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14133 2017-09-21 Martin Sebor <msebor@redhat.com>
14136 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14137 code (in C++) or code that triggers warnings.
14139 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14141 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14143 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14145 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14146 * sched-rgn.c: Ditto.
14147 * sel-sched-ir.c: Ditto.
14149 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14151 * toplev.h (set_random_seed): Adjust return type.
14152 * toplev.c (init_local_tick): Move eager initialization of random_seed
14153 to get_random_seed. Adjust comment.
14154 (init_random_seed): Inline to get_random_seed, delete.
14155 (get_random_seed): Initialize random_seed lazily.
14156 (set_random_seed): Do not return previous value.
14157 (print_switch_value): Do not call get_random_seed.
14159 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14161 * cgraph.c (delete_function_version): New, broken out from...
14162 (cgraph_node::delete_function_version): ...here. Rename to
14163 cgraph_node::delete_function_version_by_decl. Update all uses.
14164 (cgraph_node::remove): Call delete_function_version.
14166 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14169 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14170 VAR_DECLs to which addressable non-volatile parameters are mapped
14171 and for id->retvar after the return value assignment. Clear
14172 id->retval and id->retbnd after inlining.
14174 2017-09-21 Richard Biener <rguenther@suse.de>
14176 PR tree-optimization/82276
14177 PR tree-optimization/82244
14178 * tree-vrp.c (build_assert_expr_for): Set
14179 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14181 (remove_range_assertions): Revert earlier change.
14183 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14186 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14188 2017-09-21 Richard Biener <rguenther@suse.de>
14190 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14191 Restore valid IL after code generation errors.
14192 * graphite.c (graphite_transform_loops): Diagnose code
14193 generation issues as MSG_MISSED_OPTIMIZATION and continue
14194 with processing SCOPs.
14196 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14197 Alan Hayward <alan.hayward@arm.com>
14198 David Sherwood <david.sherwood@arm.com>
14200 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14201 rather than choosing between plus_constant and gen_rtx_<CODE>.
14202 * expr.c (emit_push_insn): Likewise.
14203 (expand_expr_real_2): Likewise.
14205 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14206 Alan Hayward <alan.hayward@arm.com>
14207 David Sherwood <david.sherwood@arm.com>
14209 * loop-unroll.c (split_iv): Call copy_rtx on the step.
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 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14216 calling tree_to_uhwi.
14218 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14219 Alan Hayward <alan.hayward@arm.com>
14220 David Sherwood <david.sherwood@arm.com>
14222 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14223 INTEGER_CST rather than a negative test for ADDR_EXPR.
14225 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14226 Alan Hayward <alan.hayward@arm.com>
14227 David Sherwood <david.sherwood@arm.com>
14229 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14230 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14232 2017-09-21 Richard Biener <rguenther@suse.de>
14234 PR tree-optimization/71351
14235 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14236 graphite_create_new_loop_guard): Remove, fold remaining parts
14238 (translate_isl_ast_node_for): ... here and simplify.
14240 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14243 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14244 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14245 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14246 alternative always use QI mode, for -Os imov (=R,R) alternative
14247 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14250 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14251 Jeff Law <law@redhat.com>
14253 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14254 (allocate_stack_space): New function, partially extracted from
14255 s390_emit_prologue.
14256 (s390_emit_prologue): Track offset to most recent stack probe.
14257 Code to allocate space moved into allocate_stack_space.
14258 Dump actions when no stack is allocated.
14259 (s390_prologue_plus_offset): New function.
14260 (s390_emit_stack_probe): Likewise.
14262 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14264 * common.opt (Wa, Wl, Wp, g, gz=): Add
14266 (gno-column-info): Remove.
14267 (gcolumn-info): Drop RejectNegative.
14268 (gno-): New prefix.
14269 (gno-record-gcc-switches): Remove.
14270 (grecord-gcc-switches): Drop RejectNegative.
14271 (gno-split-dwarf): Remove.
14272 (gsplit-dwarf): Drop RejectNegative.
14273 (gno-strict-dwarf): Remove.
14274 (gstrict-dwarf): Drop RejectNegative.
14275 * config/darwin.opt (gfull, gused): Add RejectNegative.
14276 * dwarf2out.c (gen_producer_string): Drop
14277 gno-record-gcc-switches handler.
14278 * optc-gen.awk: Add g to prefixes with negative forms.
14279 * opts-common.c (remapping_prefix_p): New.
14280 (find_opt): Check it.
14281 (generate_canonical_option): Test g prefix.
14282 (option_map): Add -gno- mapping.
14283 (add_misspelling_candidates): Check remapping_prefix_p.
14285 2017-09-20 Jeff Law <law@redhat.com>
14287 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14288 thinko in stack clash protection support.
14290 * explow.c (compute_stack_clash_protection_loop_data): Use
14291 CONST_INT_P instead of explicit test. Verify object is a
14292 CONST_INT_P before looking at INTVAL.
14293 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14294 instead of explicit test.
14296 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14299 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14300 address instead of to r1 and r11.
14302 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14304 * config.gcc: Support "knm".
14305 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14306 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14308 * config/i386/i386.c (m_KNM): Define.
14309 (processor_target_table): Add "knm".
14311 (ix86_option_override_internal): Add "knm".
14312 (ix86_issue_rate): Add PROCESSOR_KNM.
14313 (ix86_adjust_cost): Ditto.
14314 (ia32_multipass_dfa_lookahead): Ditto.
14315 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14316 (fold_builtin_cpu): Add M_INTEL_KNM.
14317 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14318 (processor_type): Add PROCESSOR_KNM.
14319 * config/i386/x86-tune.def: Add m_KNM.
14320 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14322 2017-09-20 Richard Biener <rguenther@suse.de>
14324 PR tree-optimization/80213
14325 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14326 are allowed in empty BBs as well.
14327 (canonicalize_loop_closed_ssa): Also look for other complex
14329 (scop_detection::get_sese): Include the loop-closed PHI block
14331 (scop_detection::merge_sese): Remove code adding extra blocks.
14332 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14333 (build_scops): Assert the final returned scop is invalid.
14335 2017-09-20 Richard Biener <rguenther@suse.de>
14337 PR tree-optimization/82264
14338 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14340 (vn_phi_lookup): Likewise.
14341 (vn_phi_insert): Likewise.
14343 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14345 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14346 that fits into uhwi or shwi, add DW_AT_const_value regardless
14347 of early_dwarf without going through RTL, using add_AT_unsigned
14350 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14351 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14352 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14353 (DEBUG_MACRO_SECTION): Likewise.
14354 (DEBUG_LTO_MACRO_SECTION): Likewise.
14355 (DEBUG_STR_DWO_SECTION): Likewise.
14356 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14357 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14358 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14359 (DEBUG_LTO_LINE_STR_SECTION): Define.
14360 (init_sections_and_labels): Initialize debug_line_str_section
14361 variable. Initialize debug_loc_section for -gdwarf-5 to
14362 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14364 2017-09-20 Richard Biener <rguenther@suse.de>
14366 * graphite-sese-to-poly.c (extract_affine): Properly handle
14367 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14369 2017-09-20 Richard Biener <rguenther@suse.de>
14371 PR tree-optimization/81373
14372 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14373 Force SESE live-out defs to be handled even if they are
14376 2017-09-19 Jeff Law <law@redhat.com>
14378 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14379 nothing for stack adjustments with REG_STACK_CHECK.
14380 * sched-deps.c (parse_add_or_inc): Reject insns with
14381 REG_STACK_CHECK from dependency breaking.
14382 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14383 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14384 * reg-notes.def (STACK_CHECK): New note.
14386 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14387 (ix86_expand_prologue): Dump stack clash info as needed.
14388 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14390 * function.c (dump_stack_clash_frame_info): New function.
14391 * function.h (dump_stack_clash_frame_info): Prototype.
14392 (enum stack_clash_probes): New enum.
14394 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14395 flag_stack_clash_protection.
14396 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14397 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14398 (arm_frame_pointer_required): Likewise.
14399 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14400 (ia64_expand_prologue): Likewise.
14401 * config/mips/mips.c (mips_expand_prologue): Likewise.
14402 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14403 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14404 (sparc_flat_expand_prologue): Likewise.
14405 * config/spu/spu.c (spu_expand_prologue): Likewise.
14407 * explow.c: Include "params.h".
14408 (anti_adjust_stack_and_probe_stack_clash): New function.
14409 (get_stack_check_protect): Likewise.
14410 (compute_stack_clash_protection_loop_data): Likewise.
14411 (emit_stack_clash_protection_loop_start): Likewise.
14412 (emit_stack_clash_protection_loop_end): Likewise.
14413 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14414 Use anti_adjust_stack_and_probe_stack_clash.
14415 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14416 (emit_stack_clash_protection_loop_start): Likewise.
14417 (emit_stack_clash_protection_loop_end): Likewise.
14418 * rtl.h (get_stack_check_protect): Prototype.
14419 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14420 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14421 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14423 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14425 * doc/tm.texi: Rebuilt.
14426 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14427 get_stack_check_protect.
14428 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14429 * config/arm/arm.c (arm_expand_prologue): Likewise.
14430 (arm_frame_pointer_required): Likewise.
14431 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14432 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14433 * config/mips/mips.c (mips_expand_prologue): Likewise.
14434 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14435 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14436 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14437 (sparc_flat_expand_prologue): Likewise.
14439 * common.opt (-fstack-clash-protection): New option.
14440 * flag-types.h (enum stack_check_type): Note difference between
14441 -fstack-check= and -fstack-clash-protection.
14442 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14443 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14444 * toplev.c (process_options): Issue warnings/errors for cases
14445 not handled with -fstack-clash-protection.
14446 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14447 (-fstack-check): Note additional problem with -fstack-check=generic.
14448 Note that -fstack-check is primarily for Ada and refer users
14449 to -fstack-clash-protection for stack-clash-protection.
14450 Document new params for stack clash protection.
14452 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14454 * config/i386/i386.c (ix86_split_long_move): Do not handle
14455 address used for LEA in a special way.
14457 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14459 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14461 2017-09-19 Martin Sebor <msebor@redhat.com>
14464 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14465 of incompatible types.
14467 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14469 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14470 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14471 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14472 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14474 2017-09-19 Richard Biener <rguenther@suse.de>
14476 PR tree-optimization/82244
14477 * tree-vrp.c (remove_range_assertions): Do not propagate
14478 a constant to abnormals but replace the assert with a copy.
14480 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14482 PR rtl-optimization/57878
14483 PR rtl-optimization/68988
14484 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14485 avoidance test involving non_reload_pseudos. Move frequency test
14486 below the general fragmentation avoidance test.
14488 2017-09-19 Richard Biener <rguenther@suse.de>
14490 PR tree-optimization/69728
14491 * graphite-sese-to-poly.c (schedule_error): New global.
14492 (add_loop_schedule): Handle empty domain by failing the
14494 (build_original_schedule): Handle schedule_error.
14496 2017-09-19 Richard Biener <rguenther@suse.de>
14498 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14499 Do not iterate to sibling loops but only to siblings of inner
14502 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14505 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14506 that the registers are different.
14508 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14510 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14511 to processor_model and "amdfam17h" to arch_names_table.
14512 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14514 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14517 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14519 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14520 Alan Hayward <alan.hayward@arm.com>
14521 David Sherwood <david.sherwood@arm.com>
14523 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14525 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14527 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14528 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14529 vect_schedule_slp_instance.
14530 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14531 Update call to vect_slp_analyze_node_operations. Simplify return
14533 (vect_slp_analyze_bb_1): Update call accordingly.
14534 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14535 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14536 (vect_schedule_slp): Update call accordingly.
14538 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14539 Alan Hayward <alan.hayward@arm.com>
14540 David Sherwood <david.sherwood@arm.com>
14542 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14543 with types that aren't in fact scalar.
14545 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14547 * tree-vect-slp.c (vect_record_max_nunits): New function,
14549 (vect_build_slp_tree_1): ...here.
14550 (vect_build_slp_tree_2): Call it for phis too.
14552 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14554 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14555 to vect_get_vec_def_for_operand when getting the mask operand.
14557 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14558 Alan Hayward <alan.hayward@arm.com>
14559 David Sherwood <david.sherwood@arm.com>
14561 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14564 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14565 Alan Hayward <alan.hayward@arm.com>
14566 David Sherwood <david.sherwood@arm.com>
14568 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14569 calculation for vector booleans.
14571 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14572 Alan Hayward <alan.hayward@arm.com>
14573 David Sherwood <david.sherwood@arm.com>
14575 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14577 (vect_transform_stmt): ...here.
14578 (vect_analyze_stmt): Use it instead of calling
14579 vectorizable_live_operation directly.
14581 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14583 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14584 non-SIMT targets in acc vector loops.
14586 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14588 * configure.ac: Add arc and check if assembler supports gdwarf2.
14589 * configure: Regenerate.
14591 2017-09-18 Richard Biener <rguenther@suse.de>
14593 PR tree-optimization/82220
14594 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14595 epilogue niters from the min_profitable_iters compute.
14597 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14600 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14601 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14602 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14604 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14607 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14608 switching to a new text section.
14610 2017-09-18 Richard Biener <rguenther@suse.de>
14612 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14614 (build_alias_set): Reject aliases with no access function.
14616 2017-09-18 Richard Biener <rguenther@suse.de>
14618 PR tree-optimization/79622
14619 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14621 (build_cross_bb_scalars_use): Likewise.
14623 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14625 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14627 2017-09-18 Alan Modra <amodra@gmail.com>
14630 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14631 stack_pointer_rtx for count 0. Update comments. Break up
14632 large rtl expression.
14634 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14636 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14637 Increase to 20 bytes.
14638 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14639 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14640 or avx version of the stub.
14642 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14645 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14646 compute the minimum stack alignment. Also update preferred stack
14647 boundary for leaf functions.
14649 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14651 PR tree-optimization/82228
14652 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14655 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14657 * common/config/nds32/nds32-common.c
14658 (nds32_option_optimization_table): Refine formatting.
14659 (nds32_option_optimization_table): Use -fsched-pressure and
14660 -fomit-frame-pointer for specific optimization level.
14662 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14664 * config/nds32/nds32.c: Refine formatting and comments.
14665 * config/nds32/nds32.h: Likewise.
14666 * config/nds32/nds32.md: Likewise.
14667 * config/nds32/nds32-cost.c: Likewise.
14668 * config/nds32/nds32-isr.c: Likewise.
14669 * config/nds32/nds32-md-auxiliary.c: Likewise.
14670 * config/nds32/nds32-multiple.md: Likewise.
14671 * config/nds32/nds32-predicates.c: Likewise.
14673 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14674 Jakub Jelinek <jakub@redhat.com>
14676 Add support for -std=c++2a.
14677 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14679 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14681 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14684 * doc/extend.texi (Common Function Attributes): Add
14685 references to ARM, AArch64, and S/390 specific attributes.
14686 (Function Specific Option Pragmas): Add AArch64 and S/390
14687 to list of back ends that support the target pragma.
14689 2017-09-15 Nathan Sidwell <nathan@acm.org>
14691 * doc/standards.texi: Fix C++17 description. Update URLs for
14694 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14696 * common.opt (Wcast-align=strict): New warning option.
14697 * doc/invoke.texi: Document -Wcast-align=strict.
14699 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14701 * cgraph.h (cgraph_thunk_info): Add comments.
14702 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14703 assert for VIRTUAL_* arguments stricter.
14705 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14707 PR tree-optimization/71026
14708 * match.pd: Move RDIV patterns from fold-const.c
14709 * fold-const.c (distribute_real_division): Removed.
14710 (fold_binary_loc): Remove calls to distribute_real_divison.
14712 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14714 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14715 c++1z and gnu++1z as deprecated. Change other references to
14716 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14717 Change -Wc++1z-compat to -Wc++17-compat.
14718 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14719 * dwarf2out.c (highest_c_language): Handle C++17.
14720 (gen_compile_unit_die): Likewise.
14722 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14724 PR rtl-optimization/82192
14725 * combine.c (make_extraction): Don't look through non-paradoxical
14726 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14729 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14730 Alan Hayward <alan.hayward@arm.com>
14731 David Sherwood <david.sherwood@arm.com>
14733 * target.def (function_arg_offset): New hook.
14734 * targhooks.h (default_function_arg_offset): Declare.
14735 * targhooks.c (default_function_arg_offset): New function.
14736 * function.c (locate_and_pad_parm): Use
14737 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14738 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14739 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14740 * doc/tm.texi: Regenerate.
14741 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14742 * config/spu/spu.c (spu_function_arg_offset): New function.
14743 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14744 * system.h (FUNCTION_ARG_OFFSET): Poison.
14746 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14747 Alan Hayard <alan.hayward@arm.com>
14748 David Sherwood <david.sherwood@arm.com>
14750 * target.def (truly_noop_truncation): New hook.
14751 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14752 than TRULY_NOOP_TRUNCATION.
14753 * hooks.h (hook_bool_uint_uint_true): Declare.
14754 * hooks.c (hook_bool_uint_uint_true): New function.
14755 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14756 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14757 * doc/tm.texi: Regenerate.
14758 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14759 rather than TRULY_NOOP_TRUNCATION in comments.
14760 (simplify_comparison): Likewise.
14761 (record_truncated_value): Likewise.
14762 * expmed.c (extract_bit_field_1): Likewise.
14763 (extract_split_bit_field): Likewise.
14764 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14765 instead of TRULY_NOOP_TRUNCATION.
14766 * function.c (assign_parm_setup_block): Likewise.
14767 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14768 * rtlhooks.c: Include target.h.
14769 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14770 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14771 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14772 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14773 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14774 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14775 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14776 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14777 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14778 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14779 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14780 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14781 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14782 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14783 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14784 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14785 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14786 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14787 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14788 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14789 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14790 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14791 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14792 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14793 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14794 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14795 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14796 rather than TRULY_NOOP_TRUNCATION in comments.
14797 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14798 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14799 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14800 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14801 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14802 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14803 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14804 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14805 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14806 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14807 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14808 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14809 rather than TRULY_NOOP_TRUNCATION in comments.
14810 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14811 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14812 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14813 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14814 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14815 TRULY_NOOP_TRUNCATION condition.
14816 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14817 (TRULY_NOOP_TRUNCATION): Delete.
14818 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14819 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14820 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14821 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14822 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14823 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14824 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14825 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14826 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14827 rather than TRULY_NOOP_TRUNCATION in comments.
14828 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14829 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14830 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14831 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14832 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14833 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14835 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14838 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14839 (*cmp_ior): Likewise.
14840 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14841 (*ior_scc_scc_cmp): Likewise.
14842 (*and_scc_scc): Likewise.
14843 (*and_scc_scc_cmp): Likewise.
14845 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14846 Alan Hayard <alan.hayward@arm.com>
14847 David Sherwood <david.sherwood@arm.com>
14849 * target.def (can_change_mode_class): New hook.
14850 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14851 (hard_regno_nregs): Likewise.
14852 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14853 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14854 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14855 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14856 (LOAD_EXTEND_OP): Update accordingly.
14857 * doc/tm.texi: Regenerate.
14858 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14859 CANNOT_CHANGE_MODE_CLASS.
14860 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14861 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14862 * combine.c (simplify_set): Update accordingly.
14863 * emit-rtl.c (validate_subreg): Likewise.
14864 * recog.c (general_operand): Likewise.
14865 * regcprop.c (mode_change_ok): Likewise.
14866 * reload1.c (choose_reload_regs): Likewise.
14867 (inherit_piecemeal_p): Likewise.
14868 * rtlanal.c (simplify_subreg_regno): Likewise.
14869 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14870 instead of CANNOT_CHANGE_MODE_CLASS.
14871 (reload_cse_simplify_operands): Likewise.
14872 * reload.c (push_reload): Use targetm.can_change_mode_class
14873 instead of CANNOT_CHANGE_MODE_CLASS.
14874 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14875 REG_CANNOT_CHANGE_MODE_P.
14876 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14877 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14878 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14879 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14880 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14881 (arm_can_change_mode_class): New function.
14882 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14883 than CANNOT_CHANGE_MODE_CLASS in comments.
14884 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14885 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14886 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14887 (ix86_can_change_mode_class): ...this new function, inverting the
14888 sense of the return value.
14889 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14890 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14891 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14892 (ia64_can_change_mode_class): New function.
14893 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14894 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14895 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14896 (m32c_can_change_mode_class): ...this new function, inverting the
14897 sense of the return value.
14898 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14899 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14900 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14901 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14902 (mips_can_change_mode_class): ...this new function, inverting the
14903 sense of the return value.
14904 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14905 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14906 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14907 (msp430_can_change_mode_class): New function.
14908 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14909 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14910 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14911 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14912 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14913 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14914 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14915 (pa_cannot_change_mode_class): Replace with...
14916 (pa_can_change_mode_class): ...this new function, inverting the
14917 sense of the return value.
14918 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14919 than CANNOT_CHANGE_MODE_CLASS in comments.
14920 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14921 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14922 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14923 (pdp11_cannot_change_mode_class): Replace with...
14924 (pdp11_can_change_mode_class): ...this new function, inverting the
14925 sense of the return value.
14926 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14927 * config/powerpcspe/powerpcspe-protos.h
14928 (rs6000_cannot_change_mode_class_ptr): Delete.
14929 * config/powerpcspe/powerpcspe.c
14930 (rs6000_cannot_change_mode_class_ptr): Delete.
14931 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14932 (rs6000_option_override_internal): Assign to
14933 targetm.can_change_mode_class instead of
14934 rs6000_cannot_change_mode_class_ptr.
14935 (rs6000_cannot_change_mode_class): Replace with...
14936 (rs6000_can_change_mode_class): ...this new function, inverting the
14937 sense of the return value.
14938 (rs6000_debug_cannot_change_mode_class): Replace with...
14939 (rs6000_debug_can_change_mode_class): ...this new function.
14940 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14941 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14942 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14943 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14944 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14946 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14947 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14948 (rs6000_option_override_internal): Assign to
14949 targetm.can_change_mode_class instead of
14950 rs6000_cannot_change_mode_class_ptr.
14951 (rs6000_cannot_change_mode_class): Replace with...
14952 (rs6000_can_change_mode_class): ...this new function, inverting the
14953 sense of the return value.
14954 (rs6000_debug_cannot_change_mode_class): Replace with...
14955 (rs6000_debug_can_change_mode_class): ...this new function.
14956 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14957 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14958 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14959 (s390_can_change_mode_class): ...this new function, inverting the
14960 sense of the return value.
14961 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14962 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14963 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14964 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14965 (sh_cannot_change_mode_class): Replace with...
14966 (sh_can_change_mode_class): ...this new function, inverting the
14967 sense of the return value.
14968 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14969 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14970 (sparc_can_change_mode_class): New function.
14971 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14972 * config/spu/spu.c (spu_can_change_mode_class): New function.
14973 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14974 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14975 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14976 (visium_can_change_mode_class): New function.
14977 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14979 2017-09-15 Richard Biener <rguenther@suse.de>
14981 PR tree-optimization/82217
14982 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14983 but not undefined case.
14985 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14988 * postreload.c (reload_cse_simplify_operands): Skip
14989 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14991 2017-09-15 Richard Biener <rguenther@suse.de>
14993 PR tree-optimization/68823
14994 * graphite-scop-detection.c (build_alias_set): If we have a
14995 possible dependence check whether we can handle them by just
14996 looking at the DRs DR_ACCESS_FNs.
14997 (build_scops): If build_alias_set fails, fail the SCOP.
14999 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15001 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15002 to support float128 built-in functions that require the ISA 3.0
15004 (BU_FLOAT128_3_HW): Likewise.
15005 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15006 built-in functions.
15007 (FMAF128): Likewise.
15009 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15010 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15011 floating point instructions.
15012 (rs6000_invalid_builtin): Likewise.
15013 (rs6000_builtin_mask_names): Likewise.
15014 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15015 (RS6000_BTM_FLOAT128_HW): Likewise.
15016 (RS6000_BTM_COMMON): Likewise.
15017 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15019 * doc/extend.texi (RS/6000 built-in functions): Document the
15020 IEEE 128-bit floating point square root and fused multiply-add
15021 built-in functions.
15023 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15025 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15026 reg (r2) isn't in the set of registers defined in the prologue.
15028 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15029 Alan Hayward <alan.hayward@arm.com>
15030 David Sherwood <david.sherwood@arm.com>
15032 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15033 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15034 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15035 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15036 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15038 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15039 max_vectorization_factor.
15040 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15042 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15043 Alan Hayward <alan.hayward@arm.com>
15044 David Sherwood <david.sherwood@arm.com>
15046 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15047 (vect_worthwhile_without_simd_p): Declare.
15048 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15049 (vectorizable_reduction): Use it.
15050 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15051 (vectorizable_operation): Likewise.
15053 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15054 Alan Hayward <alan.hayward@arm.com>
15055 David Sherwood <david.sherwood@arm.com>
15057 * tree-vectorizer.h (vect_get_num_copies): New function.
15058 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15059 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15060 (vectorizable_induction): Likewise.
15061 (vectorizable_live_operation): Likewise.
15062 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15063 (vectorizable_bswap): Likewise.
15064 (vectorizable_call): Likewise.
15065 (vectorizable_conversion): Likewise.
15066 (vectorizable_assignment): Likewise.
15067 (vectorizable_shift): Likewise.
15068 (vectorizable_operation): Likewise.
15069 (vectorizable_store): Likewise.
15070 (vectorizable_load): Likewise.
15071 (vectorizable_condition): Likewise.
15072 (vectorizable_comparison): Likewise.
15073 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15075 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15076 Alan Hayward <alan.hayward@arm.com>
15077 David Sherwood <david.sherwood@arm.com>
15079 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15080 of vect_init_vector.
15082 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15083 Alan Hayward <alan.hayward@arm.com>
15084 David Sherwood <david.sherwood@arm.com>
15086 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15087 an inline wrapper that provides a location.
15088 (gimple_build_vector): Likewise.
15089 * gimple-fold.c (gimple_build_vector_from_val): New function.
15090 (gimple_build_vector): Likewise.
15091 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15092 functions to build the initial value. Always return a gimple value.
15093 (get_initial_defs_for_reduction): Likewise. Only compute
15095 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15096 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15097 (vectorizable_induction): Use gimple_build_vector rather than
15100 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15101 Alan Hayward <alan.hayward@arm.com>
15102 David Sherwood <david.sherwood@arm.com>
15104 * target.h (vec_perm_indices): New typedef.
15105 (auto_vec_perm_indices): Likewise.
15106 * optabs-query.h: Include target.h
15107 (can_vec_perm_p): Take a vec_perm_indices *.
15108 * optabs-query.c (can_vec_perm_p): Likewise.
15109 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15110 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15111 * tree-vect-generic.c (lower_vec_perm): Likewise.
15112 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15113 (vect_grouped_load_supported): Likewise.
15114 (vect_shift_permute_load_chain): Likewise.
15115 (vect_permute_store_chain): Use auto_vec_perm_indices.
15116 (vect_permute_load_chain): Likewise.
15117 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15118 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15119 Update uses of can_vec_perm_p.
15120 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15121 mode with a number of elements. Take a vec_perm_indices *.
15122 (vect_create_epilog_for_reduction): Update accordingly.
15123 Use auto_vec_perm_indices.
15124 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15125 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15126 (vect_transform_slp_perm_load): Likewise.
15127 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15128 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15129 (vect_gen_perm_mask_checked): Likewise.
15130 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15131 (vect_gen_perm_mask_checked): Likewise.
15132 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15133 (vectorizable_store): Likewise.
15134 (vectorizable_load): Likewise.
15135 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15136 (vectorizable_bswap): Likewise.
15138 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15139 Alan Hayward <alan.hayward@arm.com>
15140 David Sherwood <david.sherwood@arm.com>
15142 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15143 * tree.c (build_vector): Likewise.
15144 (build_vector_from_ctor): Update accordingly.
15145 (build_vector_from_val): Likewise.
15146 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15147 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15148 * tree-vect-generic.c (add_rshift): Likewise.
15149 (expand_vector_divmod): Likewise.
15150 (optimize_vector_constructor): Likewise.
15151 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15152 (vect_transform_slp_perm_load): Likewise.
15153 (vect_schedule_slp_instance): Likewise.
15154 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15155 (vectorizable_call): Likewise.
15156 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15157 * expmed.c (make_tree): Likewise.
15158 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15159 a vector passed to build_vector.
15160 (fold_convert_const): Likewise.
15161 (exact_inverse): Likewise.
15162 (fold_ternary_loc): Likewise.
15163 (fold_relational_const): Likewise.
15164 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15165 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15166 (const_unop): Likewise. Store the reduction accumulator in a
15167 variable rather than an array.
15168 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15169 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15170 the new vector, rather than constructing it after the input arrays.
15171 (native_interpret_vector): Use auto_vec<tree> when building
15172 a vector passed to build_vector. Add elements in order.
15173 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15174 auto_vec<tree> when building a vector passed to build_vector.
15175 (vect_create_epilog_for_reduction): Likewise.
15176 (vectorizable_induction): Likewise.
15177 (get_initial_def_for_reduction): Likewise. Fix indentation of
15179 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15181 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15182 passed to build_vector.
15184 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15185 Alan Hayward <alan.hayward@arm.com>
15186 David Sherwood <david.sherwood@arm.com>
15188 * tree-core.h (tree_base::u): Add an "nelts" field.
15189 (tree_vector): Use VECTOR_CST_NELTS as the length.
15190 * tree.c (tree_size): Likewise.
15191 (make_vector): Initialize VECTOR_CST_NELTS.
15192 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15193 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15194 TYPE_VECTOR_SUBPARTS.
15195 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15196 as the number of units, setting it from VECTOR_CST_NELTS.
15197 (const_vector_from_tree): Likewise.
15198 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15199 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15200 (fold_negate_expr_1): Likewise.
15201 (fold_convert_const): Likewise.
15202 (const_binop): Likewise. Differentiate the number of output and
15204 (const_unop): Likewise.
15205 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15206 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15207 in cases that did the opposite.
15209 2017-09-14 Richard Biener <rguenther@suse.de>
15211 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15214 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15216 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15218 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15221 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15222 if and where to split a bb, except for splitting before debug insn
15223 sequences followed by non-label real insn. Delete debug insns
15224 in between basic blocks.
15226 * combine.c (make_compound_operation_int): Formatting fixes.
15228 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15229 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15230 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15231 * config/sol2.h (LINK_EH_SPEC): Likewise.
15232 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15233 * config/s390/linux.h (LINK_SPEC): Likewise.
15234 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15235 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15236 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15237 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15238 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15239 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15240 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15241 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15243 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15245 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15247 (ENDFILE_LINUX_SPEC): Likewise.
15248 (LINK_EH_SPEC): Likewise.
15249 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15250 (LINK_OS_LINUX_SPEC32): Likewise.
15251 (LINK_OS_LINUX_SPEC64): Likewise.
15252 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15253 (LINK_OS_LINUX_SPEC): Likewise.
15255 2017-09-13 Martin Liska <mliska@suse.cz>
15257 PR middle-end/82154
15258 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15259 CASE_HIGH is NULL_TREE.
15261 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15262 Alan Hayward <alan.hayward@arm.com>
15263 David Sherwood <david.sherwood@arm.com>
15265 * target.def (secondary_memory_needed): New hook.
15266 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15267 instead of SECONDARY_MEMORY_NEEDED.
15268 (secondary_memory_needed_mode): Likewise.
15269 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15270 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15271 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15272 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15273 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15274 * doc/tm.texi: Regenerate.
15275 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15276 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15277 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15278 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15279 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15280 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15281 mode argument first and change the reg_class arguments to reg_class_t.
15282 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15283 Make static. Update the call to inline_secondary_memory_needed.
15284 (ix86_register_move_cost): Update the call to
15285 inline_secondary_memory_needed.
15286 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15287 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15289 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15290 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15292 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15293 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15294 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15295 and match hook interface. Add comment from mips.h.
15296 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15297 * config/mmix/mmix.md (truncdfsf2): Refer to
15298 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15300 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15301 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15302 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15303 (pa_secondary_memory_needed): New function.
15304 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15305 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15306 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15307 (pdp11_secondary_memory_needed): Make static and match hook interface.
15308 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15309 * config/powerpcspe/powerpcspe-protos.h
15310 (rs6000_secondary_memory_needed_ptr): Delete.
15311 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15313 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15314 (rs6000_option_override_internal): Assign to
15315 targetm.secondary_memory_needed rather than
15316 rs6000_secondary_memory_needed_ptr.
15317 (rs6000_secondary_memory_needed): Match hook interface.
15318 (rs6000_debug_secondary_memory_needed): Likewise.
15319 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15320 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15321 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15322 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15323 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15324 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15326 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15327 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15328 (rs6000_option_override_internal): Assign to
15329 targetm.secondary_memory_needed rather than
15330 rs6000_secondary_memory_needed_ptr.
15331 (rs6000_secondary_memory_needed): Match hook interface.
15332 (rs6000_debug_secondary_memory_needed): Likewise.
15333 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15334 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15335 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15336 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15337 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15338 (sparc_secondary_memory_needed): New function.
15339 * lra-constraints.c (check_and_process_move): Refer to
15340 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15342 (curr_insn_transform): Likewise.
15343 (process_alt_operands): Use targetm.secondary_memory_needed
15344 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15345 (check_secondary_memory_needed_p): Likewise.
15346 (choose_split_class): Likewise.
15347 * reload.c: Unconditionally include code that was previously
15348 conditional on SECONDARY_MEMORY_NEEDED.
15349 (push_secondary_reload): Use targetm.secondary_memory_needed
15350 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15351 (push_reload): Likewise.
15352 * reload1.c: Unconditionally include code that was previously
15353 conditional on SECONDARY_MEMORY_NEEDED.
15354 (choose_reload_regs): Use targetm.secondary_memory_needed
15355 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15356 (gen_reload): Likewise.
15357 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15359 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15360 Alan Hayward <alan.hayward@arm.com>
15361 David Sherwood <david.sherwood@arm.com>
15363 * target.def (secondary_memory_needed_mode): New hook:
15364 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15365 * targhooks.h (default_secondary_memory_needed_mode): New function.
15366 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15367 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15368 * doc/tm.texi: Regenerate.
15369 * lra-constraints.c (check_and_process_move): Use
15370 targetm.secondary_memory_needed_mode instead of
15371 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15372 (curr_insn_transform): Likewise.
15373 * reload.c (get_secondary_mem): Likewise.
15374 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15375 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15377 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15378 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15379 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15380 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15381 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15383 * config/powerpcspe/powerpcspe-protos.h
15384 (rs6000_secondary_memory_needed_mode): Delete.
15385 * config/powerpcspe/powerpcspe.c
15386 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15387 (rs6000_secondary_memory_needed_mode): Make static.
15388 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15389 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15391 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15393 (rs6000_secondary_memory_needed_mode): Make static.
15394 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15395 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15396 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15397 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15398 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15400 (sparc_secondary_memory_needed_mode): New function.
15401 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15403 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15405 * config/aarch64/constraints.md (Umq): New constraint.
15406 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15408 (mov<mode>): Update condition.
15410 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15412 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15413 when bitposition is the same.
15415 2017-09-13 Richard Biener <rguenther@suse.de>
15417 * dwarf2out.c (output_die_symbol): Remove.
15418 (output_die): Do not output a DIEs symbol.
15420 2017-09-13 Richard Biener <rguenther@suse.de>
15422 PR middle-end/82128
15423 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15424 default-def to avoid breaking iterator update with the weird
15425 interaction with cgraph_update_edges_for_call_stmt_node.
15427 2017-09-13 Richard Biener <rguenther@suse.de>
15429 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15430 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15431 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15432 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15434 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15436 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15437 Disable pc relative literal load irrespective of
15438 TARGET_FIX_ERR_A53_84341 for default.
15440 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15442 * config/sparc/sparc.c (output_return): Output the source location of
15443 the insn in the delay slot, if any.
15444 (output_sibcall): Likewise.
15446 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15449 * common.opt (-static-pie): New alias.
15450 (shared): Negate static-pie.
15451 (-no-pie): Update help text.
15453 (static-pie): New option.
15454 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15455 -static-pie support.
15456 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15457 (LINK_EH_SPEC): Likewise.
15458 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15459 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15460 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15461 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15462 (init_gcc_specs): Likewise.
15463 (init_spec): Likewise.
15464 (display_help): Update help message for -pie.
15465 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15468 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15470 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15471 (movdi_aarch64): Likewise.
15472 (movti_aarch64): Likewise.
15474 2017-09-12 Simon Wright <simon@pushface.org>
15477 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15478 calculation of the minor version, always output as 0.
15480 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15483 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15484 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15485 on it early, rather than manual conversion late. For
15486 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15487 instead of performing manual conversion.
15489 2017-09-12 Carl Love <cel@us.ibm.com>
15491 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15492 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15494 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15495 VMULOSW): Add definitions.
15496 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15497 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15498 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15499 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15500 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15502 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15504 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15506 (movti_aarch64): Likewise.
15507 (movdf_aarch64): Likewise.
15508 (movtf_aarch64): Likewise.
15509 (load_pairdi): Likewise.
15510 (store_pairdi): Likewise.
15511 (load_pairdf): Likewise.
15512 (store_pairdf): Likewise.
15513 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15514 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15515 (ldr_got_small_<mode>): Likewise.
15516 (ldr_got_small_28k_<mode>): Likewise.
15517 (ldr_got_tiny): Likewise.
15518 * config/aarch64/iterators.md (ldst_sz): New.
15519 (ldpstp_sz): Likewise.
15520 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15522 (thunderx_load): Split load_8 to load_16.
15523 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15525 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15526 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15527 (xgene1_store_pair): Split store_8 to store_16.
15528 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15529 (falkor_st_0_st_sd): Split store_8 to store_16.
15531 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15533 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15534 and store1/2/3/4 to store_4/8/12/16.
15535 * config/aarch64/aarch64.md: Update for rename.
15536 * config/arm/arm.md: Likewise.: Likewise.
15537 * config/arm/arm.c: Likewise.
15538 * config/arm/thumb1.md: Likewise.
15539 * config/arm/thumb2.md: Likewise.
15540 * config/arm/vfp.md: Likewise.
15541 * config/arm/arm-generic.md: Likewise.
15542 * config/arm/arm1020e.md: Likewise.
15543 * config/arm/arm1026ejs.md: Likewise.
15544 * config/arm/arm1136jfs.md: Likewise.
15545 * config/arm/arm926ejs.md: Likewise.
15546 * config/arm/cortex-a15.md: Likewise.
15547 * config/arm/cortex-a17.md: Likewise.
15548 * config/arm/cortex-a5.md: Likewise.
15549 * config/arm/cortex-a53.md: Likewise.
15550 * config/arm/cortex-a57.md: Likewise.
15551 * config/arm/cortex-a7.md: Likewise.
15552 * config/arm/cortex-a8.md: Likewise.
15553 * config/arm/cortex-a9.md: Likewise.
15554 * config/arm/cortex-m4.md: Likewise.
15555 * config/arm/cortex-m7.md: Likewise.
15556 * config/arm/cortex-r4.md: Likewise.
15557 * config/arm/exynos-m1.md: Likewise.
15558 * config/arm/fa526.md: Likewise.
15559 * config/arm/fa606te.md: Likewise.
15560 * config/arm/fa626te.md: Likewise.
15561 * config/arm/fa726te.md: Likewise.
15562 * config/arm/fmp626.md: Likewise.
15563 * config/arm/iwmmxt.md: Likewise.
15564 * config/arm/ldmstm.md: Likewise.
15565 * config/arm/marvell-pj4.md: Likewise.
15566 * config/arm/xgene1.md: Likewise.
15567 * config/aarch64/thunderx.md: Likewise.
15568 * config/aarch64/thunderx2t99.md: Likewise.
15569 * config/aarch64/falkor.md: Likewise.
15571 2017-09-12 Martin Liska <mliska@suse.cz>
15573 * attribs.c (private_lookup_attribute): New function.
15574 * attribs.h (private_lookup_attribute): Declared here.
15575 (lookup_attribute): Called from this place.
15577 2017-09-12 Richard Biener <rguenther@suse.de>
15579 PR tree-optimization/82157
15580 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15581 stmts with side-effects.
15583 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15584 Alan Hayward <alan.hayward@arm.com>
15585 David Sherwood <david.sherwood@arm.com>
15587 * target.def (hard_regno_nregs): New hook.
15588 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15589 * targhooks.h (default_hard_regno_nregs): Declare.
15590 * targhooks.c (default_hard_regno_nregs): New function.
15591 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15592 (TARGET_HARD_REGNO_NREGS): ...this hook.
15593 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15594 (CLASS_MAX_NREGS): Likewise.
15595 * doc/tm.texi: Regenerate.
15596 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15597 instead of HARD_REGNO_NREGS.
15598 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15599 HARD_REGNO_NREGS in the comment.
15600 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15601 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15602 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15603 Return an unsigned int.
15604 (TARGET_HARD_REGNO_NREGS): Redefine.
15605 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15606 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15607 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15608 (arc_hard_regno_nregs): New function.
15609 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15610 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15611 (arm_hard_regno_nregs): New function.
15612 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15613 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15614 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15615 (TARGET_HARD_REGNO_NREGS): Redefine.
15616 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15617 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15618 (HARD_REGNO_NREGS): Delete.
15619 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15620 (cr16_hard_regno_nregs): New function.
15621 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15622 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15623 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15624 (cris_hard_regno_nregs): New function.
15625 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15626 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15627 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15628 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15629 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15630 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15631 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15632 (frv_hard_regno_nregs): Make static. Take and return an
15634 (frv_class_max_nregs): Remove outdated copy of documentation.
15635 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15636 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15637 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15638 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15639 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15640 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15641 (TARGET_HARD_REGNO_NREGS): Redefine.
15642 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15643 (CLASS_MAX_NREGS): Update comment.
15644 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15645 (ia64_hard_regno_nregs): New function.
15646 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15647 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15648 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15649 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15650 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15652 (m32c_hard_regno_nregs): Likewise. Make static.
15653 (TARGET_HARD_REGNO_NREGS): Redefine.
15654 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15655 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15656 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15657 (m68k_hard_regno_nregs): New function.
15658 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15659 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15660 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15661 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15662 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15663 Take and return an unsigned int.
15664 (TARGET_HARD_REGNO_NREGS): Redefine.
15665 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15666 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15667 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15668 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15669 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15670 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15671 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15672 (msp430_hard_regno_nregs): Make static. Take and return an
15674 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15675 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15676 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15677 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15678 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15679 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15680 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15681 (TARGET_HARD_REGNO_NREGS): Redefine.
15682 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15683 (PA_HARD_REGNO_NREGS): ...this.
15684 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15685 (PA_HARD_REGNO_NREGS): ...this.
15686 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15687 (pa_hard_regno_nregs): New function.
15688 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15689 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15690 (pdp11_hard_regno_nregs): New function.
15691 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15692 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15693 (rs6000_hard_regno_nregs_hook): New function.
15694 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15695 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15696 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15697 Take and return an unsigned int. Move earlier in file.
15698 (TARGET_HARD_REGNO_NREGS): Redefine.
15699 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15700 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15701 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15702 (rl78_hard_regno_nregs): Make static. Take and return an
15704 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15705 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15706 (rs6000_hard_regno_nregs_hook): New function.
15707 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15708 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15709 (TARGET_HARD_REGNO_NREGS): Redefine.
15710 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15711 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15712 instead of HARD_REGNO_NREGS.
15713 (s390_hard_regno_nregs): New function.
15714 (s390_hard_regno_mode_ok): Add comment from s390.h.
15715 (TARGET_HARD_REGNO_NREGS): Redefine.
15716 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15717 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15718 (sh_hard_regno_nregs): New function.
15719 (sh_pass_in_reg_p): Use it.
15720 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15721 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15722 (sparc_hard_regno_nregs): New function.
15723 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15724 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15725 (spu_function_arg_advance): Use it, supplying a valid register number.
15726 (TARGET_HARD_REGNO_NREGS): Redefine.
15727 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15728 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15729 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15730 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15731 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15732 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15733 (CLASS_MAX_NREGS): Remove copy of old documentation.
15734 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15735 (visium_hard_regno_nregs): New function.
15736 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15737 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15738 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15739 xtensa_hard_regno_nregs): New function.
15740 * system.h (HARD_REGNO_NREGS): Poison.
15742 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15744 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15745 hard_regno_nregs instead of HARD_REGNO_NREGS.
15746 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15747 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15748 (c6x_expand_epilogue): Likewise.
15749 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15750 (frv_read_iacc_argument): Likewise.
15751 * config/sh/sh.c: Include regs.h.
15752 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15753 (regs_used): Likewise.
15754 (output_stack_adjust): Likewise.
15755 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15756 * expmed.c: Include regs.h.
15757 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15758 * ree.c: Include regs.h.
15759 (combine_reaching_defs): Use hard_regno_nregs instead of
15761 (add_removable_extension): Likewise.
15763 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15765 * regs.h (hard_regno_nregs): Turn into a function.
15766 (end_hard_regno): Update accordingly.
15767 * caller-save.c (setup_save_areas): Likewise.
15768 (save_call_clobbered_regs): Likewise.
15769 (replace_reg_with_saved_mem): Likewise.
15770 (insert_restore): Likewise.
15771 (insert_save): Likewise.
15772 * combine.c (can_change_dest_mode): Likewise.
15773 (move_deaths): Likewise.
15774 (distribute_notes): Likewise.
15775 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15776 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15777 (rs6000_split_multireg_move): Likewise.
15778 (rs6000_register_move_cost): Likewise.
15779 (rs6000_memory_move_cost): Likewise.
15780 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15781 (rs6000_split_multireg_move): Likewise.
15782 (rs6000_register_move_cost): Likewise.
15783 (rs6000_memory_move_cost): Likewise.
15784 * cselib.c (cselib_reset_table): Likewise.
15785 (cselib_lookup_1): Likewise.
15786 * emit-rtl.c (set_mode_and_regno): Likewise.
15787 * function.c (aggregate_value_p): Likewise.
15788 * ira-color.c (setup_profitable_hard_regs): Likewise.
15789 (check_hard_reg_p): Likewise.
15790 (calculate_saved_nregs): Likewise.
15791 (assign_hard_reg): Likewise.
15792 (improve_allocation): Likewise.
15793 (calculate_spill_cost): Likewise.
15794 * ira-emit.c (modify_move_list): Likewise.
15795 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15796 (ira_hard_reg_in_set_p): Likewise.
15797 * ira.c (setup_reg_mode_hard_regset): Likewise.
15798 (clarify_prohibited_class_mode_regs): Likewise.
15799 (check_allocation): Likewise.
15800 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15801 (lra_setup_reg_renumber): Likewise.
15802 (setup_try_hard_regno_pseudos): Likewise.
15803 (spill_for): Likewise.
15804 (assign_hard_regno): Likewise.
15805 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15806 * lra-constraints.c (in_class_p): Likewise.
15807 (lra_constraint_offset): Likewise.
15808 (simplify_operand_subreg): Likewise.
15809 (lra_constraints): Likewise.
15810 (split_reg): Likewise.
15811 (split_if_necessary): Likewise.
15812 (invariant_p): Likewise.
15813 (inherit_in_ebb): Likewise.
15814 * lra-lives.c (process_bb_lives): Likewise.
15815 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15816 (get_hard_regs): Likewise.
15817 (do_remat): Likewise.
15818 * lra-spills.c (assign_spill_hard_regs): Likewise.
15819 * mode-switching.c (create_pre_exit): Likewise.
15820 * postreload.c (reload_combine_recognize_pattern): Likewise.
15821 * recog.c (peep2_find_free_register): Likewise.
15822 * regcprop.c (kill_value_regno): Likewise.
15823 (set_value_regno): Likewise.
15824 (copy_value): Likewise.
15825 (maybe_mode_change): Likewise.
15826 (find_oldest_value_reg): Likewise.
15827 (copyprop_hardreg_forward_1): Likewise.
15828 * regrename.c (check_new_reg_p): Likewise.
15829 (regrename_do_replace): Likewise.
15830 * reload.c (push_reload): Likewise.
15831 (combine_reloads): Likewise.
15832 (find_dummy_reload): Likewise.
15833 (operands_match_p): Likewise.
15834 (find_reloads): Likewise.
15835 (find_equiv_reg): Likewise.
15836 (reload_adjust_reg_for_mode): Likewise.
15837 * reload1.c (count_pseudo): Likewise.
15838 (count_spilled_pseudo): Likewise.
15839 (find_reg): Likewise.
15840 (clear_reload_reg_in_use): Likewise.
15841 (free_for_value_p): Likewise.
15842 (allocate_reload_reg): Likewise.
15843 (choose_reload_regs): Likewise.
15844 (reload_adjust_reg_for_temp): Likewise.
15845 (emit_reload_insns): Likewise.
15846 (delete_output_reload): Likewise.
15847 * rtlanal.c (subreg_get_info): Likewise.
15848 * sched-deps.c (sched_analyze_reg): Likewise.
15849 * sel-sched.c (init_regs_for_mode): Likewise.
15850 (mark_unavailable_hard_regs): Likewise.
15851 (choose_best_reg_1): Likewise.
15852 (verify_target_availability): Likewise.
15853 * valtrack.c (dead_debug_insert_temp): Likewise.
15854 * var-tracking.c (track_loc_p): Likewise.
15855 (emit_note_insn_var_location): Likewise.
15856 * varasm.c (make_decl_rtl): Likewise.
15857 * reginfo.c (choose_hard_reg_mode): Likewise.
15858 (init_reg_modes_target): Refer directly to
15859 this_target_regs->x_hard_regno_nregs.
15861 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15863 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15864 instead of hard_regno_nregs.
15866 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15868 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15869 end_hard_regno instead of hard_regno_nregs.
15870 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15871 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15872 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15873 * ira-color.c (improve_allocation): Likewise.
15874 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15875 * lra-lives.c (mark_regno_live): Likewise.
15876 (mark_regno_dead): Likewise.
15877 * lra-remat.c (operand_to_remat): Likewise.
15878 * lra.c (collect_non_operand_hard_regs): Likewise.
15879 * postreload.c (reload_combine_note_store): Likewise.
15880 (move2add_valid_value_p): Likewise.
15881 * reload.c (regno_clobbered_p): Likewise.
15883 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15885 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15887 * config/v850/v850.c (v850_reorg): Likewise.
15888 * reload.c (refers_to_regno_for_reload_p): Likewise.
15889 (find_equiv_reg): Likewise.
15890 * reload1.c (reload_reg_reaches_end_p): Likewise.
15892 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15894 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15896 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15897 * config/arm/arm.c (output_move_neon): Likewise.
15898 (arm_attr_length_move_neon): Likewise.
15899 (neon_split_vcombine): Likewise.
15900 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15901 (c6x_mark_reg_written): Likewise.
15902 (c6x_dwarf_register_span): Likewise.
15903 * config/i386/i386.c (ix86_save_reg): Likewise.
15904 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15905 (rws_access_reg): Likewise.
15906 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15907 * mode-switching.c (create_pre_exit): Likewise.
15908 * ree.c (combine_reaching_defs): Likewise.
15909 (add_removable_extension): Likewise.
15910 * regcprop.c (find_oldest_value_reg): Likewise.
15911 (copyprop_hardreg_forward_1): Likewise.
15912 * reload.c (reload_inner_reg_of_subreg): Likewise.
15913 (push_reload): Likewise.
15914 (combine_reloads): Likewise.
15915 (find_dummy_reload): Likewise.
15916 (reload_adjust_reg_for_mode): Likewise.
15917 * reload1.c (find_reload_regs): Likewise.
15918 (forget_old_reloads_1): Likewise.
15919 (reload_reg_free_for_value_p): Likewise.
15920 (reload_adjust_reg_for_temp): Likewise.
15921 (emit_reload_insns): Likewise.
15922 (delete_output_reload): Likewise.
15923 * sel-sched.c (choose_best_reg_1): Likewise.
15924 (choose_best_pseudo_reg): Likewise.
15926 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15927 Alan Hayward <alan.hayward@arm.com>
15928 David Sherwood <david.sherwood@arm.com>
15930 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15931 * target.def (slow_unaligned_access): New hook.
15932 * targhooks.h (default_slow_unaligned_access): Declare.
15933 * targhooks.c (default_slow_unaligned_access): New function.
15934 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15935 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15936 * doc/tm.texi: Regenerate.
15937 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15938 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15939 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15941 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15942 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15944 (rs6000_slow_unaligned_access): New function.
15945 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15946 (expand_block_compare): Likewise.
15947 (expand_strn_compare): Likewise.
15948 (rs6000_rtx_costs): Likewise.
15949 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15950 (riscv_slow_unaligned_access): Likewise.
15951 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15952 (riscv_slow_unaligned_access_p): ...this and make static.
15953 (riscv_option_override): Update accordingly.
15954 (riscv_slow_unaligned_access): New function.
15955 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15956 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15957 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15958 (rs6000_slow_unaligned_access): New function.
15959 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15960 (rs6000_rtx_costs): Likewise.
15961 * config/rs6000/rs6000-string.c (expand_block_compare)
15962 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15963 of SLOW_UNALIGNED_ACCESS.
15964 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15965 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15966 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15967 of SLOW_UNALIGNED_ACCESS.
15968 * expmed.c (simple_mem_bitfield_p): Likewise.
15969 * expr.c (alignment_for_piecewise_move): Likewise.
15970 (emit_group_load_1): Likewise.
15971 (emit_group_store): Likewise.
15972 (copy_blkmode_from_reg): Likewise.
15973 (emit_push_insn): Likewise.
15974 (expand_assignment): Likewise.
15975 (store_field): Likewise.
15976 (expand_expr_real_1): Likewise.
15977 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15978 * lra-constraints.c (simplify_operand_subreg): Likewise.
15979 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15980 * gimple-ssa-store-merging.c: Likewise in block comment at start
15982 * tree-ssa-strlen.c: Include target.h.
15983 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15984 of SLOW_UNALIGNED_ACCESS.
15985 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15987 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15989 PR rtl-optimization/82185
15990 * expmed.c (emit_store_flag_int): Only test tem if it has been
15993 2017-09-12 Richard Biener <rguenther@suse.de>
15995 PR middle-end/82149
15996 * match.pd ((FTYPE) N CMP CST): Fix typo.
15998 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16000 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16002 (mips_near_type_p): Add 'short_call' attribute as a synonym
16004 * doc/extend.texi (short_call): Document new function attribute.
16006 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16009 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16010 assertion check that in the condition.
16011 (get_atomic_generic_size): Likewise. Before testing if parameter
16012 has pointer type, if it has array type, call for C++
16013 default_conversion to perform array-to-pointer conversion.
16015 2017-09-12 Richard Biener <rguenther@suse.de>
16017 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16018 for operations we cannot scalarize.
16020 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16022 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16023 vectors heap vectors. Clean up comments.
16024 Make visited_bbs a reference.
16025 (profitable_jump_thread_path): Make GC
16026 vectors heap vectors. Clean up comments.
16028 (convert_and_register_jump_thread_path): Make GC vectors heap
16030 (check_subpath_and_update_thread_path): Same. Clean up comments.
16031 Make visited_bbs a reference.
16032 (handle_phi): Abstract common code to to
16033 register_jump_thread_path_if_profitable.
16034 Rename VAR_BB to DEF_BB.
16036 Make GC vectors heap vectors.
16037 Make visited_bbs a reference.
16038 (handle_assignment): Same.
16039 (register_jump_thread_path_if_profitable): New.
16040 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16042 Make GC vectors heap vectors. Clean up comments.
16043 Make visited_bbs a reference.
16044 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16045 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16048 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16051 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16052 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16054 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16056 Revert r251800 and r251799.
16058 2017-09-11 Martin Jambor <mjambor@suse.cz>
16061 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16062 arguments in advance.
16063 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16064 use it to find predecessor edges.
16065 (naive_outof_ssa): Collect vector of predecessors.
16067 2017-09-08 Jason Merrill <jason@redhat.com>
16069 PR c++/70029 - ICE with ref-qualifier and -flto
16070 * langhooks.h (struct lang_hooks_for_types): Add
16071 copy_lang_qualifiers.
16072 * attribs.c (build_type_attribute_qual_variant): Use it.
16073 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16075 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16076 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16078 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16081 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16082 (*mulsi3_sp64): New instruction.
16083 (mulsi3): New expander.
16085 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16087 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16089 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16091 * sancov.c: Include memmodel.h.
16093 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16096 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16099 2017-09-07 Carl Love <cel@us.ibm.com>
16101 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16102 the sldi instruction.
16104 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16106 * sancov.c: Include tm_p.h.
16108 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16111 * output.h (switch_to_other_text_partition): New declaration.
16112 * varasm.c (switch_to_other_text_partition): New function.
16113 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16114 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16115 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16116 to the other text partition before emitting LCL label and switch back
16117 after emitting the word after it.
16119 2017-09-07 Richard Biener <rguenther@suse.de>
16121 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16122 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16123 critical edges here, after loop init.
16124 (pass_data_pre): Remove PROP_no_crit_edges flags.
16125 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16126 for valueization of call args to avoid leaking VN_TOP.
16127 (visit_use): Assert we do not visit default defs.
16128 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16129 Use error_mark_node to more easily detect leaking VN_TOP.
16130 All default-defs are varying, not VN_TOP. Mark them visited.
16131 (run_scc_vn): Make code match comment.
16133 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16135 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16136 OPTION_MASK_FLOAT128_KEYWORD.
16137 (POWERPC_MASKS): Likewise.
16138 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16139 support for the -mfloat128-type option, and make -mfloat128
16140 default on PowerPC Linux systems. Define or undefine
16141 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16142 Define __float128 to be __ieee128 if IEEE 128-bit support is
16143 enabled, or undefine it.
16144 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16145 Delete defining __FLOAT128_TYPE__.
16146 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16147 -mfloat128-type option and make -mfloat128 default on PowerPC
16149 (TARGET_FLOAT128_TYPE): Likewise.
16150 (-mfloat128-type): Likewise.
16151 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16152 Delete the -mfloat128-type option and make -mfloat128 default on
16153 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16154 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16155 128-bit floating point is enabled. Change tests from using
16156 -mfloat128-type to -mfloat128.
16157 (rs6000_mangle_type): Use the correct mangling for the __float128
16158 type even if normal long double is restricted to 64-bits.
16159 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16161 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16162 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16163 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16165 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16166 documentation for -mfloat128.
16168 2017-09-06 Olivier Hainque <hainque@adacore.com>
16170 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16172 2017-09-06 Wish Wu <wishwu007@gmail.com>
16173 Jakub Jelinek <jakub@redhat.com>
16175 * asan.c (initialize_sanitizer_builtins): Add
16176 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16177 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16178 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16179 BT_FN_VOID_UINT64_PTR variables.
16180 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16181 (BT_FN_VOID_UINT16_UINT16): Likewise.
16182 (BT_FN_VOID_UINT32_UINT32): Likewise.
16183 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16184 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16185 (BT_FN_VOID_UINT64_PTR): Likewise.
16186 * common.opt (flag_sanitize_coverage): New variable.
16187 (fsanitize-coverage=trace-pc): Remove.
16188 (fsanitize-coverage=): Add.
16189 * flag-types.h (enum sanitize_coverage_code): New enum.
16190 * fold-const.c (fold_range_test): Disable non-short-circuit
16191 optimization if flag_sanitize_coverage.
16192 (fold_truth_andor): Likewise.
16193 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16194 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16195 (coverage_sanitizer_opts): New array.
16196 (get_closest_sanitizer_option): Add OPTS argument, handle also
16197 OPT_fsanitize_coverage_.
16198 (parse_sanitizer_options): Adjusted to also handle
16199 OPT_fsanitize_coverage_.
16200 (common_handle_option): Add OPT_fsanitize_coverage_.
16201 * sancov.c (instrument_comparison, instrument_switch): New function.
16202 (sancov_pass): Add trace-cmp support.
16203 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16204 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16205 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16206 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16207 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16208 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16209 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16210 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16211 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16212 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16214 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16216 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16217 error. Only quit immediately if parsing is complete.
16218 (BEGIN): Initialize fatal_err and parse_done.
16219 (begin fpu, end fpu): Check number of arguments.
16220 (begin arch, end arch): Likewise.
16221 (begin cpu, end cpu): Likewise.
16222 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16223 (optalias): Likewise.
16225 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16227 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16228 * config/arm/arm-isa.h: Delete. Move definitions to ...
16229 * arm-cpus.in: ... here. Use new feature and fgroup values.
16230 * config/arm/arm.c (arm_option_override): Use lower case for feature
16232 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16233 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16234 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16236 (print_isa_bits_for): New function.
16237 (gen_isa): New function.
16238 (gen_comm_data): Use print_isa_bits_for.
16239 (define feature): New keyword.
16240 (define fgroup): New keyword.
16241 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16242 (arm-isa.h): Add rule to generate file.
16243 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16244 case for feature bit names.
16246 2017-09-06 Richard Biener <rguenther@suse.de>
16248 * tree-ssa-pre.c (NECESSARY): Remove.
16249 (create_expression_by_pieces): Do not touch pass-local flags.
16250 (insert_into_preds_of_block): Likewise.
16251 (do_pre_regular_insertion): Likewise.
16252 (eliminate_insert): Likewise.
16253 (eliminate_dom_walker::before_dom_children): Likewise.
16254 (fini_eliminate): Do not look at inserted_exprs.
16255 (mark_operand_necessary): Remove.
16256 (remove_dead_inserted_code): Replace with simple work-list
16257 algorithm based on inserted_exprs and SSA uses.
16258 (pass_pre::execute): Re-order fini_eliminate and
16259 remove_dead_inserted_code.
16261 2017-09-06 Olivier Hainque <hainque@adacore.com>
16263 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16264 for VxWorks 7. Adjust surrounding comments.
16266 2017-09-06 Richard Biener <rguenther@suse.de>
16268 * gimple-ssa-strength-reduction.c
16269 (find_candidates_dom_walker::before_dom_children): Also allow
16272 2017-09-06 Richard Biener <rguenther@suse.de>
16274 PR tree-optimization/82108
16275 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16276 for gap in the non-permutation SLP case.
16278 2017-09-06 Martin Jambor <mjambor@suse.cz>
16280 PR tree-optimization/82078
16281 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16282 add_access_to_work_queue...
16283 (build_accesses_from_assign): ...here.
16284 (propagate_all_subaccesses): Make sure racc is the group
16285 representative, if there is one.
16287 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16289 PR middle-end/82095
16290 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16293 2017-09-06 Richard Biener <rguenther@suse.de>
16295 * gimple-ssa-strength-reduction.c
16296 (find_candidates_dom_walker::before_doom_children): Use a
16297 type and not a mode check.
16299 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16302 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16304 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16306 2017-09-05 Jeff Law <law@redhat.com>
16308 PR tree-optimization/64910
16309 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16310 cases where we have 3 or more operands.
16312 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16314 PR middle-end/81768
16315 * omp-low.c (lower_omp_for): Recompute tree invariant if
16316 gimple_omp_for_initial/final is ADDR_EXPR.
16318 PR middle-end/81768
16319 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16320 into gimple val before gimplification fo the COND_EXPR.
16322 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16324 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16325 REGION_COPY argument.
16326 (thread_through_all_blocks): Remove unused argument to
16327 duplicate_thread_path.
16329 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16330 Alan Hayward <alan.hayward@arm.com>
16331 David Sherwood <david.sherwood@arm.com>
16333 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16334 Take a scalar_mode rather than a machine_mode.
16335 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16336 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16337 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16338 (aarch64_gen_adjusted_ldpstp): Likewise.
16339 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16341 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16342 Alan Hayward <alan.hayward@arm.com>
16343 David Sherwood <david.sherwood@arm.com>
16345 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16346 Take a scalar_int_mode instead of a machine_mode.
16347 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16348 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16349 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16350 (aarch64_simd_attr_length_rglist): Delete.
16351 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16352 a scalar_int_mode instead of a machine_mode.
16353 (aarch64_add_offset): Likewise.
16354 (aarch64_internal_mov_immediate): Likewise
16355 (aarch64_add_constant_internal): Likewise.
16356 (aarch64_add_constant): Likewise.
16357 (aarch64_movw_imm): Likewise.
16358 (aarch64_rtx_arith_op_extract_p): Likewise.
16359 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16360 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16361 Remove assert that the mode isn't a vector.
16362 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16363 (aarch64_expand_mov_immediate): Update calls after above changes.
16364 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16365 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16366 (aarch64_move_imm): Likewise.
16367 (aarch64_can_const_movi_rtx_p): Likewise.
16368 (aarch64_strip_extend): Likewise.
16369 (aarch64_extr_rtx_p): Likewise.
16370 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16371 a CONST_INT when the mode parameter is VOIDmode.
16372 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16374 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16376 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16377 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16378 (bitwise_type_for_mode): Update accordingly.
16380 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16382 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16383 * stor-layout.c (mode_for_size_tree): Likewise.
16384 (mode_for_array): Update accordingly.
16385 (layout_decl): Likewise.
16386 (compute_record_mode): Likewise. Only set the mode once.
16388 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16390 * target.def (get_mask_mode): Change return type to opt_mode.
16392 * doc/tm.texi: Regenerate.
16393 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16394 * targhooks.c (default_get_mask_mode): Likewise.
16395 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16396 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16397 targetm.get_mask_mode.
16398 * tree.c (build_truth_vector_type): Likewise.
16400 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16402 * machmode.h (mode_for_vector): Return an opt_mode.
16403 * stor-layout.c (mode_for_vector): Likewise.
16404 (mode_for_int_vector): Update accordingly.
16405 (layout_type): Likewise.
16406 * config/i386/i386.c (emit_memmov): Likewise.
16407 (ix86_expand_set_or_movmem): Likewise.
16408 (ix86_expand_vector_init): Likewise.
16409 (ix86_get_mask_mode): Likewise.
16410 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16412 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16413 * expmed.c (extract_bit_field_1): Likewise.
16414 * expr.c (expand_expr_real_2): Likewise.
16415 * optabs-query.c (can_vec_perm_p): Likewise.
16416 (can_vec_mask_load_store_p): Likewise.
16417 * optabs.c (expand_vec_perm): Likewise.
16418 * targhooks.c (default_get_mask_mode): Likewise.
16419 * tree-vect-stmts.c (vectorizable_store): Likewise.
16420 (vectorizable_load): Likewise.
16421 (get_vectype_for_scalar_type_and_size): Likewise.
16423 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16425 * machmode.h (mode_for_int_vector): New function.
16426 * stor-layout.c (mode_for_int_vector): Likewise.
16427 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16428 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16429 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16430 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16431 (s390_expand_vcond): Likewise.
16433 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16435 * machmode.h (opt_machine_mode): New type.
16436 (opt_mode<T>): Allow construction from anything that can be
16438 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16439 (mode_for_size): Return an opt_machine_mode.
16440 * stor-layout.c (mode_for_size): Likewise.
16441 (mode_for_size_tree): Update call accordingly.
16442 (bitwise_mode_for_mode): Likewise.
16443 (make_fract_type): Likewise.
16444 (make_accum_type): Likewise.
16445 * caller-save.c (replace_reg_with_saved_mem): Update call
16447 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16448 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16449 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16450 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16451 * expmed.c (extract_bit_field_1): Likewise.
16452 * reload.c (get_secondary_mem): Likewise.
16453 * varasm.c (assemble_integer): Likewise.
16454 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16457 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16459 * machmode.h (decimal_float_mode_for_size): New function.
16460 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16461 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16462 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16463 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16464 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16465 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16467 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16469 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16470 (get_builtin_sync_mode): Likewise.
16471 (expand_ifn_atomic_compare_exchange): Likewise.
16472 (expand_builtin_atomic_clear): Likewise.
16473 (expand_builtin_atomic_test_and_set): Likewise.
16474 (fold_builtin_atomic_always_lock_free): Likewise.
16475 * calls.c (compute_argument_addresses): Likewise.
16476 (emit_library_call_value_1): Likewise.
16477 (store_one_arg): Likewise.
16478 * combine.c (combine_instructions): Likewise.
16479 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16480 * config/arm/arm.c (arm_function_value): Likewise.
16481 (aapcs_allocate_return_reg): Likewise.
16482 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16483 * config/i386/i386.c (construct_container): Likewise.
16484 (ix86_gimplify_va_arg): Likewise.
16485 (ix86_expand_sse_cmp): Likewise.
16486 (emit_memmov): Likewise.
16487 (emit_memset): Likewise.
16488 (expand_small_movmem_or_setmem): Likewise.
16489 (ix86_expand_pextr): Likewise.
16490 (ix86_expand_pinsr): Likewise.
16491 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16492 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16494 * config/mips/mips.c (mips_function_value_1) Likewise.
16495 (mips_block_move_straight): Likewise.
16496 (mips_expand_ins_as_unaligned_store): Likewise.
16497 * config/powerpcspe/powerpcspe.c
16498 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16499 (rs6000_darwin64_record_arg_flush): Likewise.
16500 * config/rs6000/rs6000.c
16501 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16502 (rs6000_darwin64_record_arg_flush): Likewise.
16503 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16504 (sparc_function_value_1): Likewise.
16505 * config/spu/spu.c (adjust_operand): Likewise.
16506 (spu_emit_branch_or_set): Likewise.
16507 (arith_immediate_p): Likewise.
16508 * emit-rtl.c (gen_lowpart_common): Likewise.
16509 * expr.c (expand_expr_real_1): Likewise.
16510 * function.c (assign_parm_setup_block): Likewise.
16511 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16512 * reload1.c (alter_reg): Likewise.
16513 * stor-layout.c (mode_for_vector): Likewise.
16514 (layout_type): Likewise.
16516 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16518 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16519 (spu_convert_move): Likewise.
16520 * lower-subreg.c (resolve_simple_move): Likewise.
16522 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16525 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16526 define_insn to a define_expand.
16527 (altivec_vsum2sws_direct): New define_insn.
16528 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16530 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16532 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16535 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16540 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16541 realign stack if stack alignment needed is less than incoming
16544 2017-09-05 Marek Polacek <polacek@redhat.com>
16547 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16550 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16552 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16554 2017-09-05 Richard Biener <rguenther@suse.de>
16556 PR tree-optimization/82084
16557 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16559 2017-09-05 Richard Biener <rguenther@suse.de>
16561 PR tree-optimization/82102
16562 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16564 2017-09-05 Martin Liska <mliska@suse.cz>
16566 PR tree-optimization/82032
16567 * tree-cfg.c (generate_range_test): New function.
16568 * tree-cfg.h (generate_range_test): Declared here.
16569 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16570 (cleanup_control_expr_graph): Use it.
16571 * tree-switch-conversion.c (try_switch_expansion): Remove
16573 (emit_case_nodes): Use generate_range_test.
16575 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16578 * config/i386/i386.md (*<btsc><mode>_mask): Add
16579 TARGET_USE_BT to insn constraint.
16580 (*btr<mode>_mask): Ditto.
16582 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16584 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16585 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16587 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16590 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16591 TARGET_NEON and TARGET_IWMMXT.
16592 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16593 TARGET_NEON and TARGET_IWMMXT.
16594 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16596 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16598 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16599 (ix86_rewrite_tls_address): Ditto.
16600 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16601 (ix86_rewrite_tls_address_1): Ditto.
16602 (ix86_rewrite_tls_address): Ditto.
16603 * config/i386/predicates.md (tls_address_pattern): New predicate.
16604 * config/i386/i386.md (TLS address splitter): New splitter.
16606 2017-09-04 Richard Biener <rguenther@suse.de>
16608 PR tree-optimization/82084
16609 * fold-const.h (can_native_encode_string_p): Declare.
16610 * fold-const.c (can_native_encode_string_p): Factor out from ...
16611 (native_encode_string): ... here.
16612 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16613 vectorizing stores from constants we later cannot handle.
16615 2017-09-04 Marek Polacek <polacek@redhat.com>
16618 * doc/invoke.texi: Update -Wtautological-compare documentation.
16620 2017-09-04 Jeff Law <law@redhat.com>
16622 PR tree-optimization/64910
16623 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16624 swap the first and last operand if the last is a constant.
16626 2017-09-04 Marek Polacek <polacek@redhat.com>
16629 * convert.c (do_narrow): When sanitizing signed integer overflows,
16630 bail out for signed types.
16631 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16633 2017-09-04 Richard Biener <rguenther@suse.de>
16635 PR tree-optimization/82060
16636 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16637 Move devirtualization after stmt folding and before EH/AB/noreturn
16638 cleanup to get the stmt refs canonicalized. Use a bool instead
16639 of gimple_modified_p since that doesn't work for NOPs. Schedule
16640 NOPs generated by folding for removal.
16642 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16643 Alan Hayward <alan.hayward@arm.com>
16644 David Sherwood <david.sherwood@arm.com>
16646 * coretypes.h (pad_direction): New enum.
16647 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16648 (FUNCTION_ARG_PADDING): Likewise.
16649 * target.def (function_arg_padding): New hook.
16650 * targhooks.h (default_function_arg_padding): Declare.
16651 * targhooks.c (default_function_arg_padding): New function.
16652 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16653 (TARGET_FUNCTION_ARG_PADDING): ...this.
16654 * doc/tm.texi: Regenerate.
16655 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16656 instead of direction.
16657 (compute_argument_addresses): Likewise.
16658 (load_register_parameters): Likewise.
16659 (emit_library_call_value_1): Likewise.
16660 (store_one_arg): Use targetm.calls.function_arg_padding instead
16661 of FUNCTION_ARG_PADDING.
16662 (must_pass_in_stack_var_size_or_pad): Likewise.
16663 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16664 (emit_group_store): Likewise.
16665 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16666 instead of FUNCTION_ARG_PADDING.
16667 (emit_push_insn): Likewise, and propagate enum change throughout
16669 * function.h (direction): Delete.
16670 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16672 * function.c (assign_parm_find_stack_rtl): Likewise.
16673 (assign_parm_setup_block_p): Likewise.
16674 (assign_parm_setup_block): Likewise.
16675 (gimplify_parameters): Likewise.
16676 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16677 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16679 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16680 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16681 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16682 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16683 (aarch64_function_arg_padding): ...this new function.
16684 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16685 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16686 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16687 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16688 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16689 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16690 (arm_pad_arg_upward): Replace with...
16691 (arm_function_arg_padding): ...this new function.
16692 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16694 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16695 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16696 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16697 (ia64_hpux_function_arg_padding): Replace with...
16698 (ia64_function_arg_padding): ...this new function. Use pad_direction
16699 instead of direction. Check for TARGET_HPUX.
16700 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16701 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16702 (iq2000_function_arg_padding): New function.
16703 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16704 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16705 (mips_function_arg_padding): ...this new function.
16706 (mips_pad_reg_upward): Update accordingly.
16707 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16708 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16709 targetm.calls.function_arg_padding.
16710 (FUNCTION_ARG_PADDING): Delete.
16711 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16712 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16713 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16714 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16715 (nios2_block_reg_padding): Return pad_direction instead of direction.
16716 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16717 instead of direction.
16718 (nios2_function_arg_padding): Likewise. Make static.
16719 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16720 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16721 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16722 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16723 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16724 (pa_function_arg_padding): Make static. Return pad_direction instead
16726 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16727 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16728 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16729 instead of direction. Use targetm.calls.function_arg_padding.
16730 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16731 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16732 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16733 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16734 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16736 (function_arg_padding): Rename to...
16737 (rs6000_function_arg_padding): ...this. Make static. Return
16738 pad_direction instead of direction.
16739 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16740 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16741 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16742 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16743 instead of direction. Use targetm.calls.function_arg_padding.
16744 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16745 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16746 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16747 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16748 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16749 (function_arg_padding): Rename to...
16750 (rs6000_function_arg_padding): ...this. Make static. Return
16751 pad_direction instead of direction.
16752 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16753 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16754 * config/s390/s390.c (s390_function_arg_padding): New function.
16755 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16756 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16757 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16758 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16759 (function_arg_padding): Rename to...
16760 (sparc_function_arg_padding): ...this. Make static. Return
16761 pad_direction instead of direction.
16762 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16763 * config/spu/spu.c (spu_function_arg_padding): New function.
16764 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16765 * system.h (FUNCTION_ARG_PADDING): Poison.
16767 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16768 Alan Hayward <alan.hayward@arm.com>
16769 David Sherwood <david.sherwood@arm.com>
16771 * target.def (modes_tieable_p): New hook.
16772 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16773 (TARGET_MODES_TIEABLE_P): ...this.
16774 * doc/tm.texi.in: Regenerate.
16775 * hooks.h (hook_bool_mode_mode_true): Declare.
16776 * hooks.c (hook_bool_mode_mode_true): New function.
16777 * combine.c (subst): Use targetm.modes_tieable_p instead of
16779 * dse.c (find_shift_sequence): Likewise.
16780 * expmed.c (extract_low_bits): Likewise.
16781 * lower-subreg.c: Include target.h.
16782 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16784 * rtlanal.c (rtx_cost): Likewise.
16785 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16786 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16787 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16788 (TARGET_MODES_TIEABLE_P): Redefine.
16789 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16790 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16791 (TARGET_MODES_TIEABLE_P): Redefine.
16792 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16793 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16794 (arc_modes_tieable_p): New function.
16795 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16796 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16797 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16798 (arm_modes_tieable_p): Make static.
16799 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16800 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16801 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16802 (TARGET_MODES_TIEABLE_P): Redefine.
16803 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16804 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16805 (TARGET_MODES_TIEABLE_P): Redefine.
16806 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16807 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16808 (cr16_modes_tieable_p): New function.
16809 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16810 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16811 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16812 (TRULY_NOOP_TRUNCATION): Update comment.
16813 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16814 (TRULY_NOOP_TRUNCATION): Update comment.
16815 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16816 (frv_modes_tieable_p): New function.
16817 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16818 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16819 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16820 (TARGET_MODES_TIEABLE_P): Redefine.
16821 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16822 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16823 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16824 (TARGET_MODES_TIEABLE_P): Redefine.
16825 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16826 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16827 (ia64_modes_tieable_p): New function.
16828 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16829 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16830 (iq2000_modes_tieable_p): New function.
16831 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16832 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16833 (lm32_modes_tieable_p): New function.
16834 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16835 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16836 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16837 (TARGET_MODES_TIEABLE_P): Redefine.
16838 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16839 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16840 (m32r_modes_tieable_p): New function.
16841 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16842 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16843 (m68k_modes_tieable_p): New function.
16844 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16845 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16846 (mcore_modes_tieable_p): New function.
16847 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16848 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16850 (TARGET_MODES_TIEABLE_P): Redefine.
16851 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16852 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16853 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16854 (TARGET_MODES_TIEABLE_P): Redefine.
16855 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16856 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16857 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16858 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16859 (mn10300_modes_tieable_p): ...this and make static.
16860 (TARGET_MODES_TIEABLE_P): Redefine.
16861 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16862 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16863 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16864 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16865 (msp430_modes_tieable_p): Make static.
16866 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16867 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16868 (TARGET_MODES_TIEABLE_P): Redefine.
16869 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16870 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16871 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16872 (TARGET_MODES_TIEABLE_P): Redefine.
16873 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16874 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16875 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16876 (TARGET_MODES_TIEABLE_P): Redefine.
16877 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16878 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16879 (pdp11_modes_tieable_p): New function.
16880 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16881 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16882 (rs6000_modes_tieable_p): New function.
16883 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16884 * config/powerpcspe/powerpcspe.md: Update comment.
16885 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16886 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16887 (TARGET_MODES_TIEABLE_P): Redefine.
16888 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16889 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16890 (rl78_modes_tieable_p): New function.
16891 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16892 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16893 (rs6000_modes_tieable_p): New function.
16894 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16895 * config/rs6000/rs6000.md: Update comment.
16896 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16897 * config/rx/rx.c (rx_modes_tieable_p): New function.
16898 (TARGET_MODES_TIEABLE_P): Redefine.
16899 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16900 * config/s390/s390.c (s390_modes_tieable_p): New function.
16901 (TARGET_MODES_TIEABLE_P): Redefine.
16902 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16903 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16904 (sh_modes_tieable_p): New function.
16905 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16906 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16907 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16908 (sparc_modes_tieable_p): Make static.
16909 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16910 * config/spu/spu.c (spu_modes_tieable_p): New function.
16911 (TARGET_MODES_TIEABLE_P): Redefine.
16912 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16913 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16914 (TARGET_MODES_TIEABLE_P): Redefine.
16915 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16916 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16917 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16918 * config/v850/v850.c (v850_modes_tieable_p): New function.
16919 (TARGET_MODES_TIEABLE_P): Redefine.
16920 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16921 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16922 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16923 (visium_modes_tieable_p): New function.
16924 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16925 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16926 (xtensa_modes_tieable_p): New function.
16927 * system.h (MODES_TIEABLE_P): Poison.
16929 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16930 Alan Hayward <alan.hayward@arm.com>
16931 David Sherwood <david.sherwood@arm.com>
16933 * target.def (hard_regno_mode_ok): New hook.
16934 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16935 (TARGET_HARD_REGNO_MODE_OK): ...this.
16936 * doc/tm.texi.in: Regenerate.
16937 * hooks.h (hook_bool_uint_mode_true): Declare.
16938 * hooks.c (hook_bool_uint_mode_true): New function.
16939 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16940 HARD_REGNO_MODE_OK.
16941 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16942 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16943 instead of HARD_REGNO_MODE_OK.
16944 * caller-save.c: Include target.h.
16945 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16946 HARD_REGNO_MODE_OK.
16947 * combine.c (can_combine_p): Likewise.
16948 (combinable_i3pat): Likewise.
16949 (can_change_dest_mode): Likewise.
16950 * expr.c (init_expr_target): Likewise.
16951 (convert_move): Likewise.
16952 (convert_modes): Likewise.
16953 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16954 (setup_prohibited_mode_move_regs): Likewise.
16955 * ira.h (target_ira): Likewise.
16956 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16957 * lra-constraints.c (process_alt_operands): Likewise.
16958 (split_reg): Likewise.
16959 * recog.c (peep2_find_free_register): Likewise.
16960 * ree.c (combine_reaching_defs): Likewise.
16961 * regcprop.c (maybe_mode_change): Likewise.
16962 * reginfo.c (init_reg_sets_1): Likewise.
16963 (choose_hard_reg_mode): Likewise.
16964 (simplifiable_subregs): Likewise.
16965 * regrename.c (check_new_reg_p): Likewise.
16966 * reload.c (find_valid_class): Likewise.
16967 (find_valid_class_1): Likewise.
16968 (reload_inner_reg_of_subreg): Likewise.
16969 (push_reload): Likewise.
16970 (combine_reloads): Likewise.
16971 (find_dummy_reload): Likewise.
16972 (find_reloads): Likewise.
16973 * reload1.c (find_reg): Likewise.
16974 (set_reload_reg): Likewise.
16975 (allocate_reload_reg): Likewise.
16976 (choose_reload_regs): Likewise.
16977 (reload_adjust_reg_for_temp): Likewise.
16978 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16979 (simplify_subreg_regno): Likewise.
16980 * sel-sched.c (init_regs_for_mode): Likewise.
16981 * varasm.c (make_decl_rtl): Likewise.
16982 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16983 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16984 HARD_REGNO_MODE_OK.
16985 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16986 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16987 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16988 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16989 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16990 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16991 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16992 (arc_mode_class): Delete.
16993 (HARD_REGNO_MODE_OK): Delete.
16994 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16995 (arc_hard_regno_mode_ok): Rename old array to...
16996 (arc_hard_regno_mode_ok_modes): ...this.
16997 (arc_conditional_register_usage): Update accordingly.
16998 (arc_mode_class): Make static.
16999 (arc_hard_regno_mode_ok): New function.
17000 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17001 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17002 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17003 (arm_hard_regno_mode_ok): Make static.
17004 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17005 HARD_REGNO_MODE_OK.
17006 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17007 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17008 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17010 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17011 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17012 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17013 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17014 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17015 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17016 * config/bfin/predicates.md (valid_reg_operand): Use
17017 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17018 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17019 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17020 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17021 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17022 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17023 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17024 (cr16_hard_regno_mode_ok): Make static and return a bool.
17025 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17026 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17027 (cris_hard_regno_mode_ok): New function.
17028 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17029 (epiphany_mode_class): Delete.
17030 (HARD_REGNO_MODE_OK): Delete.
17031 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17032 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17033 (hard_regno_mode_ok): Rename to...
17034 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17035 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17036 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17037 HARD_REGNO_MODE_OK.
17038 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17039 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17040 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17041 (frv_hard_regno_mode_ok): Make static and return a bool.
17042 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17043 HARD_REGNO_MODE_OK.
17044 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17045 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17046 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17047 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17049 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17050 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17051 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17052 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17054 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17055 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17056 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17057 (ia64_hard_regno_mode_ok): New function.
17058 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17059 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17060 (iq2000_hard_regno_mode_ok): New function.
17061 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17062 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17063 (lm32_hard_regno_mode_ok): New function.
17064 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17065 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17066 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17067 instead of HARD_REGNO_MODE_OK.
17068 (m32c_hard_regno_ok): Rename to...
17069 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17070 (m32c_cannot_change_mode_class): Update accordingly.
17071 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17072 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17073 (m32r_mode_class): Delete.
17074 (HARD_REGNO_MODE_OK): Delete.
17075 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17076 (m32r_hard_regno_mode_ok): Rename to...
17077 (m32r_hard_regno_modes): ...this.
17078 (m32r_mode_class): Make static.
17079 (m32r_hard_regno_mode_ok): New function.
17080 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17081 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17082 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17083 (m68k_hard_regno_mode_ok): Make static.
17084 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17085 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17086 (mcore_hard_regno_mode_ok): New function.
17087 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17088 (HARD_REGNO_MODE_OK): Delete.
17089 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17091 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17092 (microblaze_hard_regno_mode_ok): New function.
17093 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17094 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17095 (mips_hard_regno_mode_ok): Delete.
17096 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17097 (mips_hard_regno_mode_ok_p): ...this and make static.
17098 (mips_hard_regno_mode_ok_p): Rename to...
17099 (mips_hard_regno_mode_ok_uncached): ...this.
17100 (mips_hard_regno_mode_ok): New function.
17101 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17102 of HARD_REGNO_MODE_OK.
17103 (mips_option_override): Update after above name changes.
17104 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17105 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17106 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17107 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17108 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17109 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17110 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17111 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17112 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17113 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17114 (msp430_hard_regno_mode_ok): Make static and return a bool.
17115 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17116 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17117 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17119 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17120 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17121 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17122 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17123 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17124 (PA_HARD_REGNO_MODE_OK): ...this
17125 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17126 (PA_HARD_REGNO_MODE_OK): ...this.
17127 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17128 (pa_hard_regno_mode_ok): New function.
17129 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17130 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17131 (pdp11_hard_regno_mode_ok): New function.
17132 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17133 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17135 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17137 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17138 (rs6000_hard_regno_mode_ok): Rename to...
17139 (rs6000_hard_regno_mode_ok_uncached): ...this.
17140 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17141 (rs6000_hard_regno_mode_ok): New function.
17142 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17143 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17144 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17145 (riscv_hard_regno_mode_ok): ...this and make static.
17146 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17147 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17148 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17149 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17150 (rl78_hard_regno_mode_ok): Make static and return bool.
17151 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17152 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17154 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17155 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17156 (rs6000_hard_regno_mode_ok): Rename to...
17157 (rs6000_hard_regno_mode_ok_uncached): ...this.
17158 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17159 (rs6000_hard_regno_mode_ok): New function.
17160 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17161 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17162 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17163 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17164 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17165 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17166 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17167 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17168 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17169 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17170 (sh_hard_regno_mode_ok): Make static.
17171 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17172 instead of HARD_REGNO_MODE_OK.
17173 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17174 (sparc_mode_class): Delete.
17175 (HARD_REGNO_MODE_OK): Delete.
17176 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17177 (hard_regno_mode_classes): Make static.
17178 (sparc_mode_class): Likewise.
17179 (sparc_hard_regno_mode_ok): New function.
17180 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17181 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17182 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17184 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17185 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17186 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17187 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17188 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17189 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17190 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17191 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17192 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17193 (visium_hard_regno_mode_ok): New function.
17194 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17195 instead of HARD_REGNO_MODE_OK.
17196 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17197 (HARD_REGNO_MODE_OK): Delete.
17198 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17199 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17200 (xtensa_option_override): Update accordingly.
17201 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17202 (xtensa_hard_regno_mode_ok): New function.
17203 * system.h (HARD_REGNO_MODE_OK): Poison.
17205 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17206 Alan Hayward <alan.hayward@arm.com>
17207 David Sherwood <david.sherwood@arm.com>
17209 * target.def (hard_regno_call_part_clobbered): New hook.
17210 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17211 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17212 * doc/tm.texi: Regenerate.
17213 * hooks.h (hook_bool_uint_mode_false): Declare.
17214 * hooks.c (hook_bool_uint_mode_false): New function.
17215 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17216 * cselib.c (cselib_process_insn): Use
17217 targetm.hard_regno_call_part_clobbered instead of
17218 HARD_REGNO_CALL_PART_CLOBBERED.
17219 * ira-conflicts.c (ira_build_conflicts): Likewise.
17220 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17221 * lra-constraints.c (need_for_call_save_p): Likewise.
17222 * lra-lives.c: Include target.h.
17223 (check_pseudos_live_through_calls): Use
17224 targetm.hard_regno_call_part_clobbered instead of
17225 HARD_REGNO_CALL_PART_CLOBBERED.
17226 * regcprop.c: Include target.h.
17227 (copyprop_hardreg_forward_1): Use
17228 targetm.hard_regno_call_part_clobbered instead of
17229 HARD_REGNO_CALL_PART_CLOBBERED.
17230 * reginfo.c (choose_hard_reg_mode): Likewise.
17231 * regrename.c (check_new_reg_p): Likewise.
17232 * reload.c (find_equiv_reg): Likewise.
17233 * reload1.c (emit_reload_insns): Likewise.
17234 * sched-deps.c (deps_analyze_insn): Likewise.
17235 * sel-sched.c (init_regs_for_mode): Likewise.
17236 (mark_unavailable_hard_regs): Likewise.
17237 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17238 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17239 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17241 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17242 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17243 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17245 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17247 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17248 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17249 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17251 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17252 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17253 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17255 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17256 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17258 * config/powerpcspe/powerpcspe.c
17259 (rs6000_hard_regno_call_part_clobbered): New function.
17260 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17261 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17262 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17264 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17265 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17266 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17268 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17269 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17270 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17272 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17273 Alan Hayward <alan.hayward@arm.com>
17274 David Sherwood <david.sherwood@arm.com>
17276 * rtl.h (subreg_memory_offset): Declare.
17277 * emit-rtl.c (subreg_memory_offset): New function.
17278 * expmed.c (store_bit_field_1): Use it.
17279 * expr.c (undefined_operand_subword_p): Likewise.
17280 * simplify-rtx.c (simplify_subreg): Likewise.
17282 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17284 PR rtl-optimization/57448
17287 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17288 using atomic_load pattern.
17289 (expand_atomic_store): Likewise.
17291 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17294 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17295 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17296 handling. Use replace_call_with_value with NULL instead of
17297 gsi_replace, unlink_stmt_vdef and release_defs.
17299 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17302 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17304 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17307 * rtl.h (emit_library_call_value_1): Declare.
17308 (emit_library_call): Replace declaration with a series of overloads.
17309 Remove the parameter count argument.
17310 (emit_library_call_value): Likewise.
17311 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17312 with an "rtx_mode_t *".
17313 (emit_library_call_value): Delete.
17314 (emit_library_call): Likewise.
17315 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17316 (asan_emit_allocas_unpoison): Likewise.
17317 * builtins.c (expand_builtin_powi): Likewise.
17318 (expand_asan_emit_allocas_unpoison): Likewise.
17319 * cfgexpand.c (expand_main_function): Likewise.
17320 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17321 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17322 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17323 * config/arm/arm.c (arm_trampoline_init): Likewise.
17324 (arm_call_tls_get_addr): Likewise.
17325 (arm_expand_divmod_libfunc): Likewise.
17326 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17327 (smulsi3_highpart): Likewise.
17328 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17329 (c6x_expand_compare): Likewise.
17330 (c6x_expand_movmem): Likewise.
17331 * config/frv/frv.c (frv_trampoline_init): Likewise.
17332 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17333 (ix86_expand_divmod_libfunc): Likewise.
17334 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17335 (ia64_expand_compare): Likewise.
17336 (ia64_profile_hook): Likewise.
17337 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17338 (nonlocal_goto): Likewise.
17339 (restore_stack_nonlocal): Likewise.
17340 * config/m32r/m32r.c (block_move_call): Likewise.
17341 (m32r_trampoline_init): Likewise.
17342 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17343 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17344 (m68k_call_m68k_read_tp): Likewise.
17345 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17346 (microblaze_expand_divide): Likewise.
17347 * config/mips/mips.h (mips_args): Likewise.
17348 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17349 (MIPS_ICACHE_SYNC): Likewise.
17350 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17351 (nios2_trampoline_init): Likewise.
17352 * config/pa/pa.c (hppa_tls_call): Likewise.
17353 (pa_trampoline_init): Likewise.
17354 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17355 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17356 (expand_strn_compare): Likewise.
17357 (rs6000_generate_compare): Likewise.
17358 (rs6000_expand_float128_convert): Likewise.
17359 (output_profile_hook): Likewise.
17360 (rs6000_trampoline_init): Likewise.
17361 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17362 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17363 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17364 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17365 (rs6000_generate_compare): Likewise.
17366 (rs6000_expand_float128_convert): Likewise.
17367 (output_profile_hook): Likewise.
17368 (rs6000_trampoline_init): Likewise.
17369 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17370 * config/sh/sh.c (sh_trampoline_init): Likewise.
17371 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17372 (sparc_emit_float_lib_cmp): Likewise.
17373 (sparc32_initialize_trampoline): Likewise.
17374 (sparc64_initialize_trampoline): Likewise.
17375 (sparc_profile_hook): Likewise.
17376 * config/spu/spu.c (ea_load_store): Likewise.
17377 * config/spu/spu.md (floatunssidf2): Likewise.
17378 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17379 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17380 * config/visium/visium.c (expand_block_move_4): Likewise.
17381 (expand_block_move_2): Likewise.
17382 (expand_block_move_1): Likewise.
17383 (expand_block_set_4): Likewise.
17384 (expand_block_set_2): Likewise.
17385 (expand_block_set_1): Likewise.
17386 (visium_trampoline_init): Likewise.
17387 (visium_profile_hook): Likewise.
17388 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17389 (xtensa_setup_frame_addresses): Likewise.
17390 (xtensa_trampoline_init): Likewise.
17391 * except.c (sjlj_emit_function_enter): Likewise.
17392 (sjlj_emit_function_exit): Likewise.
17393 * explow.c (allocate_dynamic_stack_space): Likewise.
17394 (probe_stack_range): Likewise.
17395 * expr.c (convert_mode_scalar): Likewise.
17396 * optabs.c (expand_binop): Likewise.
17397 (expand_twoval_binop_libfunc): Likewise.
17398 (expand_unop): Likewise.
17399 (prepare_cmp_insn): Likewise.
17400 (prepare_float_lib_cmp): Likewise.
17401 (expand_float): Likewise.
17402 (expand_fix): Likewise.
17403 (expand_fixed_convert): Likewise.
17404 (maybe_emit_sync_lock_test_and_set): Likewise.
17405 (expand_atomic_compare_and_swap): Likewise.
17406 (expand_mem_thread_fence): Likewise.
17407 (expand_atomic_fetch_op): Likewise.
17409 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17411 * doc/generic.texi (OpenACC): Adjust URL.
17412 * doc/invoke.texi (C Dialect Options): Ditto.
17414 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17416 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17417 predicate for operand 1. Add (m,<S>) constraint.
17418 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17419 Prevent memory operand 1 with register operand 2.
17421 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17423 PR rtl-optimization/82024
17424 * combine.c (try_combine): If the combination result is a PARALLEL,
17425 and we only need to retain the SET in there that would be placed
17426 at I2, check that we can place that at I3 instead, before doing so.
17428 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17431 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17433 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17434 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17437 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17438 Jeff Law <law@redhat.com>
17440 * varasm.c (bss_initializer_p): Do not put constants into .bss
17441 (categorize_decl_for_section): Handle bss_initializer_p returning
17442 false when DECL_INITIAL is NULL.
17444 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17447 * config/s390/s390.c (s390_can_inline_p): New function.
17449 2017-09-01 Jeff Law <law@redhat.com>
17451 PR tree-optimization/82052
17452 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17453 Always initialize the returned slot after a hash table miss
17454 when INSERT is true.
17456 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17458 * config/s390/s390.md (mem_signal_fence): Remove.
17459 * doc/md.texi (mem_signal_fence): Remove.
17460 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17462 * target-insns.def (mem_signal_fence): Remove.
17464 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17467 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17470 * asan.c (create_odr_indicator): Strip name encoding from assembler
17471 name before appending it after __odr_asan_.
17473 2017-09-01 Martin Liska <mliska@suse.cz>
17475 PR tree-optimization/82059
17476 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17477 frequency only when an edge is redirected.
17479 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17481 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17482 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17483 (arc_conditional_register_usage): Remove ARC600 lp_count
17485 (arc_file_start): Emit Tag_ARC_CPU_variation.
17486 (arc_can_use_doloop_p): New conditions to use ZOLs.
17487 (hwloop_fail): New function.
17488 (hwloop_optimize): Likewise.
17489 (hwloop_pattern_reg): Likewise.
17490 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17491 (arc_reorg_loops): New function, calls reorg_loops.
17492 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17493 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17495 (arc_loop_hazard): Remove function, functionality moved into
17497 (arc_hazard): Remove arc_loop_hazard call.
17498 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17499 into hwloop_optimize.
17500 (arc_label_align): Remove ZOL handling.
17501 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17502 * config/arc/arc.md (doloop_begin): Remove pattern.
17503 (doloop_begin_i): Likewise.
17504 (doloop_end_i): Likewise.
17505 (doloop_fallback): Likewise.
17506 (doloop_fallback_m): Likewise.
17507 (doloop_end): Reimplement expand.
17508 (arc_lp): New pattern for LP instruction.
17509 (loop_end): New pattern.
17510 (loop_fail): Likewise.
17511 (decrement_and_branch_until_zero): Likewise.
17512 * config/arc/arc.opt (mlpc-width): New option.
17513 * doc/invoke.texi (mlpc-width): Document option.
17515 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17517 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17518 (arc_ccfsm_advance): Fix checking for delay slots.
17519 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17521 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17523 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17525 (movhi_insn): Update store instruction constraint which are saving
17526 6-bit short immediates.
17527 (movsi_insn): Consider also short scaled load operations.
17528 (zero_extendhisi2_i): Use Usd constraint instead of T.
17529 (extendhisi2_i): Add q constraint.
17530 (arc_clzsi2): Add type and length attributes.
17531 (arc_ctzsi2): Likewise.
17532 * config/arc/constraints.md (Usc): Update constraint, the
17533 assembler can parse two relocations for a single instruction.
17535 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17537 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17538 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17540 2017-08-31 Olivier Hainque <hainque@adacore.com>
17542 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17543 match as powerpc-wrs-vxworks*.
17545 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17547 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17548 register constraint for by-element operand.
17549 (aarch64_mls_elt_merge<mode>): Likewise.
17551 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17553 * config/arc/arc.c (arc_can_follow_jump): Check for short
17556 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17558 * config.gcc: Use g.opt for arc.
17559 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17560 functionality moved to ...
17561 (legitimate_scaled_address_p): New function, ...here.
17562 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17563 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17564 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17566 (arc_override_options): Handle G option.
17567 (arc_output_pic_addr_const): Correct function definition.
17568 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17569 (arc_decl_anon_ns_mem_p): Delete.
17570 (arc_in_small_data_p): Overhaul this function to take into
17571 consideration the value given via G option.
17572 (arc_rewrite_small_data_1): Renamed and corrected old
17573 arc_rewrite_small_data function.
17574 (arc_rewrite_small_data): New function.
17575 (small_data_pattern): Don't use pic_offset_table_rtx.
17576 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17577 * config/arc/simdext.md (movmisalignv2hi): Use
17578 prepare_move_operands function.
17580 (movmisalign*): Likewise.
17581 * doc/invoke.texi (ARC options): Document -G option.
17583 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17585 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17587 * config/arc/arc.c (arc_print_operand): Output scalled address for
17588 sdata whenever is possible.
17589 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17590 load/stores are available.
17591 (compact_sda_memory_operand): Check for the alignment required by
17592 code density instructions.
17593 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17595 * config/arc/constraints.md (Usd): Update constraint.
17596 (Us0): New constraint.
17597 (Usc): Update constraint.
17599 2017-08-31 Richard Biener <rguenther@suse.de>
17601 PR middle-end/82054
17602 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17603 function only once.
17605 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17607 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17608 Resize type_signature.
17610 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17611 Alan Hayward <alan.hayward@arm.com>
17612 David Sherwood <david.sherwood@arm.com>
17614 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17615 subregs whose inner modes can be stored in GPRs.
17616 (aarch64_classify_index): Likewise.
17618 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17619 Alan Hayward <alan.hayward@arm.com>
17620 David Sherwood <david.sherwood@arm.com>
17622 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17623 (V_INT_EQUIV): ...this.
17624 (v_cmp_result): Rename to...
17625 (v_int_equiv): ...this.
17626 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17627 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17628 (copysign<mode>3): Likewise.
17629 (aarch64_simd_bsl<mode>_internal): Likewise.
17630 (aarch64_simd_bsl<mode>): Likewise.
17631 (vec_cmp<mode><mode>): Likewise.
17632 (vcond<mode><mode>): Likewise.
17633 (vcond<v_cmp_mixed><mode>): Likewise.
17634 (vcondu<mode><v_cmp_mixed>): Likewise.
17635 (aarch64_cm<optab><mode>): Likewise.
17636 (aarch64_cmtst<mode>): Likewise.
17637 (aarch64_fac<optab><mode>): Likewise.
17638 (vec_perm_const<mode>): Likewise.
17639 (vcond_mask_<mode><v_cmp_result>): Rename to...
17640 (vcond_mask_<mode><v_int_equiv>): ...this.
17641 (vec_cmp<mode><v_cmp_result>): Rename to...
17642 (vec_cmp<mode><v_int_equiv>): ...this.
17644 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17645 Alan Hayward <alan.hayward@arm.com>
17646 David Sherwood <david.sherwood@arm.com>
17648 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17650 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17651 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17652 (UNSPEC_LD4_DREG): New unspecs.
17653 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17654 (aarch64_ld2<mode>_dreg_be): Replace with...
17655 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17657 (aarch64_ld3<mode>_dreg_le)
17658 (aarch64_ld3<mode>_dreg_be): Replace with...
17659 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17661 (aarch64_ld4<mode>_dreg_le)
17662 (aarch64_ld4<mode>_dreg_be): Replace with...
17663 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17666 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17668 PR tree-optimization/81987
17669 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17670 insert an initializer in a location not dominated by the stride
17673 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17675 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17676 on the entire header of the finally block in the fallthru case.
17678 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17680 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17682 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17684 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17685 rs6000_emit_move_from_cr and call renamed function.
17686 (rs6000_emit_prologue): Call renamed functions.
17687 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17688 movesi_from_cr, remove volatile CRs.
17690 2017-08-30 Jon Beniston <jon@beniston.com>
17691 Richard Biener <rguenther@suse.de>
17693 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17694 of VECTOR_MODE_P check.
17695 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17696 element vector types.
17698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17700 * df.h (df_read_modify_subreg_p): Remove in favor of...
17701 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17702 const_rtx instead of an rtx.
17703 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17704 * df-problems.c (df_word_lr_mark_ref): Likewise.
17705 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17706 (mark_pseudo_reg_dead): Likewise.
17707 (mark_ref_dead): Likewise.
17708 * reginfo.c (init_subregs_of_mode): Likewise.
17709 * sched-deps.c (sched_analyze_1): Likewise.
17710 * df-scan.c (df_def_record_1): Likewise.
17711 (df_uses_record): Likewise.
17712 (df_read_modify_subreg_p): Remove in favor of...
17713 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17714 const_rtx instead of an rtx.
17716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17717 Alan Hayward <alan.hayward@arm.com>
17718 David Sherwood <david.sherwood@arm.com>
17720 * rtl.h (partial_subreg_p): New function.
17721 * caller-save.c (save_call_clobbered_regs): Use it.
17722 * calls.c (expand_call): Likewise.
17723 * combine.c (combinable_i3pat): Likewise.
17724 (simplify_set): Likewise.
17725 (make_extraction): Likewise.
17726 (make_compound_operation_int): Likewise.
17727 (gen_lowpart_or_truncate): Likewise.
17728 (force_to_mode): Likewise.
17729 (make_field_assignment): Likewise.
17730 (reg_truncated_to_mode): Likewise.
17731 (record_truncated_value): Likewise.
17732 (move_deaths): Likewise.
17733 * cse.c (record_jump_cond): Likewise.
17734 (cse_insn): Likewise.
17735 * cselib.c (cselib_lookup_1): Likewise.
17736 * expmed.c (extract_bit_field_using_extv): Likewise.
17737 * function.c (assign_parm_setup_reg): Likewise.
17738 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17739 * ira-build.c (create_insn_allocnos): Likewise.
17740 * lra-coalesce.c (merge_pseudos): Likewise.
17741 * lra-constraints.c (match_reload): Likewise.
17742 (simplify_operand_subreg): Likewise.
17743 (curr_insn_transform): Likewise.
17744 * lra-lives.c (process_bb_lives): Likewise.
17745 * lra.c (new_insn_reg): Likewise.
17746 (lra_substitute_pseudo): Likewise.
17747 * regcprop.c (mode_change_ok): Likewise.
17748 (maybe_mode_change): Likewise.
17749 (copyprop_hardreg_forward_1): Likewise.
17750 * reload.c (push_reload): Likewise.
17751 (find_reloads): Likewise.
17752 (find_reloads_subreg_address): Likewise.
17753 * reload1.c (alter_reg): Likewise.
17754 (eliminate_regs_1): Likewise.
17755 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17757 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17759 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17760 back to if statements, including unpack.
17762 2017-08-30 Martin Liska <mliska@suse.cz>
17764 PR inline-asm/82001
17765 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17767 (func_checker::compare_asm_inputs_outputs): ... this function.
17768 (func_checker::compare_gimple_asm): Use the function to compare
17769 also ASM constrains.
17770 * ipa-icf-gimple.h: Rename the function.
17772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17773 Alan Hayward <alan.hayward@arm.com>
17774 David Sherwood <david.sherwood@arm.com>
17776 * coretypes.h (complex_mode): New type.
17777 * gdbhooks.py (build_pretty_printer): Handle it.
17778 * machmode.h (complex_mode): New class.
17779 (complex_mode::includes_p): New function.
17780 (is_complex_int_mode): Likewise.
17781 (is_complex_float_mode): Likewise.
17782 * genmodes.c (get_mode_class): Handle complex mode classes.
17783 * function.c (expand_function_end): Use is_complex_int_mode.
17785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17786 Alan Hayward <alan.hayward@arm.com>
17787 David Sherwood <david.sherwood@arm.com>
17789 * coretypes.h (scalar_mode_pod): New typedef.
17790 * gdbhooks.py (build_pretty_printer): Handle it.
17791 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17792 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17793 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17794 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17795 as_a <scalar_mode>.
17797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17798 Alan Hayward <alan.hayward@arm.com>
17799 David Sherwood <david.sherwood@arm.com>
17801 * machmode.h (mode_for_vector): Take a scalar_mode instead
17803 * stor-layout.c (mode_for_vector): Likewise.
17804 * explow.c (promote_mode): Use as_a <scalar_mode>.
17805 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17808 Alan Hayward <alan.hayward@arm.com>
17809 David Sherwood <david.sherwood@arm.com>
17811 * target.def (preferred_simd_mode): Take a scalar_mode
17812 instead of a machine_mode.
17813 * targhooks.h (default_preferred_simd_mode): Likewise.
17814 * targhooks.c (default_preferred_simd_mode): Likewise.
17815 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17816 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17817 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17818 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17819 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17820 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17821 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17822 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17824 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17825 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17826 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17827 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17828 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17829 * doc/tm.texi: Regenerate.
17830 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17834 Alan Hayward <alan.hayward@arm.com>
17835 David Sherwood <david.sherwood@arm.com>
17837 * target.def (scalar_mode_supported_p): Take a scalar_mode
17838 instead of a machine_mode.
17839 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17840 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17841 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17842 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17843 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17844 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17845 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17846 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17847 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17848 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17849 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17850 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17851 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17852 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17854 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17855 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17856 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17857 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17858 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17860 * doc/tm.texi: Regenerate.
17862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17863 Alan Hayward <alan.hayward@arm.com>
17864 David Sherwood <david.sherwood@arm.com>
17866 * coretypes.h (opt_scalar_mode): New typedef.
17867 * gdbhooks.py (build_pretty_printers): Handle it.
17868 * machmode.h (mode_iterator::get_2xwider): Add overload for
17870 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17872 * expr.c (convert_mode_scalar): Likewise.
17873 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17874 * optabs.c (expand_float): Likewise.
17875 (expand_fix): Likewise.
17876 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17879 Alan Hayward <alan.hayward@arm.com>
17880 David Sherwood <david.sherwood@arm.com>
17882 * optabs.c (expand_float): Explicitly check for scalars before
17883 using a branching expansion.
17884 (expand_fix): Likewise.
17886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17887 Alan Hayward <alan.hayward@arm.com>
17888 David Sherwood <david.sherwood@arm.com>
17890 * expr.c (convert_mode): Split scalar handling out into...
17891 (convert_mode_scalar): ...this new function. Treat the modes
17894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17895 Alan Hayward <alan.hayward@arm.com>
17896 David Sherwood <david.sherwood@arm.com>
17898 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17900 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17903 Alan Hayward <alan.hayward@arm.com>
17904 David Sherwood <david.sherwood@arm.com>
17906 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17907 rather than a machine_mode.
17908 (fixed_from_string): Likewise.
17909 (fixed_convert): Likewise.
17910 (fixed_convert_from_int): Likewise.
17911 (fixed_convert_from_real): Likewise.
17912 (real_convert_from_fixed): Likewise.
17913 * fixed-value.c (fixed_from_double_int): Likewise.
17914 (fixed_from_string): Likewise.
17915 (fixed_convert): Likewise.
17916 (fixed_convert_from_int): Likewise.
17917 (fixed_convert_from_real): Likewise.
17918 (real_convert_from_fixed): Likewise.
17919 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17922 Alan Hayward <alan.hayward@arm.com>
17923 David Sherwood <david.sherwood@arm.com>
17925 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17926 of separate mode class checks. Do not allow vector modes here.
17927 (immed_wide_int_const): Use as_a <scalar_mode>.
17928 * explow.c (trunc_int_for_mode): Likewise.
17929 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17930 (wi::shwi): Likewise.
17931 (wi::min_value): Likewise.
17932 (wi::max_value): Likewise.
17933 * dwarf2out.c (loc_descriptor): Likewise.
17934 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17935 for CONST_WIDE_INT.
17937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17938 Alan Hayward <alan.hayward@arm.com>
17939 David Sherwood <david.sherwood@arm.com>
17941 * tree.h (SCALAR_TYPE_MODE): New macro.
17942 * expr.c (expand_expr_addr_expr_1): Use it.
17943 (expand_expr_real_2): Likewise.
17944 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17945 (fold_convert_const_fixed_from_int): Likewise.
17946 (fold_convert_const_fixed_from_real): Likewise.
17947 (native_encode_fixed): Likewise
17948 (native_encode_complex): Likewise
17949 (native_encode_vector): Likewise.
17950 (native_interpret_fixed): Likewise.
17951 (native_interpret_real): Likewise.
17952 (native_interpret_complex): Likewise.
17953 (native_interpret_vector): Likewise.
17954 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17955 (simd_clone_adjust_argument_types): Likewise.
17956 (simd_clone_init_simd_arrays): Likewise.
17957 (simd_clone_adjust): Likewise.
17958 * stor-layout.c (layout_type): Likewise.
17959 * tree.c (build_minus_one_cst): Likewise.
17960 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17961 * tree-inline.c (estimate_move_cost): Likewise.
17962 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17963 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17964 (vectorizable_reduction): Likewise.
17965 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17966 (vect_recog_mixed_size_cond_pattern): Likewise.
17967 (check_bool_pattern): Likewise.
17968 (adjust_bool_pattern): Likewise.
17969 (search_type_for_mask_1): Likewise.
17970 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17971 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17972 (vectorizable_load): Likewise.
17973 (vectorizable_store): Likewise.
17974 * ubsan.c (ubsan_encode_value): Likewise.
17975 * varasm.c (output_constant): Likewise.
17977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17978 Alan Hayward <alan.hayward@arm.com>
17979 David Sherwood <david.sherwood@arm.com>
17981 * coretypes.h (scalar_mode): New class.
17982 * machmode.h (scalar_mode): Likewise.
17983 (scalar_mode::includes_p): New function.
17984 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17985 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17986 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17987 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17988 * expmed.c (store_bit_field_1): Likewise.
17989 (extract_bit_field_1): Likewise.
17990 * expr.c (write_complex_part): Likewise.
17991 (read_complex_part): Likewise.
17992 (emit_move_complex_push): Likewise.
17993 (expand_expr_real_2): Likewise.
17994 * function.c (assign_parm_setup_reg): Likewise.
17995 (assign_parms_unsplit_complex): Likewise.
17996 * optabs.c (expand_binop): Likewise.
17997 * rtlanal.c (subreg_get_info): Likewise.
17998 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17999 * varasm.c (output_constant_pool_2): Likewise.
18001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18002 Alan Hayward <alan.hayward@arm.com>
18003 David Sherwood <david.sherwood@arm.com>
18005 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18007 (expmed_mult_highpart_optab): Likewise.
18008 (expmed_mult_highpart): Likewise.
18010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18011 Alan Hayward <alan.hayward@arm.com>
18012 David Sherwood <david.sherwood@arm.com>
18014 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18015 instead of a machine_mode.
18016 (builtin_memset_read_str): Likewise.
18017 * builtins.c (c_readstr): Likewise.
18018 (builtin_memcpy_read_str): Likewise.
18019 (builtin_strncpy_read_str): Likewise.
18020 (builtin_memset_read_str): Likewise.
18021 (builtin_memset_gen_str): Likewise.
18022 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18023 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18024 instead of a machine_mode.
18025 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18027 (make_extraction): Likewise.
18028 (try_widen_shift_mode): Take and return scalar_int_modes instead
18030 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18031 a scalar_int_mode instead of a machine_mode.
18032 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18033 (avr_addr_space_pointer_mode): Likewise.
18034 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18035 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18036 (msp430_unwind_word_mode): Likewise.
18037 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18038 (spu_addr_space_pointer_mode): Likewise.
18039 (spu_addr_space_address_mode): Likewise.
18040 (spu_libgcc_cmp_return_mode): Likewise.
18041 (spu_libgcc_shift_count_mode): Likewise.
18042 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18043 (rl78_addr_space_pointer_mode): Likewise.
18044 (fl78_unwind_word_mode): Likewise.
18045 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18047 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18048 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18049 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18050 (mips_valid_pointer_mode): Likewise.
18051 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18052 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18053 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18055 (ft32_addr_space_address_mode): Likewise.
18056 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18057 scalar_int_mode instead of a machine_mode.
18058 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18060 (m32c_addr_space_address_mode): Likewise.
18061 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18062 (rs6000_eh_return_filter_mode): Likewise.
18063 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18064 (rs6000_eh_return_filter_mode): Likewise.
18065 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18066 (s390_libgcc_shift_count_mode): Likewise.
18067 (s390_unwind_word_mode): Likewise.
18068 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18070 * target.def (mode_rep_extended): Likewise.
18071 (valid_pointer_mode): Likewise.
18072 (addr_space.valid_pointer_mode): Likewise.
18073 (eh_return_filter_mode): Return a scalar_int_mode rather than
18075 (libgcc_cmp_return_mode): Likewise.
18076 (libgcc_shift_count_mode): Likewise.
18077 (unwind_word_mode): Likewise.
18078 (addr_space.pointer_mode): Likewise.
18079 (addr_space.address_mode): Likewise.
18080 * doc/tm.texi: Regenerate.
18081 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18083 (do_jump): Use scalar_int_mode for local variables.
18084 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18085 rather than a machine_mode.
18086 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18087 (scompare_loc_descriptor_wide): Likewise.
18088 (scompare_loc_descriptor_narrow): Likewise.
18089 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18091 * except.c (sjlj_emit_dispatch_table): Likewise.
18092 (expand_builtin_eh_copy_values): Likewise.
18093 * explow.c (convert_memory_address_addr_space_1): Likewise.
18094 Take a scalar_int_mode rather than a machine_mode.
18095 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18096 than a machine_mode.
18097 (memory_address_addr_space): Use scalar_int_mode for local variables.
18098 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18099 rather than a machine_mode.
18100 * expmed.c (mask_rtx): Likewise.
18101 (init_expmed_one_conv): Likewise.
18102 (expand_mult_highpart_adjust): Likewise.
18103 (extract_high_half): Likewise.
18104 (expmed_mult_highpart_optab): Likewise.
18105 (expmed_mult_highpart): Likewise.
18106 (expand_smod_pow2): Likewise.
18107 (expand_sdiv_pow2): Likewise.
18108 (emit_store_flag_int): Likewise.
18109 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18111 (extract_low_bits): Likewise.
18112 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18114 * expr.c (pieces_addr::adjust): Likewise.
18115 (can_store_by_pieces): Likewise.
18116 (store_by_pieces): Likewise.
18117 (clear_by_pieces_1): Likewise.
18118 (expand_expr_addr_expr_1): Likewise.
18119 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18120 (expand_expr_real_1): Likewise.
18121 (try_casesi): Likewise.
18122 * final.c (shorten_branches): Likewise.
18123 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18124 type of "mode" to machine_mode.
18125 * internal-fn.c (expand_arith_overflow_result_store): Take a
18126 scalar_int_mode rather than a machine_mode.
18127 (expand_mul_overflow): Use scalar_int_mode for local variables.
18128 * loop-doloop.c (doloop_modify): Likewise.
18129 (doloop_optimize): Likewise.
18130 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18131 than a machine_mode.
18132 (expand_doubleword_shift_condmove): Likewise.
18133 (expand_doubleword_shift): Likewise.
18134 (expand_doubleword_clz): Likewise.
18135 (expand_doubleword_popcount): Likewise.
18136 (expand_doubleword_parity): Likewise.
18137 (expand_absneg_bit): Use scalar_int_mode for local variables.
18138 (prepare_float_lib_cmp): Likewise.
18139 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18140 rather than a machine_mode.
18141 (convert_memory_address_addr_space): Likewise.
18142 (get_mode_bounds): Likewise.
18143 (get_address_mode): Return a scalar_int_mode rather than a
18145 * rtlanal.c (get_address_mode): Likewise.
18146 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18147 than a machine_mode.
18148 * targhooks.c (default_mode_rep_extended): Likewise.
18149 (default_valid_pointer_mode): Likewise.
18150 (default_addr_space_valid_pointer_mode): Likewise.
18151 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18152 than a machine_mode.
18153 (default_libgcc_cmp_return_mode): Likewise.
18154 (default_libgcc_shift_count_mode): Likewise.
18155 (default_unwind_word_mode): Likewise.
18156 (default_addr_space_pointer_mode): Likewise.
18157 (default_addr_space_address_mode): Likewise.
18158 * targhooks.h (default_eh_return_filter_mode): Likewise.
18159 (default_libgcc_cmp_return_mode): Likewise.
18160 (default_libgcc_shift_count_mode): Likewise.
18161 (default_unwind_word_mode): Likewise.
18162 (default_addr_space_pointer_mode): Likewise.
18163 (default_addr_space_address_mode): Likewise.
18164 (default_mode_rep_extended): Take a scalar_int_mode rather than
18166 (default_valid_pointer_mode): Likewise.
18167 (default_addr_space_valid_pointer_mode): Likewise.
18168 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18170 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18171 rather than a machine_mode.
18172 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18173 for local variables.
18174 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18175 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18176 than a machine_mode.
18178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18179 Alan Hayward <alan.hayward@arm.com>
18180 David Sherwood <david.sherwood@arm.com>
18182 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18183 the mode argument to scalar_int_mode.
18184 (do_jump_by_parts_zero_rtx): Likewise.
18185 (do_jump_by_parts_equality_rtx): Likewise.
18186 (do_jump_by_parts_greater): Take a mode argument.
18187 (do_jump_by_parts_equality): Likewise.
18188 (do_jump_1): Update calls accordingly.
18190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18191 Alan Hayward <alan.hayward@arm.com>
18192 David Sherwood <david.sherwood@arm.com>
18194 * is-a.h (safe_dyn_cast): New function.
18195 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18196 (jump_table_for_label): Likewise.
18197 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18198 instead of an rtx_insn *.
18199 (shorten_branches): Use dyn_cast instead of LABEL_P and
18200 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18201 rtx_jump_table_data::get_data_mode.
18202 (final_scan_insn): Likewise.
18204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18205 Alan Hayward <alan.hayward@arm.com>
18206 David Sherwood <david.sherwood@arm.com>
18208 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18209 trying to combine a full-register integer set with a subreg
18212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18213 Alan Hayward <alan.hayward@arm.com>
18214 David Sherwood <david.sherwood@arm.com>
18216 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18217 that is always either address_mode or pointer_mode.
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 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18224 when the two are known to be equal.
18226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18227 Alan Hayward <alan.hayward@arm.com>
18228 David Sherwood <david.sherwood@arm.com>
18230 * simplify-rtx.c (simplify_const_unary_operation): Use
18231 is_a <scalar_int_mode> instead of checking for a nonzero
18232 precision. Forcibly convert op_mode to a scalar_int_mode
18233 in that case. More clearly differentiate the operand and
18234 result modes and use the former when deciding what the value
18235 of a count-bits operation should be. Use is_int_mode instead
18236 of checking for a MODE_INT. Remove redundant check for whether
18237 this mode has a zero precision.
18239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18240 Alan Hayward <alan.hayward@arm.com>
18241 David Sherwood <david.sherwood@arm.com>
18243 * optabs.c (widen_leading): Change the type of the mode argument
18244 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18245 (widen_bswap): Likewise.
18246 (expand_parity): Likewise.
18247 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18248 (expand_ffs): Likewise.
18249 (epand_unop): Check for scalar integer modes before calling the
18252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18253 Alan Hayward <alan.hayward@arm.com>
18254 David Sherwood <david.sherwood@arm.com>
18256 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18258 (expand_expr_real_1): Update call accordingly.
18260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18261 Alan Hayward <alan.hayward@arm.com>
18262 David Sherwood <david.sherwood@arm.com>
18264 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18265 value_mode arguments. Use scalar_int_mode internally.
18266 (store_bit_field_1): Rename the new integer mode from imode
18267 to op0_mode and use it instead of GET_MODE (op0). Update calls
18268 to store_split_bit_field, store_bit_field_using_insv and
18269 store_fixed_bit_field.
18270 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18271 Use scalar_int_mode internally. Use a bit count rather than a mode
18272 when calculating the largest bit size for get_best_mode.
18273 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18274 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18275 Remove assertion that OP0 has a scalar integer mode.
18276 (store_split_bit_field): Add op0_mode and value_mode arguments.
18277 Update calls to extract_fixed_bit_field.
18278 (extract_bit_field_using_extv): Add an op0_mode argument.
18279 Use scalar_int_mode internally.
18280 (extract_bit_field_1): Rename the new integer mode from imode to
18281 op0_mode and use it instead of GET_MODE (op0). Update calls to
18282 extract_split_bit_field, extract_bit_field_using_extv and
18283 extract_fixed_bit_field.
18284 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18285 to extract_split_bit_field and extract_fixed_bit_field_1.
18286 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18287 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18288 on the target mode.
18289 (extract_split_bit_field): Add an op0_mode argument. Update call
18290 to extract_fixed_bit_field.
18292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18293 Alan Hayward <alan.hayward@arm.com>
18294 David Sherwood <david.sherwood@arm.com>
18296 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18297 * explow.c (hard_function_value): Likewise.
18298 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18299 convert_to_mode call outside the loop.
18300 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18301 for the mode iterator. Require the mode specified by max_pieces
18303 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18305 (copy_blkmode_to_reg): Likewise.
18306 (set_storage_via_setmem): Likewise.
18307 * optabs.c (prepare_cmp_insn): Likewise.
18308 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18309 * stor-layout.c (finish_bitfield_representative): Likewise.
18311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18312 Alan Hayward <alan.hayward@arm.com>
18313 David Sherwood <david.sherwood@arm.com>
18315 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18316 * expr.c (convert_move): Use them.
18317 (convert_modes): Likewise.
18318 (store_expr_with_bounds): Likewise.
18320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18321 Alan Hayward <alan.hayward@arm.com>
18322 David Sherwood <david.sherwood@arm.com>
18324 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18325 parameter for the mode of "x". Remove the "known_x", "known_mode"
18326 and "known_ret" arguments. Change the type of the mode argument
18327 to scalar_int_mode.
18328 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18329 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18330 (reg_num_sign_bit_copies_for_combine): Likewise.
18331 * rtlanal.c (nonzero_bits1): Likewise.
18332 (num_sign_bit_copies1): Likewise.
18333 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18334 (reg_num_sign_bit_copies_general): Likewise.
18335 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18336 (reg_nonzero_bits_general): Likewise.
18338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18339 Alan Hayward <alan.hayward@arm.com>
18340 David Sherwood <david.sherwood@arm.com>
18342 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18343 than in subroutines. Return 1 for non-integer modes.
18344 (cached_num_sign_bit_copies): Change the type of the mode parameter
18345 to scalar_int_mode.
18346 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18347 classes. Handle CONST_INT_P first and then check whether X also
18348 has a scalar integer mode. Check the same thing for inner registers
18349 of a SUBREG and for values that are being extended or truncated.
18351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18352 Alan Hayward <alan.hayward@arm.com>
18353 David Sherwood <david.sherwood@arm.com>
18355 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18356 in subroutines. Return the mode mask for non-integer modes.
18357 (cached_nonzero_bits): Change the type of the mode parameter
18358 to scalar_int_mode.
18359 (nonzero_bits1): Likewise. Remove early exit for other mode
18360 classes. Handle CONST_INT_P first and then check whether X
18361 also has a scalar integer mode.
18363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18364 Alan Hayward <alan.hayward@arm.com>
18365 David Sherwood <david.sherwood@arm.com>
18367 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18368 Return a scalar_int_mode and assert that the size is greater than
18370 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18371 (op_by_pieces_d::op_by_pieces_d): Likewise.
18372 (op_by_pieces_d::run): Likewise.
18373 (can_store_by_pieces): Likewise.
18375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18376 Alan Hayward <alan.hayward@arm.com>
18377 David Sherwood <david.sherwood@arm.com>
18379 * combine.c (extract_left_shift): Add a mode argument and update
18381 (make_compound_operation_int): Change the type of the mode parameter
18382 to scalar_int_mode and update the call to extract_left_shift.
18384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18385 Alan Hayward <alan.hayward@arm.com>
18386 David Sherwood <david.sherwood@arm.com>
18388 * combine.c (simplify_and_const_int): Change the type of the mode
18389 parameter to scalar_int_mode.
18390 (simplify_and_const_int_1): Likewise. Update recursive call.
18392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18393 Alan Hayward <alan.hayward@arm.com>
18394 David Sherwood <david.sherwood@arm.com>
18396 * combine.c (simplify_compare_const): Check that the mode is a
18397 scalar_int_mode (rather than VOIDmode) before testing its
18399 (simplify_comparison): Move COMPARISON_P handling out of the
18400 loop and restrict the latter part of the loop to scalar_int_modes.
18401 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18402 and when considering SUBREG_REGs. Use is_int_mode instead of
18403 checking GET_MODE_CLASS against MODE_INT.
18405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18406 Alan Hayward <alan.hayward@arm.com>
18407 David Sherwood <david.sherwood@arm.com>
18409 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18410 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18411 shift_unit_mode and for modes involved in scalar shifts.
18413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18414 Alan Hayward <alan.hayward@arm.com>
18415 David Sherwood <david.sherwood@arm.com>
18417 * combine.c (force_int_to_mode): New function, split out from...
18418 (force_to_mode): ...here. Keep xmode up-to-date and use it
18419 instead of GET_MODE (x).
18421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18422 Alan Hayward <alan.hayward@arm.com>
18423 David Sherwood <david.sherwood@arm.com>
18425 * optabs-query.h (extraction_insn::struct_mode): Change type to
18426 opt_scalar_int_mode and update comment.
18427 (extraction_insn::field_mode): Change type to scalar_int_mode.
18428 (extraction_insn::pos_mode): Likewise.
18429 * combine.c (make_extraction): Update accordingly.
18430 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18431 (get_optab_extraction_insn): Likewise.
18432 * recog.c (simplify_while_replacing): Likewise.
18433 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18434 parameter to opt_scalar_int_mode.
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 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18441 to a scalar_int_mode instead of a machine_mode.
18442 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18443 (get_best_mode): Return a boolean and use a pointer argument to store
18444 the selected mode. Replace the limit mode parameter with a bit limit.
18445 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18446 for the values returned by bit_field_mode_iterator::next_mode.
18447 (store_bit_field): Update call to get_best_mode.
18448 (store_fixed_bit_field): Likewise.
18449 (extract_fixed_bit_field): Likewise.
18450 * expr.c (optimize_bitfield_assignment_op): Likewise.
18451 * fold-const.c (optimize_bit_field_compare): Likewise.
18452 (fold_truth_andor_1): Likewise.
18453 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18454 Update for new type of m_mode.
18455 (get_best_mode): As above.
18457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18458 Alan Hayward <alan.hayward@arm.com>
18459 David Sherwood <david.sherwood@arm.com>
18461 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18462 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18463 (store_bit_field): Check is_a <scalar_int_mode> before calling
18464 strict_volatile_bitfield_p.
18465 (extract_bit_field): Likewise.
18467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18468 Alan Hayward <alan.hayward@arm.com>
18469 David Sherwood <david.sherwood@arm.com>
18471 * target.def (cstore_mode): Return a scalar_int_mode.
18472 * doc/tm.texi: Regenerate.
18473 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18474 * targhooks.h (default_cstore_mode): Likewise.
18475 * targhooks.c (default_cstore_mode): Likewise, using a forced
18477 * expmed.c (emit_cstore): Expect the target of the cstore to be
18480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18481 Alan Hayward <alan.hayward@arm.com>
18482 David Sherwood <david.sherwood@arm.com>
18484 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18486 (niter_desc): Likewise mode.
18487 (iv_analyze): Add a mode parameter.
18489 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18490 and change its type to scalar_int_mode.
18491 * loop-iv.c: Update commentary at head of file.
18492 (iv_constant): Pass the mode paraeter before the rtx it describes
18493 and change its type to scalar_int_mode. Remove VOIDmode handling.
18494 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18495 (iv_extend): Likewise.
18496 (shorten_into_mode): Likewise.
18497 (iv_add): Use scalar_int_mode.
18498 (iv_mult): Likewise.
18499 (iv_shift): Likewise.
18500 (canonicalize_iv_subregs): Likewise.
18501 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18502 it describes and change its mode to scalar_int_mode. Also change
18503 the type of the returned inner_mode to scalar_int_mode.
18504 (get_biv_step): Likewise, turning outer_mode from a pointer
18505 into a direct parameter. Update call to get_biv_step_1.
18506 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18507 iv_constant and get_biv_step.
18508 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18509 and change its type to scalar_int_mode. Don't initialise iv->mode
18510 to VOIDmode and remove later checks for its still being VOIDmode.
18511 Update calls to iv_analyze_op and iv_analyze_expr. Check
18512 is_a <scalar_int_mode> when changing the mode under consideration.
18513 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18514 Update call to iv_analyze_expr.
18515 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18516 inner register is not also a scalar_int_mode. Update call to
18518 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18519 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18520 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18521 separate mode class checks. Update calls to iv_analyze. Remove
18522 fix-up of VOIDmodes after iv_analyze_biv.
18523 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18524 don't have a scalar_int_mode. Update call to biv_p.
18526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18527 Alan Hayward <alan.hayward@arm.com>
18528 David Sherwood <david.sherwood@arm.com>
18530 * cfgexpand.c (convert_debug_memory_address): Use
18531 as_a <scalar_int_mode>.
18532 * combine.c (expand_compound_operation): Likewise.
18533 (make_extraction): Likewise.
18534 (change_zero_ext): Likewise.
18535 (simplify_comparison): Likewise.
18536 * cse.c (cse_insn): Likewise.
18537 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18538 (mem_loc_descriptor): Likewise.
18539 (loc_descriptor): Likewise.
18540 * expmed.c (init_expmed_one_mode): Likewise.
18541 (synth_mult): Likewise.
18542 (emit_store_flag_1): Likewise.
18543 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18544 of a comparison with size.
18545 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18546 (reduce_to_bit_field_precision): Likewise.
18547 * function.c (expand_function_end): Likewise.
18548 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18549 * loop-doloop.c (doloop_modify): Likewise.
18550 * optabs.c (expand_binop): Likewise.
18551 (expand_unop): Likewise.
18552 (expand_copysign_absneg): Likewise.
18553 (prepare_cmp_insn): Likewise.
18554 (maybe_legitimize_operand): Likewise.
18555 * recog.c (const_scalar_int_operand): Likewise.
18556 * rtlanal.c (get_address_mode): Likewise.
18557 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18558 (simplify_cond_clz_ctz): Likewise.
18559 * tree-nested.c (get_nl_goto_field): Likewise.
18560 * tree.c (build_vector_type_for_mode): Likewise.
18561 * var-tracking.c (use_narrower_mode): Likewise.
18563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18564 Alan Hayward <alan.hayward@arm.com>
18565 David Sherwood <david.sherwood@arm.com>
18567 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18568 * builtins.c (expand_builtin_signbit): Use it.
18569 * cfgexpand.c (expand_debug_expr): Likewise.
18570 * dojump.c (do_jump): Likewise.
18571 (do_compare_and_jump): Likewise.
18572 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18573 * expmed.c (make_tree): Likewise.
18574 * expr.c (expand_expr_real_2): Likewise.
18575 (expand_expr_real_1): Likewise.
18576 (try_casesi): Likewise.
18577 * fold-const-call.c (fold_const_call_ss): Likewise.
18578 * fold-const.c (unextend): Likewise.
18579 (extract_muldiv_1): Likewise.
18580 (fold_single_bit_test): Likewise.
18581 (native_encode_int): Likewise.
18582 (native_encode_string): Likewise.
18583 (native_interpret_int): Likewise.
18584 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18585 * internal-fn.c (expand_addsub_overflow): Likewise.
18586 (expand_neg_overflow): Likewise.
18587 (expand_mul_overflow): Likewise.
18588 (expand_arith_overflow): Likewise.
18589 * match.pd: Likewise.
18590 * stor-layout.c (layout_type): Likewise.
18591 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18592 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18593 * tree-ssanames.c (get_range_info): Likewise.
18594 * tree-switch-conversion.c (array_value_type) Likewise.
18595 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18596 (vect_recog_divmod_pattern): Likewise.
18597 (vect_recog_mixed_size_cond_pattern): Likewise.
18598 * tree-vrp.c (extract_range_basic): Likewise.
18599 (simplify_float_conversion_using_ranges): Likewise.
18600 * tree.c (int_fits_type_p): Likewise.
18601 * ubsan.c (instrument_bool_enum_load): Likewise.
18602 * varasm.c (mergeable_string_section): Likewise.
18603 (narrowing_initializer_constant_valid_p): Likewise.
18604 (output_constant): Likewise.
18606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18607 Alan Hayward <alan.hayward@arm.com>
18608 David Sherwood <david.sherwood@arm.com>
18610 * machmode.h (NARROWEST_INT_MODE): New macro.
18611 * expr.c (alignment_for_piecewise_move): Use it instead of
18612 GET_CLASS_NARROWEST_MODE (MODE_INT).
18613 (push_block): Likewise.
18614 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18616 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18619 Alan Hayward <alan.hayward@arm.com>
18620 David Sherwood <david.sherwood@arm.com>
18622 * postreload.c (move2add_valid_value_p): Change the type of the
18623 mode parameter to scalar_int_mode.
18624 (move2add_use_add2_insn): Add a mode parameter and use it instead
18626 (move2add_use_add3_insn): Likewise.
18627 (reload_cse_move2add): Update accordingly.
18629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18630 Alan Hayward <alan.hayward@arm.com>
18631 David Sherwood <david.sherwood@arm.com>
18633 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18635 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18636 * optabs.c (expand_unop): Likewise.
18638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18639 Alan Hayward <alan.hayward@arm.com>
18640 David Sherwood <david.sherwood@arm.com>
18642 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18643 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18644 (popcount_loc_descriptor): Likewise.
18645 (bswap_loc_descriptor): Likewise.
18646 (rotate_loc_descriptor): Likewise.
18647 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18648 calling the functions above.
18650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18651 Alan Hayward <alan.hayward@arm.com>
18652 David Sherwood <david.sherwood@arm.com>
18654 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18656 (try_combine): Likewise.
18657 (simplify_if_then_else): Likewise.
18658 * cse.c (cse_insn): Likewise.
18659 * dwarf2out.c (mem_loc_descriptor): Likewise.
18660 * emit-rtl.c (gen_lowpart_common): Likewise.
18661 * simplify-rtx.c (simplify_truncation): Likewise.
18662 (simplify_binary_operation_1): Likewise.
18663 (simplify_const_relational_operation): Likewise.
18664 (simplify_ternary_operation): Likewise.
18665 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18668 Alan Hayward <alan.hayward@arm.com>
18669 David Sherwood <david.sherwood@arm.com>
18671 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18672 * reload.c (push_reload): Likewise.
18673 (find_reloads): Likewise.
18675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18676 Alan Hayward <alan.hayward@arm.com>
18677 David Sherwood <david.sherwood@arm.com>
18679 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18680 (make_compound_operation_int): Likewise.
18681 (change_zero_ext): Likewise.
18682 * expr.c (convert_move): Likewise.
18683 (convert_modes): Likewise.
18684 * fwprop.c (forward_propagate_subreg): Likewise.
18685 * loop-iv.c (get_biv_step_1): Likewise.
18686 * optabs.c (widen_operand): Likewise.
18687 * postreload.c (move2add_valid_value_p): Likewise.
18688 * recog.c (simplify_while_replacing): Likewise.
18689 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18690 (simplify_binary_operation_1): Likewise. Remove redundant
18691 mode equality check.
18693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18694 Alan Hayward <alan.hayward@arm.com>
18695 David Sherwood <david.sherwood@arm.com>
18697 * combine.c (combine_simplify_rtx): Add checks for
18698 is_a <scalar_int_mode>.
18699 (simplify_if_then_else): Likewise.
18700 (make_field_assignment): Likewise.
18701 (simplify_comparison): Likewise.
18702 * ifcvt.c (noce_try_bitop): Likewise.
18703 * loop-invariant.c (canonicalize_address_mult): Likewise.
18704 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18707 Alan Hayward <alan.hayward@arm.com>
18708 David Sherwood <david.sherwood@arm.com>
18710 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18711 is_a <scalar_int_mode> instead of != BLKmode.
18713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18714 Alan Hayward <alan.hayward@arm.com>
18715 David Sherwood <david.sherwood@arm.com>
18717 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18718 instead of != VOIDmode.
18719 * combine.c (if_then_else_cond): Likewise.
18720 (change_zero_ext): Likewise.
18721 * dwarf2out.c (mem_loc_descriptor): Likewise.
18722 (loc_descriptor): Likewise.
18723 * rtlanal.c (canonicalize_condition): Likewise.
18724 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18727 Alan Hayward <alan.hayward@arm.com>
18728 David Sherwood <david.sherwood@arm.com>
18730 * simplify-rtx.c (simplify_binary_operation_1): Use
18731 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18734 Alan Hayward <alan.hayward@arm.com>
18735 David Sherwood <david.sherwood@arm.com>
18737 * wide-int.h (int_traits<unsigned char>) New class.
18738 (int_traits<unsigned short>) Likewise.
18739 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18740 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18742 * combine.c (set_nonzero_bits_and_sign_copies): Use
18743 is_a <scalar_int_mode>.
18744 (find_split_point): Likewise.
18745 (combine_simplify_rtx): Likewise.
18746 (simplify_logical): Likewise.
18747 (expand_compound_operation): Likewise.
18748 (expand_field_assignment): Likewise.
18749 (make_compound_operation): Likewise.
18750 (extended_count): Likewise.
18751 (change_zero_ext): Likewise.
18752 (simplify_comparison): Likewise.
18753 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18754 (ucompare_loc_descriptor): Likewise.
18755 (minmax_loc_descriptor): Likewise.
18756 (mem_loc_descriptor): Likewise.
18757 (loc_descriptor): Likewise.
18758 * expmed.c (init_expmed_one_mode): Likewise.
18759 * lra-constraints.c (lra_constraint_offset): Likewise.
18760 * optabs.c (prepare_libcall_arg): Likewise.
18761 * postreload.c (move2add_note_store): Likewise.
18762 * reload.c (operands_match_p): Likewise.
18763 * rtl.h (load_extend_op): Likewise.
18764 * rtlhooks.c (gen_lowpart_general): Likewise.
18765 * simplify-rtx.c (simplify_truncation): Likewise.
18766 (simplify_unary_operation_1): Likewise.
18767 (simplify_binary_operation_1): Likewise.
18768 (simplify_const_binary_operation): Likewise.
18769 (simplify_const_relational_operation): Likewise.
18770 (simplify_subreg): Likewise.
18771 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18772 * var-tracking.c (adjust_mems): Likewise.
18773 (prepare_call_arguments): Likewise.
18775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18776 Alan Hayward <alan.hayward@arm.com>
18777 David Sherwood <david.sherwood@arm.com>
18779 * machmode.h (is_int_mode): New fuction.
18780 * combine.c (find_split_point): Use it.
18781 (combine_simplify_rtx): Likewise.
18782 (simplify_if_then_else): Likewise.
18783 (simplify_set): Likewise.
18784 (simplify_shift_const_1): Likewise.
18785 (simplify_comparison): Likewise.
18786 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18787 * cse.c (notreg_cost): Likewise.
18788 (cse_insn): Likewise.
18789 * cselib.c (cselib_lookup_1): Likewise.
18790 * dojump.c (do_jump_1): Likewise.
18791 (do_compare_rtx_and_jump): Likewise.
18792 * dse.c (get_call_args): Likewise.
18793 * dwarf2out.c (rtl_for_decl_init): Likewise.
18794 (native_encode_initializer): Likewise.
18795 * expmed.c (emit_store_flag_1): Likewise.
18796 (emit_store_flag): Likewise.
18797 * expr.c (convert_modes): Likewise.
18798 (store_field): Likewise.
18799 (expand_expr_real_1): Likewise.
18800 * fold-const.c (fold_read_from_constant_string): Likewise.
18801 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18802 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18803 * optabs.c (expand_binop): Likewise.
18804 (expand_unop): Likewise.
18805 (expand_abs_nojump): Likewise.
18806 (expand_one_cmpl_abs_nojump): Likewise.
18807 * simplify-rtx.c (mode_signbit_p): Likewise.
18808 (val_signbit_p): Likewise.
18809 (val_signbit_known_set_p): Likewise.
18810 (val_signbit_known_clear_p): Likewise.
18811 (simplify_relational_operation_1): Likewise.
18812 * tree.c (vector_type_mode): Likewise.
18814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18815 Alan Hayward <alan.hayward@arm.com>
18816 David Sherwood <david.sherwood@arm.com>
18818 * machmode.h (smallest_mode_for_size): Fix formatting.
18819 (smallest_int_mode_for_size): New function.
18820 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18821 instead of smallest_mode_for_size.
18822 * combine.c (make_extraction): Likewise.
18823 * config/arc/arc.c (arc_expand_movmem): Likewise.
18824 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18825 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18826 * config/s390/s390.c (s390_expand_insv): Likewise.
18827 * config/sparc/sparc.c (assign_int_registers): Likewise.
18828 * config/spu/spu.c (spu_function_value): Likewise.
18829 (spu_function_arg): Likewise.
18830 * coverage.c (get_gcov_type): Likewise.
18831 (get_gcov_unsigned_t): Likewise.
18832 * dse.c (find_shift_sequence): Likewise.
18833 * expmed.c (store_bit_field_1): Likewise.
18834 * expr.c (convert_move): Likewise.
18835 (store_field): Likewise.
18836 * internal-fn.c (expand_arith_overflow): Likewise.
18837 * optabs-query.c (get_best_extraction_insn): Likewise.
18838 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18839 * stor-layout.c (layout_type): Likewise.
18840 (initialize_sizetypes): Likewise.
18841 * targhooks.c (default_get_mask_mode): Likewise.
18842 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18845 Alan Hayward <alan.hayward@arm.com>
18846 David Sherwood <david.sherwood@arm.com>
18848 * machmode.h (opt_mode::else_blk): New function.
18849 (int_mode_for_mode): Declare.
18850 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18851 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18853 * cfgexpand.c (expand_debug_expr): Likewise.
18854 * combine.c (gen_lowpart_or_truncate): Likewise.
18855 (gen_lowpart_for_combine): Likewise.
18856 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18857 * config/avr/avr.c (avr_to_int_mode): Likewise.
18858 (avr_out_plus_1): Likewise.
18859 (avr_out_plus): Likewise.
18860 (avr_out_round): Likewise.
18861 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18862 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18863 (s390_expand_vcond): Likewise.
18864 * config/spu/spu.c (spu_split_immediate): Likewise.
18865 (spu_expand_mov): Likewise.
18866 * dse.c (get_stored_val): Likewise.
18867 * expmed.c (store_bit_field_1): Likewise.
18868 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18870 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18871 (extract_low_bits): Likewise.
18872 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18873 handling rather than repeating the check.
18874 (emit_group_store): Likewise.
18875 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18876 * optabs.c (expand_absneg_bit): Likewise.
18877 (expand_copysign_absneg): Likewise.
18878 (expand_copysign_bit): Likewise.
18879 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18880 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18881 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18882 * var-tracking.c (prepare_call_arguments): Likewise.
18883 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18884 int_mode_for_mode instead of mode_for_size.
18885 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18888 Alan Hayward <alan.hayward@arm.com>
18889 David Sherwood <david.sherwood@arm.com>
18891 * machmode.h (int_mode_for_size): New function.
18892 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18893 instead of mode_for_size.
18894 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18896 * combine.c (expand_field_assignment): Use int_mode_for_size
18897 instead of mode_for_size.
18898 (make_extraction): Likewise.
18899 (simplify_shift_const_1): Likewise.
18900 (simplify_comparison): Likewise.
18901 * dojump.c (do_jump): Likewise.
18902 * dwarf2out.c (mem_loc_descriptor): Likewise.
18903 * emit-rtl.c (init_derived_machine_modes): Likewise.
18904 * expmed.c (flip_storage_order): Likewise.
18905 (convert_extracted_bit_field): Likewise.
18906 * expr.c (copy_blkmode_from_reg): Likewise.
18907 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18908 * internal-fn.c (expand_mul_overflow): Likewise.
18909 * lower-subreg.c (simple_move): Likewise.
18910 * optabs-libfuncs.c (init_optabs): Likewise.
18911 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18912 * tree.c (vector_type_mode): Likewise.
18913 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18914 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18915 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18916 * tree-vect-stmts.c (vectorizable_load): Likewise.
18917 (vectorizable_store): Likewise.
18919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18920 Alan Hayward <alan.hayward@arm.com>
18921 David Sherwood <david.sherwood@arm.com>
18923 * coretypes.h (pod_mode): New type.
18924 (scalar_int_mode_pod): New typedef.
18925 * machmode.h (pod_mode): New class.
18926 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18927 * genmodes.c (emit_mode_int_n): Update accordingly.
18928 * lower-subreg.h (target_lower_subreg): Change type to
18929 scalar_int_mode_pod.
18930 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18931 scalar_int_mode_pod.
18933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18934 Alan Hayward <alan.hayward@arm.com>
18935 David Sherwood <david.sherwood@arm.com>
18937 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18938 machine_mode to scalar_int_mode.
18939 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18940 (rs6000_option_override_internal): Remove cast to int.
18941 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18942 machine_mode to scalar_int_mode.
18943 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18944 (rs6000_option_override_internal): Remove cast to int.
18945 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18946 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18948 * config/s390/s390.c (s390_expand_builtin): Likewise.
18949 * coretypes.h (scalar_int_mode): New type.
18950 (opt_scalar_int_mode): New typedef.
18951 * machmode.h (scalar_int_mode): New class.
18952 (scalar_int_mode::includes_p): New function.
18953 (byte_mode): Change type to scalar_int_mode.
18954 (word_mode): Likewise.
18955 (ptr_mode): Likewise.
18956 * emit-rtl.c (byte_mode): Likewise.
18957 (word_mode): Likewise.
18958 (ptr_mode): Likewise.
18959 (init_derived_machine_modes): Update accordingly.
18960 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18961 and MODE_PARTIAL_INT.
18962 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18963 opt_scalar_int_mode.
18965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18966 Alan Hayward <alan.hayward@arm.com>
18967 David Sherwood <david.sherwood@arm.com>
18969 * target.def (libgcc_floating_mode_supported_p): Take a
18971 * doc/tm.texi: Regenerate.
18972 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18974 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18975 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18979 Alan Hayward <alan.hayward@arm.com>
18980 David Sherwood <david.sherwood@arm.com>
18982 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18983 * doc/tm.texi: Regenerate.
18984 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18985 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18986 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18987 * targhooks.h (default_floatn_mode): Likewise.
18988 * targhooks.c (default_floatn_mode): Likewise.
18989 * tree.c (build_common_tree_nodes): Update accordingly.
18991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18992 Alan Hayward <alan.hayward@arm.com>
18993 David Sherwood <david.sherwood@arm.com>
18995 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18996 (mode_iterator::iterate_p): Likewise.
18997 (mode_iterator::get_wider): Likewise.
18998 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19001 Alan Hayward <alan.hayward@arm.com>
19002 David Sherwood <david.sherwood@arm.com>
19004 * coretypes.h (opt_scalar_float_mode): New typedef.
19005 * machmode.h (float_mode_for_size): New function.
19006 * emit-rtl.c (double_mode): Delete.
19007 (init_emit_once): Use float_mode_for_size.
19008 * stor-layout.c (layout_type): Likewise.
19009 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19012 Alan Hayward <alan.hayward@arm.com>
19013 David Sherwood <david.sherwood@arm.com>
19015 * output.h (assemble_real): Take a scalar_float_mode.
19016 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19017 * config/arm/arm.md (consttable_4): Likewise.
19018 (consttable_8): Likewise.
19019 (consttable_16): Likewise.
19020 * config/mips/mips.md (consttable_float): Likewise.
19021 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19022 * varasm.c (assemble_real): Take a scalar_float_mode.
19023 (output_constant_pool_2): Update accordingly.
19024 (output_constant): Likewise.
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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19031 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19032 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19033 (native_encode_real): Likewise.
19034 (native_interpret_real): Likewise.
19035 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19036 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19039 Alan Hayward <alan.hayward@arm.com>
19040 David Sherwood <david.sherwood@arm.com>
19042 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19043 <scalar_float_mode>. Simplify.
19044 (gen_extend_conv_libfunc): Likewise.
19046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19047 Alan Hayward <alan.hayward@arm.com>
19048 David Sherwood <david.sherwood@arm.com>
19050 * coretypes.h (scalar_float_mode): New type.
19051 * machmode.h (mode_traits::from_int): Use machine_mode if
19052 USE_ENUM_MODES is defined.
19053 (is_a): New function.
19055 (dyn_cast): Likewise.
19056 (scalar_float_mode): New class.
19057 (scalar_float_mode::includes_p): New function.
19058 (is_float_mode): Likewise.
19059 * gdbhooks.py (MachineModePrinter): New class.
19060 (build_pretty_printer): Use it for scalar_float_mode.
19061 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19062 (format_helper::format_helper): Turn into a template.
19063 * genmodes.c (get_mode_class): New function.
19064 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19065 or machine_mode if none.
19066 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19067 as_a <scalar_float_mode>.
19068 * dwarf2out.c (mem_loc_descriptor): Likewise.
19069 (insert_float): Likewise.
19070 (add_const_value_attribute): Likewise.
19071 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19072 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19073 (expand_unop): Update accordingly.
19074 (expand_abs_nojump): Likewise.
19075 (expand_copysign_absneg): Take a scalar_float_mode.
19076 (expand_copysign_bit): Likewise.
19077 (expand_copysign): Update accordingly.
19079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19080 Alan Hayward <alan.hayward@arm.com>
19081 David Sherwood <david.sherwood@arm.com>
19083 * coretypes.h (opt_mode): New class.
19084 * machmode.h (opt_mode): Likewise.
19085 (opt_mode::else_void): New function.
19086 (opt_mode::require): Likewise.
19087 (opt_mode::exists): Likewise.
19088 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19089 (GET_MODE_2XWIDER_MODE): Likewise.
19090 (mode_iterator::get_wider): Update accordingly.
19091 (mode_iterator::get_2xwider): Likewise.
19092 (mode_iterator::get_known_wider): Likewise, turning into a template.
19093 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19094 forcing a wider mode to exist.
19095 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19096 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19097 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19098 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19099 * lower-subreg.c (init_lower_subreg): Likewise.
19100 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19101 on the final iteration.
19102 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19103 a wider mode exists before asking for a move pattern.
19104 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19105 forcing a wider mode to exist.
19106 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19107 returning false if no such mode exists.
19108 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19109 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19110 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19111 Avoid checking for a MODE_INT if we already know the mode is not a
19113 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19114 forcing a wider mode to exist.
19115 (expmed_mult_highpart_optab): Likewise.
19116 (expmed_mult_highpart): Likewise.
19117 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19119 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19120 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19121 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19122 * internal-fn.c (expand_mul_overflow): Update use of
19123 GET_MODE_2XWIDER_MODE.
19124 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19125 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19126 GET_MODE_WIDER_MODE.
19127 (convert_plusminus_to_widen): Likewise.
19128 * tree-switch-conversion.c (array_value_type): Likewise.
19129 * var-tracking.c (emit_note_insn_var_location): Likewise.
19130 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19131 Return false inside rather than outside the loop if no wider mode
19133 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19134 and GET_MODE_2XWIDER_MODE
19135 (can_compare_p): Use else_void.
19136 * gdbhooks.py (OptMachineModePrinter): New class.
19137 (build_pretty_printer): Use it for opt_mode.
19139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19140 Alan Hayward <alan.hayward@arm.com>
19141 David Sherwood <david.sherwood@arm.com>
19143 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19144 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19147 Alan Hayward <alan.hayward@arm.com>
19148 David Sherwood <david.sherwood@arm.com>
19150 * machmode.h (mode_traits): New structure.
19151 (get_narrowest_mode): New function.
19152 (mode_iterator::start): Likewise.
19153 (mode_iterator::iterate_p): Likewise.
19154 (mode_iterator::get_wider): Likewise.
19155 (mode_iterator::get_known_wider): Likewise.
19156 (mode_iterator::get_2xwider): Likewise.
19157 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19158 (FOR_EACH_MODE): Likewise.
19159 (FOR_EACH_MODE_FROM): Likewise.
19160 (FOR_EACH_MODE_UNTIL): Likewise.
19161 (FOR_EACH_WIDER_MODE): Likewise.
19162 (FOR_EACH_2XWIDER_MODE): Likewise.
19163 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19164 * combine.c (simplify_comparison): Likewise
19165 * config/i386/i386.c (type_natural_mode): Likewise.
19166 * cse.c (cse_insn): Likewise.
19167 * dse.c (find_shift_sequence): Likewise.
19168 * emit-rtl.c (init_derived_machine_modes): Likewise.
19169 (init_emit_once): Likewise.
19170 * explow.c (hard_function_value): Likewise.
19171 * expmed.c (extract_fixed_bit_field_1): Likewise.
19172 (extract_bit_field_1): Likewise.
19173 (expand_divmod): Likewise.
19174 (emit_store_flag_1): Likewise.
19175 * expr.c (init_expr_target): Likewise.
19176 (convert_move): Likewise.
19177 (alignment_for_piecewise_move): Likewise.
19178 (widest_int_mode_for_size): Likewise.
19179 (emit_block_move_via_movmem): Likewise.
19180 (copy_blkmode_to_reg): Likewise.
19181 (set_storage_via_setmem): Likewise.
19182 (compress_float_constant): Likewise.
19183 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19184 * optabs-query.c (get_best_extraction_insn): Likewise.
19185 * optabs.c (expand_binop): Likewise.
19186 (expand_twoval_unop): Likewise.
19187 (expand_twoval_binop): Likewise.
19188 (widen_leading): Likewise.
19189 (widen_bswap): Likewise.
19190 (expand_parity): Likewise.
19191 (expand_unop): Likewise.
19192 (prepare_cmp_insn): Likewise.
19193 (prepare_float_lib_cmp): Likewise.
19194 (expand_float): Likewise.
19195 (expand_fix): Likewise.
19196 (expand_sfix_optab): Likewise.
19197 * postreload.c (move2add_use_add2_insn): Likewise.
19198 * reg-stack.c (reg_to_stack): Likewise.
19199 * reginfo.c (choose_hard_reg_mode): Likewise.
19200 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19201 * stor-layout.c (mode_for_size): Likewise.
19202 (smallest_mode_for_size): Likewise.
19203 (mode_for_vector): Likewise.
19204 (finish_bitfield_representative): Likewise.
19205 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19206 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19207 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19208 * var-tracking.c (prepare_call_arguments): Likewise.
19210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19211 Alan Hayward <alan.hayward@arm.com>
19212 David Sherwood <david.sherwood@arm.com>
19214 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19215 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19216 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19217 * machmode.h (mode_size): Move earlier in file.
19218 (mode_precision): Likewise.
19219 (mode_inner): Likewise.
19220 (mode_nunits): Likewise.
19221 (mode_unit_size): Likewise.
19222 (unit_unit_precision): Likewise.
19223 (mode_wider): Likewise.
19224 (mode_2xwider): Likewise.
19225 (machine_mode): New class.
19226 (mode_to_bytes): New function.
19227 (mode_to_bits): Likewise.
19228 (mode_to_precision): Likewise.
19229 (mode_to_inner): Likewise.
19230 (mode_to_unit_size): Likewise.
19231 (mode_to_unit_precision): Likewise.
19232 (mode_to_nunits): Likewise.
19233 (GET_MODE_SIZE): Use mode_to_bytes.
19234 (GET_MODE_BITSIZE): Use mode_to_bits.
19235 (GET_MODE_PRECISION): Use mode_to_precision.
19236 (GET_MODE_INNER): Use mode_to_inner.
19237 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19238 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19239 (GET_MODE_NUNITS): Use mode_to_nunits.
19240 * system.h (ALWAYS_INLINE): New macro.
19241 * config/powerpcspe/powerpcspe-c.c
19242 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19243 int for arg1_mode and arg2_mode.
19245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19246 Alan Hayward <alan.hayward@arm.com>
19247 David Sherwood <david.sherwood@arm.com>
19249 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19250 Prefix mode names with E_ in case statements.
19251 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19252 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19253 (aarch64_split_simd_move): Likewise.
19254 (aarch64_gen_storewb_pair): Likewise.
19255 (aarch64_gen_loadwb_pair): Likewise.
19256 (aarch64_gen_store_pair): Likewise.
19257 (aarch64_gen_load_pair): Likewise.
19258 (aarch64_get_condition_code_1): Likewise.
19259 (aarch64_constant_pool_reload_icode): Likewise.
19260 (get_rsqrte_type): Likewise.
19261 (get_rsqrts_type): Likewise.
19262 (get_recpe_type): Likewise.
19263 (get_recps_type): Likewise.
19264 (aarch64_gimplify_va_arg_expr): Likewise.
19265 (aarch64_simd_container_mode): Likewise.
19266 (aarch64_emit_load_exclusive): Likewise.
19267 (aarch64_emit_store_exclusive): Likewise.
19268 (aarch64_expand_compare_and_swap): Likewise.
19269 (aarch64_gen_atomic_cas): Likewise.
19270 (aarch64_emit_bic): Likewise.
19271 (aarch64_emit_atomic_swap): Likewise.
19272 (aarch64_emit_atomic_load_op): Likewise.
19273 (aarch64_evpc_trn): Likewise.
19274 (aarch64_evpc_uzp): Likewise.
19275 (aarch64_evpc_zip): Likewise.
19276 (aarch64_evpc_ext): Likewise.
19277 (aarch64_evpc_rev): Likewise.
19278 (aarch64_evpc_dup): Likewise.
19279 (aarch64_gen_ccmp_first): Likewise.
19280 (aarch64_gen_ccmp_next): Likewise.
19281 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19282 (alpha_emit_xfloating_libcall): Likewise.
19283 (emit_insxl): Likewise.
19284 (alpha_arg_type): Likewise.
19285 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19286 (arc_preferred_simd_mode): Likewise.
19287 (arc_secondary_reload): Likewise.
19288 (get_arc_condition_code): Likewise.
19289 (arc_print_operand): Likewise.
19290 (arc_legitimate_constant_p): Likewise.
19291 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19292 * config/arc/arc.md (casesi_load): Likewise.
19293 (casesi_compact_jump): Likewise.
19294 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19295 (cc_use_register): Likewise.
19296 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19297 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19298 (arm_init_iwmmxt_builtins): Likewise.
19299 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19300 (neon_expand_vector_init): Likewise.
19301 (arm_attr_length_move_neon): Likewise.
19302 (maybe_get_arm_condition_code): Likewise.
19303 (arm_emit_vector_const): Likewise.
19304 (arm_preferred_simd_mode): Likewise.
19305 (arm_output_iwmmxt_tinsr): Likewise.
19306 (thumb1_output_casesi): Likewise.
19307 (thumb2_output_casesi): Likewise.
19308 (arm_emit_load_exclusive): Likewise.
19309 (arm_emit_store_exclusive): Likewise.
19310 (arm_expand_compare_and_swap): Likewise.
19311 (arm_evpc_neon_vuzp): Likewise.
19312 (arm_evpc_neon_vzip): Likewise.
19313 (arm_evpc_neon_vrev): Likewise.
19314 (arm_evpc_neon_vtrn): Likewise.
19315 (arm_evpc_neon_vext): Likewise.
19316 (arm_validize_comparison): Likewise.
19317 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19318 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19319 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19320 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19321 (c6x_preferred_simd_mode): Likewise.
19322 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19323 (epiphany_rtx_costs): Likewise.
19324 * config/epiphany/predicates.md (proper_comparison_operator):
19326 * config/frv/frv.c (condexec_memory_operand): Likewise.
19327 (frv_emit_move): Likewise.
19328 (output_move_single): Likewise.
19329 (output_condmove_single): Likewise.
19330 (frv_hard_regno_mode_ok): Likewise.
19331 (frv_matching_accg_mode): Likewise.
19332 * config/h8300/h8300.c (split_adds_subs): Likewise.
19333 (h8300_rtx_costs): Likewise.
19334 (h8300_print_operand): Likewise.
19335 (compute_mov_length): Likewise.
19336 (output_logical_op): Likewise.
19337 (compute_logical_op_length): Likewise.
19338 (compute_logical_op_cc): Likewise.
19339 (h8300_shift_needs_scratch_p): Likewise.
19340 (output_a_shift): Likewise.
19341 (compute_a_shift_length): Likewise.
19342 (compute_a_shift_cc): Likewise.
19343 (expand_a_rotate): Likewise.
19344 (output_a_rotate): Likewise.
19345 * config/i386/i386.c (classify_argument): Likewise.
19346 (function_arg_advance_32): Likewise.
19347 (function_arg_32): Likewise.
19348 (function_arg_64): Likewise.
19349 (function_value_64): Likewise.
19350 (ix86_gimplify_va_arg): Likewise.
19351 (ix86_legitimate_constant_p): Likewise.
19352 (put_condition_code): Likewise.
19353 (split_double_mode): Likewise.
19354 (ix86_avx256_split_vector_move_misalign): Likewise.
19355 (ix86_expand_vector_logical_operator): Likewise.
19356 (ix86_split_idivmod): Likewise.
19357 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19358 (ix86_build_const_vector): Likewise.
19359 (ix86_build_signbit_mask): Likewise.
19360 (ix86_match_ccmode): Likewise.
19361 (ix86_cc_modes_compatible): Likewise.
19362 (ix86_expand_branch): Likewise.
19363 (ix86_expand_sse_cmp): Likewise.
19364 (ix86_expand_sse_movcc): Likewise.
19365 (ix86_expand_int_sse_cmp): Likewise.
19366 (ix86_expand_vec_perm_vpermi2): Likewise.
19367 (ix86_expand_vec_perm): Likewise.
19368 (ix86_expand_sse_unpack): Likewise.
19369 (ix86_expand_int_addcc): Likewise.
19370 (ix86_split_to_parts): Likewise.
19371 (ix86_vectorize_builtin_gather): Likewise.
19372 (ix86_vectorize_builtin_scatter): Likewise.
19373 (avx_vpermilp_parallel): Likewise.
19374 (inline_memory_move_cost): Likewise.
19375 (ix86_tieable_integer_mode_p): Likewise.
19376 (x86_maybe_negate_const_int): Likewise.
19377 (ix86_expand_vector_init_duplicate): Likewise.
19378 (ix86_expand_vector_init_one_nonzero): Likewise.
19379 (ix86_expand_vector_init_one_var): Likewise.
19380 (ix86_expand_vector_init_concat): Likewise.
19381 (ix86_expand_vector_init_interleave): Likewise.
19382 (ix86_expand_vector_init_general): Likewise.
19383 (ix86_expand_vector_set): Likewise.
19384 (ix86_expand_vector_extract): Likewise.
19385 (emit_reduc_half): Likewise.
19386 (ix86_emit_i387_round): Likewise.
19387 (ix86_mangle_type): Likewise.
19388 (ix86_expand_round_sse4): Likewise.
19389 (expand_vec_perm_blend): Likewise.
19390 (canonicalize_vector_int_perm): Likewise.
19391 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19392 (expand_vec_perm_1): Likewise.
19393 (expand_vec_perm_interleave3): Likewise.
19394 (expand_vec_perm_even_odd_pack): Likewise.
19395 (expand_vec_perm_even_odd_1): Likewise.
19396 (expand_vec_perm_broadcast_1): Likewise.
19397 (ix86_vectorize_vec_perm_const_ok): Likewise.
19398 (ix86_expand_vecop_qihi): Likewise.
19399 (ix86_expand_mul_widen_hilo): Likewise.
19400 (ix86_expand_sse2_abs): Likewise.
19401 (ix86_expand_pextr): Likewise.
19402 (ix86_expand_pinsr): Likewise.
19403 (ix86_preferred_simd_mode): Likewise.
19404 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19405 * config/i386/sse.md (*andnot<mode>3): Likewise.
19406 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19407 (*<code><mode>3): Likewise.
19408 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19409 (ia64_expand_atomic_op): Likewise.
19410 (ia64_arg_type): Likewise.
19411 (ia64_mode_to_int): Likewise.
19412 (ia64_scalar_mode_supported_p): Likewise.
19413 (ia64_vector_mode_supported_p): Likewise.
19414 (expand_vec_perm_broadcast): Likewise.
19415 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19416 (iq2000_function_arg_advance): Likewise.
19417 (iq2000_function_arg): Likewise.
19418 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19419 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19420 (m68k_libcall_value): Likewise.
19421 (m68k_function_value): Likewise.
19422 (sched_attr_op_type): Likewise.
19423 * config/mcore/mcore.c (mcore_output_move): Likewise.
19424 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19426 (microblaze_function_arg): Likewise.
19427 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19428 (mips_print_operand): Likewise.
19429 (mips_mode_ok_for_mov_fmt_p): Likewise.
19430 (mips_vector_mode_supported_p): Likewise.
19431 (mips_preferred_simd_mode): Likewise.
19432 (mips_expand_vpc_loongson_even_odd): Likewise.
19433 (mips_expand_vec_unpack): Likewise.
19434 (mips_expand_vi_broadcast): Likewise.
19435 (mips_expand_vector_init): Likewise.
19436 (mips_expand_vec_reduc): Likewise.
19437 (mips_expand_msa_cmp): Likewise.
19438 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19439 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19440 (cc_flags_for_mode): Likewise.
19441 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19442 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19443 (nds32_output_casesi_pc_relative): Likewise.
19444 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19445 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19446 (nvptx_gen_unpack): Likewise.
19447 (nvptx_gen_pack): Likewise.
19448 (nvptx_gen_shuffle): Likewise.
19449 (nvptx_gen_wcast): Likewise.
19450 (nvptx_preferred_simd_mode): Likewise.
19451 * config/pa/pa.c (pa_secondary_reload): Likewise.
19452 * config/pa/predicates.md (base14_operand): Likewise.
19453 * config/powerpcspe/powerpcspe-c.c
19454 (altivec_resolve_overloaded_builtin): Likewise.
19455 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19457 (rs6000_preferred_simd_mode): Likewise.
19458 (output_vec_const_move): Likewise.
19459 (rs6000_expand_vector_extract): Likewise.
19460 (rs6000_split_vec_extract_var): Likewise.
19461 (reg_offset_addressing_ok_p): Likewise.
19462 (rs6000_legitimate_offset_address_p): Likewise.
19463 (rs6000_legitimize_address): Likewise.
19464 (rs6000_emit_set_const): Likewise.
19465 (rs6000_const_vec): Likewise.
19466 (rs6000_emit_move): Likewise.
19467 (spe_build_register_parallel): Likewise.
19468 (rs6000_darwin64_record_arg_recurse): Likewise.
19469 (swap_selector_for_mode): Likewise.
19470 (spe_init_builtins): Likewise.
19471 (paired_init_builtins): Likewise.
19472 (altivec_init_builtins): Likewise.
19473 (do_load_for_compare): Likewise.
19474 (rs6000_generate_compare): Likewise.
19475 (rs6000_expand_float128_convert): Likewise.
19476 (emit_load_locked): Likewise.
19477 (emit_store_conditional): Likewise.
19478 (rs6000_output_function_epilogue): Likewise.
19479 (rs6000_handle_altivec_attribute): Likewise.
19480 (rs6000_function_value): Likewise.
19481 (emit_fusion_gpr_load): Likewise.
19482 (emit_fusion_p9_load): Likewise.
19483 (emit_fusion_p9_store): Likewise.
19484 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19485 (fusion_gpr_mem_load): Likewise.
19486 (fusion_addis_mem_combo_load): Likewise.
19487 (fusion_addis_mem_combo_store): Likewise.
19488 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19489 (fusion_gpr_mem_load): Likewise.
19490 (fusion_addis_mem_combo_load): Likewise.
19491 (fusion_addis_mem_combo_store): Likewise.
19492 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19494 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19495 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19496 (rs6000_preferred_simd_mode): Likewise.
19497 (output_vec_const_move): Likewise.
19498 (rs6000_expand_vector_extract): Likewise.
19499 (rs6000_split_vec_extract_var): Likewise.
19500 (reg_offset_addressing_ok_p): Likewise.
19501 (rs6000_legitimate_offset_address_p): Likewise.
19502 (rs6000_legitimize_address): Likewise.
19503 (rs6000_emit_set_const): Likewise.
19504 (rs6000_const_vec): Likewise.
19505 (rs6000_emit_move): Likewise.
19506 (rs6000_darwin64_record_arg_recurse): Likewise.
19507 (swap_selector_for_mode): Likewise.
19508 (paired_init_builtins): Likewise.
19509 (altivec_init_builtins): Likewise.
19510 (rs6000_expand_float128_convert): Likewise.
19511 (emit_load_locked): Likewise.
19512 (emit_store_conditional): Likewise.
19513 (rs6000_output_function_epilogue): Likewise.
19514 (rs6000_handle_altivec_attribute): Likewise.
19515 (rs6000_function_value): Likewise.
19516 (emit_fusion_gpr_load): Likewise.
19517 (emit_fusion_p9_load): Likewise.
19518 (emit_fusion_p9_store): Likewise.
19519 * config/rx/rx.c (rx_gen_move_template): Likewise.
19520 (flags_from_mode): Likewise.
19521 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19522 (s390_slb_comparison): Likewise.
19523 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19524 (s390_vector_mode_supported_p): Likewise.
19525 (s390_cc_modes_compatible): Likewise.
19526 (s390_match_ccmode_set): Likewise.
19527 (s390_canonicalize_comparison): Likewise.
19528 (s390_emit_compare_and_swap): Likewise.
19529 (s390_branch_condition_mask): Likewise.
19530 (s390_rtx_costs): Likewise.
19531 (s390_secondary_reload): Likewise.
19532 (__SECONDARY_RELOAD_CASE): Likewise.
19533 (s390_expand_cs): Likewise.
19534 (s390_preferred_simd_mode): Likewise.
19535 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19536 * config/sh/sh.c (sh_print_operand): Likewise.
19537 (dump_table): Likewise.
19538 (sh_secondary_reload): Likewise.
19539 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19540 * config/sh/sh.md (casesi_worker_1): Likewise.
19541 (casesi_worker_2): Likewise.
19542 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19543 (fcc_comparison_operator): Likewise.
19544 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19545 (emit_soft_tfmode_cvt): Likewise.
19546 (sparc_preferred_simd_mode): Likewise.
19547 (output_cbranch): Likewise.
19548 (sparc_print_operand): Likewise.
19549 (sparc_expand_vec_perm_bmask): Likewise.
19550 (vector_init_bshuffle): Likewise.
19551 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19552 (spu_vector_mode_supported_p): Likewise.
19553 (spu_expand_insv): Likewise.
19554 (spu_emit_branch_or_set): Likewise.
19555 (spu_handle_vector_attribute): Likewise.
19556 (spu_builtin_splats): Likewise.
19557 (spu_builtin_extract): Likewise.
19558 (spu_builtin_promote): Likewise.
19559 (spu_expand_sign_extend): Likewise.
19560 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19561 (tilegx_simd_int): Likewise.
19562 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19563 (tilepro_simd_int): Likewise.
19564 * config/v850/v850.c (const_double_split): Likewise.
19565 (v850_print_operand): Likewise.
19566 (ep_memory_offset): Likewise.
19567 * config/vax/vax.c (vax_rtx_costs): Likewise.
19568 (vax_output_int_move): Likewise.
19569 (vax_output_int_add): Likewise.
19570 (vax_output_int_subtract): Likewise.
19571 * config/visium/predicates.md (visium_branch_operator): Likewise.
19572 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19573 (visium_print_operand_address): Likewise.
19574 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19575 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19576 (xtensa_expand_conditional_branch): Likewise.
19577 (xtensa_copy_incoming_a7): Likewise.
19578 (xtensa_output_literal): Likewise.
19579 * dfp.c (decimal_real_maxval): Likewise.
19580 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19583 Alan Hayward <alan.hayward@arm.com>
19584 David Sherwood <david.sherwood@arm.com>
19586 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19587 (mode_nunits_inline): Likewise.
19588 (mode_inner_inline): Likewise.
19589 (mode_unit_size_inline): Likewise.
19590 (mode_unit_precision_inline): Likewise.
19591 (emit_insn_modes_h): Likewise. Also emit a #define of the
19593 (emit_mode_wider): Add an E_ prefix to mode names.
19594 (emit_mode_complex): Likewise.
19595 (emit_mode_inner): Likewise.
19596 (emit_mode_adjustments): Likewise.
19597 (emit_mode_int_n): Likewise.
19598 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19599 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19600 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19601 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19602 (CRC32_BUILTIN, ENTRY): Likewise.
19603 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19604 (aarch64_pop_regs): Likewise.
19605 (aarch64_process_components): Likewise.
19606 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19607 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19608 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19609 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19610 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19611 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19613 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19614 (pcmpistr): Likewise.
19615 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19616 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19617 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19619 * config/rl78/rl78.c (mduc_regs): Likewise.
19620 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19621 (htm_expand_builtin): Likewise.
19622 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19623 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19624 * config/xtensa/xtensa.c (print_operand): Likewise.
19625 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19626 (NUM_MODE_VECTOR_INT): Likewise.
19627 * genoutput.c (null_operand): Likewise.
19628 (output_operand_data): Likewise.
19629 * genrecog.c (print_parameter_value): Likewise.
19630 * lra.c (debug_operand_data): Likewise.
19632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19634 * dwarf2out.c (scompare_loc_descriptor_wide)
19635 (scompare_loc_descriptor_narrow): New functions, split out from...
19636 (scompare_loc_descriptor): ...here.
19637 * expmed.c (emit_store_flag_int): New function, split out from...
19638 (emit_store_flag): ...here.
19640 2017-08-30 Richard Biener <rguenther@suse.de>
19642 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19643 (dwarf2out_early_finish): Move setting of AT_pubnames from
19644 early debug output to early finish.
19646 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19648 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19649 and -mdata-region to the assembler.
19651 2017-08-30 Richard Biener <rguenther@suse.de>
19653 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19655 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19656 (dwarf2out_early_global_decl): Only generate a DIE for the
19657 abstract origin if it doesn't already exist or is a declaration DIE.
19658 (resolve_addr): Do not add the linkage name twice when
19659 generating a stub DIE for the DW_TAG_GNU_call_site target.
19661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19663 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19664 Use machine_mode rather than int for arg1_mode.
19666 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19669 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19670 that the second argument of the built-in functions to unpack
19671 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19672 switch statement instead a lot of if statements.
19673 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19674 Allow 64-bit values to be in Altivec registers as well as
19675 traditional floating point registers.
19676 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19678 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19680 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19681 MAX_REGS_PER_ADDRESS == 1.
19683 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19685 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19686 * config/i386/i386.c (ix86_option_override_internal): Simplify
19687 setting of opts->x_flag_entry.
19689 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19690 Jakub Jelinek <jakub@redhat.com>
19691 Richard Biener <rguenther@suse.de>
19693 PR tree-optimization/81503
19694 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19695 folded constant fits in the target type; reorder tests for clarity.
19697 2017-08-29 Martin Liska <mliska@suse.cz>
19699 * passes.def: Include pass_lower_switch.
19700 * stmt.c (dump_case_nodes): Remove and move to
19701 tree-switch-conversion.
19702 (case_values_threshold): Likewise.
19703 (expand_switch_as_decision_tree_p): Likewise.
19704 (emit_case_decision_tree): Likewise.
19705 (expand_case): Likewise.
19706 (balance_case_nodes): Likewise.
19707 (node_has_low_bound): Likewise.
19708 (node_has_high_bound): Likewise.
19709 (node_is_bounded): Likewise.
19710 (emit_case_nodes): Likewise.
19711 (struct simple_case_node): New struct.
19712 (add_case_node): Remove.
19713 (emit_case_dispatch_table): Use vector instead of case_list.
19714 (reset_out_edges_aux): Remove.
19715 (compute_cases_per_edge): Likewise.
19716 (expand_case): Build list of simple_case_node.
19717 (expand_sjlj_dispatch_table): Use it.
19718 * tree-switch-conversion.c (struct case_node): Moved from
19719 stmt.c and adjusted.
19720 (emit_case_nodes): Likewise.
19721 (node_has_low_bound): Likewise.
19722 (node_has_high_bound): Likewise.
19723 (node_is_bounded): Likewise.
19724 (case_values_threshold): Likewise.
19725 (reset_out_edges_aux): Likewise.
19726 (compute_cases_per_edge): Likewise.
19727 (add_case_node): Likewise.
19728 (dump_case_nodes): Likewise.
19729 (balance_case_nodes): Likewise.
19730 (expand_switch_as_decision_tree_p): Likewise.
19731 (emit_jump): Likewise.
19732 (emit_case_decision_tree): Likewise.
19733 (try_switch_expansion): Likewise.
19734 (do_jump_if_equal): Likewise.
19735 (emit_cmp_and_jump_insns): Likewise.
19736 (fix_phi_operands_for_edge): New function.
19737 (record_phi_operand_mapping): Likewise.
19738 (class pass_lower_switch): New pass.
19739 (pass_lower_switch::execute): New function.
19740 (make_pass_lower_switch): Likewise.
19741 (conditional_probability):
19742 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19743 * tree-pass.h: Add make_pass_lower_switch.
19745 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19748 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19751 2017-08-29 Richard Biener <rguenther@suse.de>
19753 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19754 we do not add a DW_AT_inline attribute twice.
19755 (gen_subprogram_die): Remove code setting DW_AT_inline on
19756 DECL_ABSTRACT_P nodes.
19758 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19760 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19761 calls to internal functions.
19762 (gimplify_modify_expr): Likewise.
19763 * tree-call-cdce.c (use_internal_fn): Likewise.
19764 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19765 (convert_to_divmod): Set the nothrow flag.
19766 * tree-if-conv.c (predicate_mem_writes): Likewise.
19767 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19768 (vectorizable_call): Likewise.
19769 (vectorizable_store): Likewise.
19770 (vectorizable_load): Likewise.
19771 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19772 (vect_recog_mask_conversion_pattern): Likewise.
19774 2017-08-29 Martin Liska <mliska@suse.cz>
19777 * gcc.c (driver_handle_option): Add new argument.
19778 * opts-common.c (handle_option): Pass
19779 target_option_override_hook.
19780 * opts-global.c (lang_handle_option): Add new option.
19781 (set_default_handlers): Add new argument.
19782 (decode_options): Likewise.
19783 * opts.c (target_handle_option): Likewise.
19784 (common_handle_option): Call target_option_override_hook.
19785 * opts.h (struct cl_option_handler_func): Add hook for
19786 target option override.
19787 (struct cl_option_handlers): Likewise.
19788 (set_default_handlers): Add new argument.
19789 (decode_options): Likewise.
19790 (common_handle_option): Likewise.
19791 (target_handle_option): Likewise.
19792 * toplev.c (toplev::main): Pass targetm.target_option.override
19795 2017-08-29 Richard Biener <rguenther@suse.de>
19796 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19798 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19799 life to the active subtree.
19801 2017-08-28 Jeff Law <law@redhat.com>
19803 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19804 derive_equivalences.
19805 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19806 Code moved into....
19807 (edge_info::derive_equivalences): New private member function
19809 * tree-ssa-dom.c (class edge_info): Changed from a struct
19810 to a class. Add ctor/dtor, methods and data members.
19811 (edge_info::edge_info): Renamed from allocate_edge_info.
19812 Initialize additional members.
19813 (edge_info::~edge_info): New.
19814 (free_dom_edge_info): Delete the edge info.
19815 (record_edge_info): Use new class & associated member functions.
19816 Tighten forms for testing for edge equivalences.
19817 (record_temporary_equivalences): Iterate over the simple
19818 equivalences rather than assuming there's only one per edge.
19819 (cprop_into_successor_phis): Iterate over the simple
19820 equivalences rather than assuming there's only one per edge.
19821 (optimize_stmt): Use operand_equal_p rather than pointer
19822 equality for mini-DSE code.
19824 2017-08-28 Nathan Sidwell <nathan@acm.org>
19826 * gcc.c (execute): Fold SIGPIPE handling into switch
19827 statement. Adjust internal error message.
19829 2017-08-28 Richard Biener <rguenther@suse.de>
19832 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19833 Do nothing for removed DIEs.
19835 2017-08-28 Richard Biener <rguenther@suse.de>
19837 PR tree-optimization/81977
19838 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19841 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19844 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19845 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19846 using targetm.gen_mem_thread_fence.
19848 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19851 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19852 predicate to register_operand. Reorder operands.
19853 (*btr<mode>): Ditto.
19854 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19855 (*btr<mode>_mask): Ditto.
19857 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19859 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19860 * config/rs6000/xmmintrin.h: New file.
19861 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19863 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19866 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19867 parameter and_insn and return it.
19868 (recombine_lvx_pattern): Insert a copy to ensure availability of
19869 the base register of the copied masking operation at the point of
19870 the instruction replacement.
19871 (recombine_stvx_pattern): Likewise.
19873 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19875 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19876 undocumented switches.
19877 (-mpower9-dform-vector): Likewise.
19878 (-mpower9-dform): Likewise.
19879 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19880 comments to delete references to -mpower9-dform* switches.
19881 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19882 Delete reference to -mpower9-dform* switches, test for
19883 -mpower9-vector instead.
19884 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19885 (OTHER_P9_VECTOR_MASKS): Likewise.
19886 (POWERPC_MASKS): Likewise.
19887 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19888 tests against -mpower9-dform* to -mpower9-vector. Delete code
19889 that checked for -mpower9-dform* consistancy with other options.
19890 Add test for -mpower9-misc to enable other power9 switches.
19891 (rs6000_init_hard_regno_mode_ok): Likewise.
19892 (rs6000_option_override_internal): Likewise.
19893 (rs6000_emit_prologue): Likewise.
19894 (rs6000_emit_epilogue): Likewise.
19895 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19896 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19897 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19899 (emit_fusion_p9_store): Likewise.
19900 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19901 resetting these macros if the assembler does not support ISA 3.0
19903 (TARGET_P9_DFORM_VECTOR): Likewise.
19904 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19905 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19907 2017-08-25 Alan Modra <amodra@gmail.com>
19909 PR rtl-optimization/81747
19910 * cse.c (cse_extended_basic_block): Don't attempt to record
19911 equivalences for degenerate conditional branches that branch
19912 to their fall-through.
19914 2017-08-24 Martin Sebor <msebor@redhat.com>
19916 PR middle-end/81908
19917 * gimple-fold.c (size_must_be_zero_p): New function.
19918 (gimple_fold_builtin_memory_op): Call it.
19920 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19922 * config/rs6000/mm_malloc.h: New file.
19924 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19926 PR tree-optimization/81913
19927 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19928 analysis when either IVs in condition can wrap.
19930 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19932 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19933 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19935 2017-08-24 Richard Biener <rguenther@suse.de>
19938 * targhooks.c (default_target_can_inline_p): Properly
19939 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19940 is present and always compare.
19941 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19942 infer -mfpmath=sse from TARGET_SSE_P.
19943 (ix86_can_inline_p): Properly use target_option_default_node when
19944 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19946 2017-08-24 Richard Biener <rguenther@suse.de>
19949 * dwarf2out.c (output_die): Handle flag_generate_offload like
19951 (output_comp_unit): Likewise.
19952 (gen_array_type_die): Likewise.
19953 (dwarf2out_early_finish): Likewise.
19954 (note_variable_value_in_expr): Likewise.
19955 (dwarf2out_finish): Likewise. Adjust assert.
19956 * cgraphunit.c (symbol_table::compile): Move setting of
19957 flag_generate_offload earlier ...
19958 (symbol_table::finalize_compilation_unit): ... here, before
19959 early debug finalization.
19961 2017-08-24 Richard Biener <rguenther@suse.de>
19963 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19964 and ipa-fnsummary.h.
19965 (ix86_can_inline_p): When ix86_fpmath flags do not match
19966 check whether the callee uses FP math at all.
19968 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19970 PR middle-end/81931
19971 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19972 instead of TYPE_PRECISION.
19974 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19975 Alan Hayward <alan.hayward@arm.com>
19976 David Sherwood <david.sherwood@arm.com>
19978 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19980 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19982 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19984 (ix86_frame::stack_realign_allocate): New field.
19985 (struct machine_frame_state): Modify comments.
19986 (machine_frame_state::sp_realigned_fp_end): New field.
19987 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19988 layout calculation.
19989 (sp_valid_at): Add assertion to assure no attempt to access invalid
19990 offset of a realigned stack.
19991 (fp_valid_at): Likewise.
19992 (choose_baseaddr): Modify comments.
19993 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19994 ix86_expand_prologue.
19995 (ix86_expand_prologue): Modify stack realignment and allocation.
19996 (ix86_expand_epilogue): Modify comments.
19997 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19998 avx2_runtime, avx512f, and avx512f_runtime.
20000 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20002 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20003 (mstackrealign): Do not init to -1.
20004 * config/i386/i386.c (ix86_option_override_internal):
20005 Check opts_set, not opts when setting default value of
20006 opts->x_ix86_force_align_arg_pointer.
20008 2017-08-23 Richard Biener <rguenther@suse.de>
20010 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20011 lang_hooks.decl_printable_name.
20012 * print-rtl-function.c (print_rtx_function): Likewise.
20013 * tree-pretty-print.c (dump_function_header): Likewise.
20015 2017-08-23 Richard Biener <rguenther@suse.de>
20018 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20019 -g0 at compile-time.
20021 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20023 PR middle-end/19706
20024 * doc/sourcebuild.texi (Other hardware attributes):
20027 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20029 PR middle-end/19706
20030 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20031 Fix single-use check.
20033 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20035 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20037 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20039 * doc/install.texi: Modify to add more details on running selected
20042 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20044 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20045 is combined with -mabi=ms.
20046 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20049 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20051 PR tree-optimization/81488
20052 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20053 and cached_basis fields.
20054 (MAX_SPREAD): New constant.
20055 (alloc_cand_and_find_basis): Initialize new fields.
20056 (clear_visited): New function.
20057 (create_phi_basis_1): Rename from create_phi_basis, set visited
20058 and cached_basis fields.
20059 (create_phi_basis): New wrapper function.
20060 (phi_add_costs_1): Rename from phi_add_costs, add spread
20061 parameter, set visited field, short-circuit when limits reached.
20062 (phi_add_costs): New wrapper function.
20063 (record_phi_increments_1): Rename from record_phi_increments, set
20065 (record_phi_increments): New wrapper function.
20066 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20067 (phi_incr_cost): New wrapper function.
20068 (all_phi_incrs_profitable_1): Rename from
20069 all_phi_incrs_profitable, set visited field.
20070 (all_phi_incrs_profitable): New wrapper function.
20072 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20073 Alan Hayward <alan.hayward@arm.com>
20074 David Sherwood <david.sherwood@arm.com>
20076 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20077 that takes the outer and inner modes.
20078 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20079 comparison as the canonical test for a paradoxical subreg.
20080 * combine.c (simplify_set): Use paradoxical_subreg_p.
20081 (make_extraction): Likewise.
20082 (force_to_mode): Likewise.
20083 (rtx_equal_for_field_assignment_p): Likewise.
20084 (gen_lowpart_for_combine): Likewise.
20085 (simplify_comparison): Likewise.
20086 * cse.c (equiv_constant): Likewise.
20087 * expmed.c (store_bit_field_1): Likewise.
20088 * final.c (alter_subreg): Likewise.
20089 * fwprop.c (propagate_rtx): Likewise.
20090 (forward_propagate_subreg): Likewise.
20091 * ira-conflicts.c (ira_build_conflicts): Likewise.
20092 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20093 * lra-constraints.c (curr_insn_transform): Likewise.
20094 (split_reg): Likewise.
20095 * lra-eliminations.c (move_plus_up): Likewise.
20096 (lra_eliminate_regs_1): Likewise.
20097 * recog.c (general_operand): Likewise.
20098 * ree.c (combine_reaching_defs): Likewise.
20099 * reload.c (push_reload): Likewise.
20100 (find_reloads): Likewise.
20101 * reload1.c (elimination_effects): Likewise.
20102 (compute_reload_subreg_offset): Likewise.
20103 (choose_reload_regs): Likewise.
20104 * rtlanal.c (subreg_lsb_1): Likewise.
20105 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20106 (simplify_subreg): Likewise.
20107 * var-tracking.c (track_loc_p): Likewise.
20108 * emit-rtl.c (byte_lowpart_offset): Likewise.
20109 (paradoxical_subreg_p): Delete out-of-line definition.
20111 2017-08-22 Jeff Law <law@redhat.com>
20113 PR tree-optimization/81741
20114 PR tree-optimization/71947
20115 * tree-ssa-dom.c: Include tree-inline.h.
20116 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20117 equivalences if one is more expensive to compute than the other.
20118 * tree-ssa-scopedtables.h (class const_or_copies): Make
20119 record_const_or_copy_raw method private.
20120 (class avail_exprs_stack): New method simplify_binary_operation.
20121 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20122 avail_exprs_stack::simplify_binary_operation as needed.
20123 (avail_exprs_stack::simplify_binary_operation): New function.
20125 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20127 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20128 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20129 (DOT_SYMBOLS): Likewise.
20130 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20131 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20132 (RS6000_ABI_NAME): Likewise.
20133 (TARGET_CMODEL): Likewise.
20134 (TOC_SECTION_ASM_OP): Likewise.
20135 (SET_CMODEL): New macro.
20136 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20138 2017-08-22 Richard Biener <rguenther@suse.de>
20140 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20141 to allow for free-lang-data replacements similar to verify_type_variant.
20143 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20144 Alan Hayward <alan.hayward@arm.com>
20145 David Sherwood <david.sherwood@arm.com>
20147 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20148 VOIDmode for the LABEL_REF.
20150 2017-08-22 Richard Biener <rguenther@suse.de>
20152 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20154 2017-08-22 Richard Biener <rguenther@suse.de>
20156 * common.opt (feliminate-dwarf2-dups): Ignore.
20157 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20158 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20159 same_die_p_wrap, compute_section_prefix,
20160 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20161 (comdat_symbol_id, comdat_symbol_number): Likewise.
20162 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20164 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20165 (output_die): Mark unreachable path unreachable.
20166 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20167 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20168 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20169 (dwarf2out_early_finish): Likewise.
20171 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20173 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20175 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20178 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20179 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20180 (avr_attribute_table) <io, io_low, address>: Initialize
20181 .decl_required with true.
20183 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20185 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20186 undocumented debugging options.
20187 (-mvsx-scalar-double): Likewise.
20188 (-mallow-df-permute): Likewise.
20189 (-mvectorize-builtins): Likewise.
20190 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20191 (rs6000_builtin_vectorized_function): Likewise.
20192 (rs6000_builtin_md_vectorized_function): Likewise.
20193 (rs6000_opt_vars): Likewise.
20195 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20198 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20199 (*btrq_imm): Rename from *btrq.
20200 (*btcq_imm): Rename from *btcq.
20201 (btsc): New code attribute.
20202 (*<btsc><mode>): New insn pattern.
20203 (*btr<mode>): Ditto.
20204 (*<btsc><mode>_mask): New insn_and_split pattern.
20205 (*btr<mode>_mask): Ditto.
20207 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20208 Alan Hayward <alan.hayward@arm.com>
20209 David Sherwood <david.sherwood@arm.com>
20211 * function.c (pad_below): Simplify padding calculation.
20213 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20214 Alan Hayward <alan.hayward@arm.com>
20215 David Sherwood <david.sherwood@arm.com>
20217 * target.def (function_prologue): Remove frame size argument.
20218 (function_epilogue): Likewise.
20219 * doc/tm.texi: Regenerate.
20220 * final.c (final_start_function): Update call to function_prologue.
20221 (final_end_function): Update call to function_epilogue.
20222 (default_function_pro_epilogue): Remove frame size argument.
20223 * output.h (default_function_pro_epilogue): Likewise.
20224 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20225 (arm_output_function_prologue): Likewise.
20226 * config/frv/frv.c (frv_function_prologue): Likewise.
20227 (frv_function_epilogue): Likewise.
20228 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20229 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20230 (ia64_output_function_epilogue): Likewise.
20231 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20232 (m32r_output_function_epilogue): Likewise.
20233 * config/microblaze/microblaze.c (microblaze_function_prologue)
20234 (microblaze_function_epilogue): Likewise.
20235 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20236 (mips_output_function_epilogue): Likewise.
20237 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20238 (mmix_target_asm_function_epilogue): Likewise.
20239 * config/msp430/msp430.c (msp430_start_function): Likewise.
20240 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20241 (nds32_asm_function_epilogue): Likewise.
20242 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20243 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20244 (pa_output_function_epilogue): Likewise.
20245 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20246 (rs6000_output_function_epilogue): Likewise.
20247 * config/rl78/rl78.c (rl78_start_function): Likewise.
20248 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20249 (rs6000_output_function_epilogue): Likewise.
20250 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20251 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20252 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20253 (sparc_asm_function_epilogue): Likewise.
20255 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20257 * tree.h (type_has_mode_precision_p): New function.
20258 * convert.c (convert_to_integer_1): Use it.
20259 * expr.c (expand_expr_real_2): Likewise.
20260 (expand_expr_real_1): Likewise.
20261 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20262 * match.pd: Likewise.
20263 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20264 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20265 * tree-tailcall.c (process_assignment): Likewise.
20266 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20267 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20268 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20269 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20270 (vectorizable_assignment): Likewise.
20271 (vectorizable_shift): Likewise.
20272 (vectorizable_operation): Likewise.
20273 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20275 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20277 * match.pd: Add pow (C, x) simplification.
20279 2017-08-21 Richard Biener <rguenther@suse.de>
20281 PR tree-optimization/81900
20282 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20283 for blocks with abnormal predecessors.
20284 (compute_antic): Do not set visited flag prematurely.
20286 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20289 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20291 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20293 * stor-layout.h (vector_type_mode): Move to...
20294 * tree.h (vector_type_mode): ...here.
20295 * stor-layout.c (vector_type_mode): Move to...
20296 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20298 2017-08-21 Richard Biener <rguenther@suse.de>
20300 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20301 register_external_die hooks.
20302 (debug_false_tree_charstarstar_uhwistar): Declare.
20303 (debug_nothing_tree_charstar_uhwi): Likewise.
20304 * debug.c (do_nothing_debug_hooks): Adjust.
20305 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20306 (debug_nothing_tree_charstar_uhwi): Likewise.
20307 * dbxout.c (dbx_debug_hooks): Adjust.
20308 (xcoff_debug_hooks): Likewise.
20309 * sdbout.c (sdb_debug_hooks): Likewise.
20310 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20311 * dwarf2out.c (macinfo_label_base): New global.
20312 (dwarf2out_register_external_die): New function for the
20313 register_external_die hook.
20314 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20315 (dwarf2_debug_hooks): Use them.
20316 (dwarf2_lineno_debug_hooks): Adjust.
20317 (struct die_struct): Add with_offset flag.
20318 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20319 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20320 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20321 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20322 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20323 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20324 defining section names for the early LTO debug variants.
20325 (reset_indirect_string): New helper.
20326 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20327 (print_dw_val): Add support for offsetted symbol references.
20328 (get_ultimate_context): Split out from is_cxx.
20329 (is_cxx): Use get_ultimate_context.
20330 (is_fortran): Add decl overload.
20331 (compute_comp_unit_symbol): Split out worker from
20332 compute_section_prefix.
20333 (compute_section_prefix): Call compute_comp_unit_symbol and
20334 set comdat_type_p here.
20335 (output_die): Skip DIE symbol output for the LTO added one.
20336 Handle DIE symbol references with offset.
20337 (output_comp_unit): Guard section name mangling properly.
20338 For LTO debug sections emit a symbol at the section beginning
20339 which we use to refer to its DIEs.
20340 (add_abstract_origin_attribute): For DIEs registered via
20341 dwarf2out_register_external_die directly refer to the early
20342 DIE rather than indirectly through the shadow one we created.
20343 Remove obsolete call to dwarf2out_abstract_function for
20344 non-function/block origins.
20345 (gen_array_type_die): When generating early LTO debug do
20346 not emit DW_AT_string_length.
20347 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20348 late when in LTO. As suggested place a gcc_unreachable for
20349 the DECL_ABSTRACT_P case.
20350 (gen_subprogram_die): Avoid another specification DIE
20351 for early built declarations/definitions for the late LTO case.
20352 (gen_variable_die): Add type references for late duplicated VLA dies
20354 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20355 we have the abstract instance already.
20356 (process_scope_var): Adjust decl DIE contexts in LTO which
20357 first puts them in limbo.
20358 (gen_decl_die): Do not generate type DIEs late apart from
20359 types for VLAs or for decls we do not yet have a DIE. Do not
20360 call dwarf2out_abstract_function late.
20361 (dwarf2out_early_global_decl): Make sure to create DIEs
20362 for abstract instances of a decl first.
20363 (dwarf2out_late_global_decl): Adjust comment.
20364 (output_macinfo_op): With multiple macro sections use
20365 macinfo_label_base to distinguish labels.
20366 (output_macinfo): Likewise. Update macinfo_label_base.
20367 Pass in the line info label.
20368 (note_variable_value_in_expr): When generating LTO resolve
20369 all variable values here by generating DIEs as needed.
20370 (init_sections_and_labels): Add early LTO debug flag parameter
20371 and generate different sections and names if set. Add generation
20372 counter for the labels so we can have multiple of them.
20373 (reset_dies): Helper to allow DIEs to be output multiple times.
20374 (dwarf2out_finish): When outputting DIEs to the fat part of an
20375 LTO object first reset DIEs.
20376 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20377 (modified_type_die): Check for decl_ultimate_origin being self
20379 (gen_type_die_with_usage): Likewise.
20380 (gen_typedef_die): Allow decl_ultimate_origin being self.
20381 (set_decl_abstract_flags): Remove.
20382 (set_block_abstract_flags): Likewise.
20383 (dwarf2out_abstract_function): Treat the early generated DIEs
20384 as the abstract copy and only add DW_AT_inline and
20385 DW_AT_artificial here and call set_decl_origin_self.
20386 If the DIE has an abstract origin don't do anything.
20387 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20388 if we have none yet (Go fails to build one, PR78628).
20389 (variably_modified_type_p): Prevent endless recursion for Ada
20390 cyclic pointer types.
20391 * lto-streamer-in.c: Include debug.h.
20392 (dref_queue): New global.
20393 (lto_read_tree_1): Stream in DIE references.
20394 (lto_input_tree): Register DIE references.
20395 (input_function): Stream DECL_DEBUG_ARGS.
20396 * lto-streamer-out.c: Include debug.h.
20397 (lto_write_tree_1): Output DIE references.
20398 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20399 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20400 (output_function): Stream DECL_DEBUG_ARGS.
20401 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20402 Stream DECL_ABSTRACT_ORIGIN.
20403 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20404 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20405 DECL_CONTEXT for file-scope decls.
20406 * lto-streamer.h (struct dref_entry): Declare.
20407 (dref_queue): Likewise.
20408 * cfgexpand.c (pass_expand::execute): Do not call the
20409 outlining_inline_function hook here.
20410 * lto-wrapper.c (debug_obj): New global.
20411 (tool_cleanup): Unlink it if required.
20412 (debug_objcopy): New function.
20413 (run_gcc): Handle early debug sections in the IL files by
20414 extracting them to separate files, partially linkin them and
20415 feeding the result back as result to the linker.
20416 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20417 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20418 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20419 sections into a separate segment.
20420 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20422 (darwin_asm_dwarf_section): Likewise.
20423 (darwin_asm_output_dwarf_offset): Likewise.
20424 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20426 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20427 Alan Hayward <alan.hayward@arm.com>
20428 David Sherwood <david.sherwood@arm.com>
20430 * read-md.h (md_reader::record_potential_iterator_use): Replace
20431 pointer argument with an rtx and an index.
20432 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20433 (apply_mode_iterator): Likewise.
20434 (apply_code_iterator): Likewise.
20435 (apply_int_iterator): Likewise.
20436 (apply_subst_iterator): Likewise.
20437 (record_iterator_use): Likewise.
20438 (record_attribute_use): Likewise.
20439 (md_reader::record_potential_iterator_use): Likewise. Update calls
20440 to record_iterator_use and apply_iterator.
20441 (iterator_use): Replace ptr with x and index.
20442 (attribute_use): Likewise.
20443 (apply_attribute_uses): Update calls to apply_iterator.
20444 (apply_iterators): Likewise. Update initialization of iterator_use.
20445 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20446 and record_potential_iterator_use.
20447 (rtx_reader::read_rtx_operand): Likewise.
20449 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20450 Alan Hayward <alan.hayward@arm.com>
20451 David Sherwood <david.sherwood@arm.com>
20453 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20456 2017-08-21 Richard Biener <rguenther@suse.de>
20458 PR middle-end/81884
20459 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20460 at struct end conservatively when comparing common bases.
20462 2017-08-21 Richard Biener <rguenther@suse.de>
20464 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20465 (mem_ref_in_stmt): Remove.
20466 (determine_max_movement): Use ref index to get at the reference.
20467 (invariantness_dom_walker::before_dom_children): Deal with
20468 lim data already initialized.
20469 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20471 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20473 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20474 (TARGET_ISA_ROUND): Ditto.
20475 (TARGET_ROUND): Ditto.
20476 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20477 * config/i386/i386.md: Ditto.
20478 * config/i386/sse.md: Ditto.
20479 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20480 with OPTION_MASK_ISA_SSE4_1.
20482 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20485 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20486 __builtin_ia32_lzcnt_u16.
20488 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20491 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20492 (rs6000_set_current_function): Rewrite function to use it.
20494 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20497 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20498 and TYPE_WARN_IF_NOT_ALIGN.
20499 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20500 (handle_warn_if_not_align): New.
20501 (place_union_field): Call handle_warn_if_not_align.
20502 (place_field): Call handle_warn_if_not_align.
20503 Copy TYPE_WARN_IF_NOT_ALIGN.
20504 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20505 (layout_type): Likewise.
20506 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20508 (tree_decl_common): Add warn_if_not_align.
20509 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20510 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20511 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20512 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20513 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20514 * doc/extend.texi: Document warn_if_not_aligned attribute.
20515 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20517 2017-08-17 Martin Liska <mliska@suse.cz>
20520 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20521 (get_data_dependence): Use it as pointer type.
20522 (distribute_loop): Likewise.
20524 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20526 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20527 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20528 (p8_vmrgow_<mode>_direct): New define_insn.
20529 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20530 handle endianness for vmrgew and vmrgow permute patterns.
20532 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20534 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20535 * config/rs6000/rs6000-cpus.def: Remove comment.
20536 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20537 (POWERPC_MASKS): Likewise.
20538 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20539 use of TARGET_VSX_TIMODE.
20540 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20541 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20542 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20543 (rs6000_option_override_internal): Remove dead code.
20544 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20545 (rs6000_legitimize_reload_address): Likewise.
20546 (rs6000_legitimate_address_p): Likewise.
20547 (rs6000_opt_masks): Delete "vsx-timode".
20548 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20549 from function comment.
20550 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20551 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20552 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20554 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20555 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20556 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20557 (VSX_TI): Likewise.
20559 (define_peephole2): Likewise.
20561 2017-08-17 Martin Sebor <msebor@redhat.com>
20564 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20565 past the end of an array.
20566 (test_pp_format): Add test cases.
20568 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20570 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20571 missing ECF_NOTHROW flags.
20573 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20576 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20577 operands residing in integer registers.
20578 (*vsx_le_perm_load_<mode>): Likewise.
20579 (*vsx_le_perm_store_<mode>): Likewise.
20580 (define_peephole2): Add peepholes to optimize the above.
20582 2017-08-17 Marek Polacek <polacek@redhat.com>
20584 PR middle-end/81814
20585 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20586 to mimic what shorten_compare did. Change the return type to bool.
20587 (fold_cond_expr_with_comparison): Update call to
20588 operand_equal_for_comparison_p.
20589 (fold_ternary_loc): Likewise.
20591 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20593 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20595 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20597 2017-08-17 Richard Biener <rguenther@suse.de>
20599 * tree-ssa-structalias.c (solve_graph): When propagating
20600 to successors update the graphs succ edges and avoid duplicate work.
20602 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20605 * config/i386/i386.c (ix86_option_override_internal): Save target
20606 specific options after ix86_stack_protector_guard_reg was changed.
20608 2017-08-17 Richard Biener <rguenther@suse.de>
20610 PR tree-optimization/81827
20611 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20612 (new_var_info): Initialize it conservatively.
20613 (get_call_vi): Mark register vars.
20614 (new_scalar_tmp_constraint_exp): Likewise.
20615 (handle_rhs_call): Likewise.
20616 (handle_const_call): Likewise.
20617 (create_function_info_for): Likewise.
20618 (solve_constraints): Sort varinfos to separate register from
20619 non-register vars to pack points-to solution bitmaps during
20622 2017-08-17 Marek Polacek <polacek@redhat.com>
20624 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20626 2017-08-17 Richard Biener <rguenther@suse.de>
20628 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20629 to true when overflow is undefined and we saturated the result.
20631 2017-08-17 Alan Modra <amodra@gmail.com>
20634 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20635 Don't use store multiple if only one reg needs saving.
20636 (interesting_frame_related_regno): New function.
20637 (rs6000_frame_related): Don't emit frame info for regs that
20639 (rs6000_emit_epilogue): Likewise.
20641 2017-08-16 Nathan Sidwell <nathan@acm.org>
20643 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20644 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20645 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20646 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20647 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20648 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20649 (verify_type): Adjust for TYPE_BINFO move.
20650 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20651 process TYPE_BINFO directly.
20652 (hash_tree): Likewise.
20653 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20655 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20658 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20660 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20662 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20665 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20666 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20667 Add ix86_binary_operator_ok to insn constraint.
20668 (*iordi_1_bts): Ditto.
20669 (*xordi_1_btc): Ditto.
20670 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20671 Update corresponding peephole2 pattern.
20675 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20677 PR tree-optimization/81832
20678 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20679 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20681 2017-08-16 Marek Polacek <polacek@redhat.com>
20684 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20685 regarding size_zero_node.
20687 2017-08-16 Martin Liska <mliska@suse.cz>
20690 * config.gcc: Respect previously set extra_objs in case
20693 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20695 PR tree-optimization/81835
20696 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20697 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20698 not depend on the phi.
20700 2017-08-16 Alan Modra <amodra@gmail.com>
20702 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20705 2017-08-16 Alan Modra <amodra@gmail.com>
20707 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20708 (save_reg_p): ..into this. Update all callers.
20709 (first_reg_to_save): Simplify.
20711 2017-08-16 Alan Modra <amodra@gmail.com>
20713 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20716 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20720 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20721 constant count if that count is less than 32.
20723 2017-08-15 Nathan Sidwell <nathan@acm.org>
20725 * gcc.c (execute): Emit friendlier message if inferior is killed
20726 by an external cause.
20728 2017-08-15 Richard Biener <rguenther@suse.de>
20730 PR tree-optimization/81790
20731 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20732 CONSTRUCTORs from simplifying and VN.
20734 2017-08-14 Martin Sebor <msebor@redhat.com>
20736 * builtin-attrs.def: Add comments.
20738 2017-08-14 Martin Sebor <msebor@redhat.com>
20741 * doc/extend.texi (attribute nonstring): Document new attribute.
20743 2017-08-14 Martin Sebor <msebor@redhat.com>
20746 * tree-diagnostic.c (default_tree_printer): Handle %G.
20747 * gimple-pretty-print.h (percent_G_format): Declare new function.
20748 * gimple-pretty-print.c (percent_G_format): Define.
20749 * tree-pretty-print.c (percent_K_format): Add argument.
20751 2017-08-14 Martin Sebor <msebor@redhat.com>
20753 PR translation/79998
20754 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20755 Remove a stray space.
20757 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20760 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20761 (*iordi_1_bts): Ditto.
20762 (*xordi_1_btc): Ditto.
20764 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20767 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20768 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20770 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20771 quoted strings, and make more translator-friendly.
20772 (darwin_rs6000_override_options): Likewise.
20773 (rs6000_option_override_internal): Likewise.
20774 (rs6000_return_in_memory): Fix overlong line.
20775 (init_cmulative_args): Use quoted strings, and make more
20776 translator-friendly.
20777 (rs6000_pass_by_reference): Fix overlong line.
20778 (def_builtin): Use quoted strings.
20779 (altivec_expand_predicate_builtin): Use quoted strings, and make
20780 more translator-friendly.
20781 (htm_expand_builtin): Use quoted strings.
20782 (cpu_expand_builtin): Use quoted strings, and make more
20783 translator-friendly.
20784 (altivec_expand_builtin): Likewise.
20785 (paired_expand_predicate_builtin): Likewise.
20786 (rs6000_invalid_builtin): Likewise.
20787 (builtin_function_type): Use quoted strings.
20788 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20789 more translator-friendly.
20790 (rs6000_trampoline_init): Likewise.
20791 (rs6000_handle_altivec_attribute): Likewise.
20792 (rs6000_inner_target_options): Use quoted strings.
20793 (rs6000_disable_incompatible_switches): Likewise.
20794 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20795 strings, and make more translator-friendly.
20796 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20798 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20800 PR tree-optimization/81799
20801 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20802 cond_expr to simple gimple operand.
20804 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20806 PR middle-end/46932
20807 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20809 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20813 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20814 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20815 TARGET_GASISR_PROLOGUES.
20816 * config/avr/avr.c (avr_option_override): Same.
20817 (avr_pass_pre_proep::execute): Same.
20819 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20822 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20823 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20825 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20827 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20828 define_insn_and_split. Split to a memory load from 0 in
20829 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20830 using PTR mode iterator.
20831 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20832 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20833 (*add_tp_<mode>): Redefine as define_insn_and_split.
20834 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20835 address space. Merge with *add_tp_x32 using PTR mode iterator.
20836 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20837 Split to an add with a memory load from 0 in
20838 DEFAULT_TLS_SEG_REG address space.
20840 2017-08-12 Andrew Pinski <apinski@cavium.com>
20842 * config/aarch64/aarch64-option-extensions.def (rdma):
20843 Fix feature string to what Linux prints out in /proc/cpuinfo.
20845 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20848 * dwarf2out.c (modified_type_die): For C typedef types that have
20849 an ultimate origin, process the ultimate origin instead of the
20851 (gen_typedef_die): Assert that input DECLs have no ultimate
20853 (gen_type_die_with_usage): For typedef variants that have an
20854 ultimate origin, just call gen_decl_die on the original DECL.
20855 (process_scope_var): Avoid creating DIEs for local typedefs and
20856 concrete static variables.
20858 2017-08-12 Alan Modra <amodra@gmail.com>
20862 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20863 match gnu-user.h startfile.
20864 (ENDFILE_LINUX_SPEC): Similarly.
20866 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20869 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20871 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20873 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20874 * config/aarch64/aarch64.md (mov<mode>): Change.
20875 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20876 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20877 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20879 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20881 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20882 for storage order barriers.
20884 2017-08-11 Martin Liska <mliska@suse.cz>
20887 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20888 variables of void type.
20890 2017-08-11 Martin Liska <mliska@suse.cz>
20892 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20893 TARGET_SUPPORTS_ALIASES.
20894 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20895 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20896 (optimize_weakref): Likewise.
20897 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20898 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20899 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20901 2017-08-11 Martin Liska <mliska@suse.cz>
20904 * config/i386/i386.c (make_resolver_func): Do complete
20905 refactoring of the function.
20907 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20910 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20911 * config/i386/i386.c (ix86_stack_protect_guard): Use
20912 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20913 * doc/invoke.texi (x86 Options): Document
20914 -mstack-protector-guard-symbol= option.
20916 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20918 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20919 * config/i386/i386.c (ix86_split_stack_guard): New function.
20920 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20921 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20922 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20923 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20924 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20925 (split_stack_space_check): Call ix86_split_stack_guard.
20927 2017-08-10 Martin Sebor <msebor@redhat.com>
20929 * print-tree.c (print_node): Print location using the established
20931 Replace spaces with colons.
20932 (debug_raw, debug): Ditto.
20934 2017-08-10 Martin Sebor <msebor@redhat.com>
20937 * pretty-print.c (pp_format): Correct the handling of %s precision.
20939 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20942 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20944 (ix86_finalize_stack_frame_flags): This. Also clear
20945 frame_pointer_needed if -fno-omit-frame-pointer is used without
20947 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20948 with ix86_finalize_stack_frame_flags.
20949 (ix86_expand_epilogue): Likewise.
20950 (ix86_expand_split_stack_prologue): Likewise.
20951 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20953 2017-08-10 Martin Liska <mliska@suse.cz>
20956 * c-attribs.c (handle_target_attribute):
20957 Report warning for an empty string argument of target attribute.
20959 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20962 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20964 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20965 or DECL_NONLOCAL labels.
20966 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20967 or DECL_NONLOCAL labels here.
20969 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20971 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20972 to indicate when early gimple folding has been disabled.
20973 (rs6000_gimple_fold_builtin): Add debug content.
20974 (rs6000_invalid_builtin): Fix whitespace.
20975 (rs6000_expand_builtin): Fix whitespace.
20976 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20978 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20981 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20982 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20984 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20986 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20987 (qdf24xx): Likewise.
20988 * config/aarch64/aarch64.md: Include falkor.md.
20989 * config/aarch64/falkor.md: New.
20991 2017-08-09 Marek Polacek <polacek@redhat.com>
20994 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20995 * diagnostic.c (emit_diagnostic): Add a comment.
20996 (emit_diagnostic_valist): New function.
20998 2017-08-09 Marek Polacek <polacek@redhat.com>
21001 * input.c (make_location): New overload.
21002 * input.h (make_location): Declare.
21004 2017-08-08 Alan Modra <amodra@gmail.com>
21005 H.J. Lu <hongjiu.lu@intel.com>
21008 * gcc.c (NO_PIE_SPEC): Delete.
21009 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21011 (LINK_PIE_SPEC): ..to here.
21012 (LINK_COMMAND_SPEC): Support -no-pie.
21013 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21014 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21016 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21017 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21018 (ENDFILE_CRTEND_SPEC): Similarly.
21020 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21023 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21024 (mstack-protector-guard-offset=): Ditto.
21025 * config/i386/i386.c (ix86_option_override): Handle
21026 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21028 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21029 ix86_stack_protect_guard_offset variables.
21030 (TARGET_STACK_PROTECT_GUARD): Always define.
21031 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21032 and -mstack-protector-guard-offset= options.
21034 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21036 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21037 boundary case for the last candidate.
21039 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21041 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21042 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21044 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21046 PR middle-end/19706
21047 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21048 * config/aarch64/aarch64-builtins.c
21049 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21050 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21051 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21053 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21054 Andrew Pinski <pinskia@gmail.com>
21056 PR middle-end/19706
21057 * internal-fn.def (XORSIGN): New.
21058 * optabs.def (xorsign_optab): New.
21059 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21060 (convert_expand_mult_copysign): New.
21061 (pass_optimize_widening_mul::execute): Call
21062 convert_expand_mult_copysign.
21064 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21066 PR tree-optimization/81354
21067 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21068 Insert on edges rather than explicitly creating landing pads.
21069 (analyze_candidates_and_replace): Commit edge inserts.
21071 2017-08-08 Richard Biener <rguenther@suse.de>
21073 PR middle-end/81719
21074 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21075 (expand_simple_operations): Also look through ADDR_EXPRs with
21076 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21078 2017-08-08 Richard Biener <rguenther@suse.de>
21080 PR tree-optimization/81723
21081 * tree-vect-slp.c (struct bst_traits): New hash traits.
21082 (bst_fail): New global.
21083 (vect_build_slp_tree_2): New worker, split out from ...
21084 (vect_build_slp_tree): ... this now wrapping it with using
21085 bst_fail set to cache SLP tree build fails. Properly handle
21087 (vect_analyze_slp_instance): Allocate and free bst_fail.
21089 2017-08-08 Martin Liska <mliska@suse.cz>
21092 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21093 LABEL_DECLs that can be from a different function.
21095 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21097 PR tree-optimization/81744
21098 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21099 loop's number of iterations.
21101 2017-08-08 Martin Liska <mliska@suse.cz>
21103 * asan.c: Include header files.
21104 * attribs.c (build_decl_attribute_variant): New function moved
21106 (build_type_attribute_qual_variant): Likewise.
21107 (cmp_attrib_identifiers): Likewise.
21108 (simple_cst_list_equal): Likewise.
21109 (omp_declare_simd_clauses_equal): Likewise.
21110 (attribute_value_equal): Likewise.
21111 (comp_type_attributes): Likewise.
21112 (build_type_attribute_variant): Likewise.
21113 (lookup_ident_attribute): Likewise.
21114 (remove_attribute): Likewise.
21115 (merge_attributes): Likewise.
21116 (merge_type_attributes): Likewise.
21117 (merge_decl_attributes): Likewise.
21118 (merge_dllimport_decl_attributes): Likewise.
21119 (handle_dll_attribute): Likewise.
21120 (attribute_list_equal): Likewise.
21121 (attribute_list_contained): Likewise.
21122 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21123 (lookup_attribute_by_prefix): Likewise.
21124 * bb-reorder.c: Include header files.
21125 * builtins.c: Likewise.
21126 * calls.c: Likewise.
21127 * cfgexpand.c: Likewise.
21128 * cgraph.c: Likewise.
21129 * cgraphunit.c: Likewise.
21130 * convert.c: Likewise.
21131 * dwarf2out.c: Likewise.
21132 * final.c: Likewise.
21133 * fold-const.c: Likewise.
21134 * function.c: Likewise.
21135 * gimple-expr.c: Likewise.
21136 * gimple-fold.c: Likewise.
21137 * gimple-pretty-print.c: Likewise.
21138 * gimple.c: Likewise.
21139 * gimplify.c: Likewise.
21140 * hsa-common.c: Likewise.
21141 * hsa-gen.c: Likewise.
21142 * internal-fn.c: Likewise.
21143 * ipa-chkp.c: Likewise.
21144 * ipa-cp.c: Likewise.
21145 * ipa-devirt.c: Likewise.
21146 * ipa-fnsummary.c: Likewise.
21147 * ipa-inline.c: Likewise.
21148 * ipa-visibility.c: Likewise.
21150 * lto-cgraph.c: Likewise.
21151 * omp-expand.c: Likewise.
21152 * omp-general.c: Likewise.
21153 * omp-low.c: Likewise.
21154 * omp-offload.c: Likewise.
21155 * omp-simd-clone.c: Likewise.
21156 * opts-global.c: Likewise.
21157 * passes.c: Likewise.
21158 * predict.c: Likewise.
21159 * sancov.c: Likewise.
21160 * sanopt.c: Likewise.
21161 * symtab.c: Likewise.
21162 * toplev.c: Likewise.
21163 * trans-mem.c: Likewise.
21164 * tree-chkp.c: Likewise.
21165 * tree-eh.c: Likewise.
21166 * tree-into-ssa.c: Likewise.
21167 * tree-object-size.c: Likewise.
21168 * tree-parloops.c: Likewise.
21169 * tree-profile.c: Likewise.
21170 * tree-ssa-ccp.c: Likewise.
21171 * tree-ssa-live.c: Likewise.
21172 * tree-ssa-loop.c: Likewise.
21173 * tree-ssa-sccvn.c: Likewise.
21174 * tree-ssa-structalias.c: Likewise.
21175 * tree-ssa.c: Likewise.
21176 * tree-streamer-in.c: Likewise.
21177 * tree-vectorizer.c: Likewise.
21178 * tree-vrp.c: Likewise.
21179 * tsan.c: Likewise.
21180 * ubsan.c: Likewise.
21181 * varasm.c: Likewise.
21182 * varpool.c: Likewise.
21183 * tree.c: Remove functions moved to attribs.[ch].
21184 * tree.h: Likewise.
21185 * config/aarch64/aarch64.c: Add attrs.h header file.
21186 * config/alpha/alpha.c: Likewise.
21187 * config/arc/arc.c: Likewise.
21188 * config/arm/arm.c: Likewise.
21189 * config/avr/avr.c: Likewise.
21190 * config/bfin/bfin.c: Likewise.
21191 * config/c6x/c6x.c: Likewise.
21192 * config/cr16/cr16.c: Likewise.
21193 * config/cris/cris.c: Likewise.
21194 * config/darwin.c: Likewise.
21195 * config/epiphany/epiphany.c: Likewise.
21196 * config/fr30/fr30.c: Likewise.
21197 * config/frv/frv.c: Likewise.
21198 * config/ft32/ft32.c: Likewise.
21199 * config/h8300/h8300.c: Likewise.
21200 * config/i386/winnt.c: Likewise.
21201 * config/ia64/ia64.c: Likewise.
21202 * config/iq2000/iq2000.c: Likewise.
21203 * config/lm32/lm32.c: Likewise.
21204 * config/m32c/m32c.c: Likewise.
21205 * config/m32r/m32r.c: Likewise.
21206 * config/m68k/m68k.c: Likewise.
21207 * config/mcore/mcore.c: Likewise.
21208 * config/microblaze/microblaze.c: Likewise.
21209 * config/mips/mips.c: Likewise.
21210 * config/mmix/mmix.c: Likewise.
21211 * config/mn10300/mn10300.c: Likewise.
21212 * config/moxie/moxie.c: Likewise.
21213 * config/msp430/msp430.c: Likewise.
21214 * config/nds32/nds32-isr.c: Likewise.
21215 * config/nds32/nds32.c: Likewise.
21216 * config/nios2/nios2.c: Likewise.
21217 * config/nvptx/nvptx.c: Likewise.
21218 * config/pa/pa.c: Likewise.
21219 * config/pdp11/pdp11.c: Likewise.
21220 * config/powerpcspe/powerpcspe.c: Likewise.
21221 * config/riscv/riscv.c: Likewise.
21222 * config/rl78/rl78.c: Likewise.
21223 * config/rx/rx.c: Likewise.
21224 * config/s390/s390.c: Likewise.
21225 * config/sh/sh.c: Likewise.
21226 * config/sol2.c: Likewise.
21227 * config/sparc/sparc.c: Likewise.
21228 * config/spu/spu.c: Likewise.
21229 * config/stormy16/stormy16.c: Likewise.
21230 * config/tilegx/tilegx.c: Likewise.
21231 * config/tilepro/tilepro.c: Likewise.
21232 * config/v850/v850.c: Likewise.
21233 * config/vax/vax.c: Likewise.
21234 * config/visium/visium.c: Likewise.
21235 * config/xtensa/xtensa.c: Likewise.
21237 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21240 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21241 constraints since the -mupper-regs-* switches have been
21243 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21244 into a vector from a double word element that was extracted from
21245 another vector, and eliminate extra XXPERMDI instructions.
21246 (vsx_concat_<mode>_2): Likewise.
21247 (vsx_concat_<mode>_3): Likewise.
21248 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21249 concat to allow optimizing inserts from previous extracts.
21251 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21253 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21254 memory reference to a SSP offset in TLS address space.
21255 (ix86_print_operand) <case '@'>: Remove.
21256 (ix86_print_operand_punct_valid_p): Remove '@' code.
21257 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21258 UNSPEC_SP_TLS_TEST.
21259 (stack_tls_protect_set_<mode>): Remove.
21260 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21261 (stack_tls_protect_test_<mode>): Remove.
21262 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21264 2017-08-07 Olivier Hainque <hainque@adacore.com>
21267 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21269 2017-08-07 Douglas Rupp <rupp@adacore.com>
21271 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21272 variable was referenced as multidir in command.
21274 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21277 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21280 2017-08-07 Martin Liska <mliska@suse.cz>
21282 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21283 * config/rl78/rl78.c: Add include of attribs.h.
21284 * config/sh/sh.c: Likewise.
21285 * config/v850/v850.c: Likewise.
21287 2017-08-07 Tom de Vries <tom@codesourcery.com>
21289 PR middle-end/78266
21290 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21292 2017-08-07 Martin Liska <mliska@suse.cz>
21294 * config/mips/mips.c: Include attribs.h.
21296 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21299 * doc/invoke.texi: Document change in behvaior for -Ofast for
21302 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21304 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21306 (aarch64_pop_regs): Likewise.
21307 (aarch64_gen_load_pair): Likewise.
21308 (aarch64_save_callee_saves): Likewise.
21309 (aarch64_restore_callee_saves): Likewise.
21311 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21313 * config/i386/i386.c: Revert the last change.
21315 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21318 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21320 (ix86_finalize_stack_frame_flags): This. Also clear
21321 frame_pointer_needed if -fno-omit-frame-pointer is used without
21323 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21324 with ix86_finalize_stack_frame_flags.
21325 (ix86_expand_epilogue): Likewise.
21326 (ix86_expand_split_stack_prologue): Likewise.
21328 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21331 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21332 to P_AES for Westmere.
21334 2017-08-07 Jonathan Yong <10walls@gmail.com>
21336 * config/i386/mingw.opt (fset-stack-executable): Removed.
21337 * config/i386/cygming.opt (fset-stack-executable): Moved
21339 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21341 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21343 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21345 2017-08-07 Marek Polacek <polacek@redhat.com>
21347 PR middle-end/81737
21348 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21350 2017-08-07 Martin Liska <mliska@suse.cz>
21352 * attribs.h (canonicalize_attr_name): New function.
21353 (cmp_attribs): Move from c-format.c and adjusted.
21354 (is_attribute_p): Moved from tree.h.
21355 * tree-inline.c: Add new includes.
21356 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21357 (private_is_attribute_p): Remove.
21358 (private_lookup_attribute): Likewise.
21359 (private_lookup_attribute_by_prefix): Simplify.
21360 (remove_attribute): Use is_attribute_p.
21361 * tree.h: Remove removed declarations.
21363 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21365 PR middle-end/81698
21366 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21367 instead of computing it in the function. Formatting fix.
21368 (expand_case): Don't rely on default_edge being the first edge,
21369 clear it if removing it, pass default_edge to
21370 emit_case_dispatch_table.
21371 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21374 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21376 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21377 insn in the function, emit NOP after the insn.
21379 2017-08-06 Tom de Vries <tom@codesourcery.com>
21381 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21384 2017-08-06 Tom de Vries <tom@codesourcery.com>
21386 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21389 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21391 PR tree-optimization/57371
21392 * match.pd: New pattern.
21394 2017-08-04 Marek Polacek <polacek@redhat.com>
21396 PR middle-end/81695
21397 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21398 perform the computation in offset_int.
21400 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21402 PR tree-optimization/81136
21403 * tree-vectorizer.h: Include tree-hash-traits.h.
21404 (vec_base_alignments): New typedef.
21405 (vec_info): Add a base_alignments field.
21406 (vect_record_base_alignments): Declare.
21407 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21409 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21410 (create_data_ref): Add an is_conditional_in_stmt argument.
21411 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21412 the is_conditional_in_stmt field.
21413 (data_ref_loc): Add an is_conditional_in_stmt field.
21414 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21415 (find_data_references_in_stmt): Update call to create_data_ref.
21416 (graphite_find_data_references_in_stmt): Likewise.
21417 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21418 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21419 (vect_record_base_alignment): New function.
21420 (vect_record_base_alignments): Likewise.
21421 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21422 for nested statements even if we fail to compute a misalignment.
21423 Use pooled base alignments for unconditional references.
21424 (vect_find_same_alignment_drs): Compare base addresses instead
21426 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21427 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21429 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21431 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21432 Add an explicit name for the enum. Use auto_vec for slp_instances
21433 and grouped_stores.
21434 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21436 (_bb_vec_info): Add a constructor and destructor.
21437 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21438 (destroy_loop_vec_info): Delete.
21439 (vect_destroy_datarefs): Likewise.
21440 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21441 (vec_info::vec_info): New function.
21442 (vec_info::~vec_info): Likewise.
21443 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21444 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21445 destroy_loop_vec_info.
21446 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21447 (_loop_vec_info::_loop_vec_info): ...this.
21448 (destroy_loop_vec_info): Replace with...
21449 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21450 the stmt_vec_infos. Leave handling of vec_info information to its
21451 destructor. Remove explicit vector releases.
21452 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21453 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21454 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21455 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21456 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21457 (destroy_bb_vec_info): Replace with...
21458 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21459 information to its destructor.
21460 (vect_slp_analyze_bb_1): Use new and delete instead of
21461 new_bb_vec_info and destroy_bb_vec_info.
21462 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21465 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21467 * tree-data-ref.h (subscript): Add access_fn field.
21468 (data_dependence_relation): Add could_be_independent_p.
21469 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21470 (same_access_functions): Move to tree-data-ref.c.
21471 * tree-data-ref.c (ref_contains_union_access_p): New function.
21472 (access_fn_component_p): Likewise.
21473 (access_fn_components_comparable_p): Likewise.
21474 (dr_analyze_indices): Add a reference to access_fn_component_p.
21475 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21477 (constant_access_functions): Likewise.
21478 (add_other_self_distances): Likewise.
21479 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21480 (initialize_data_dependence_relation): Use XCNEW and remove
21481 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21482 of access functions that have the same type. Allow the
21483 subsequence to end with different bases in some circumstances.
21484 Record the chosen access functions in SUB_ACCESS_FN.
21485 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21486 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21487 (subscript_dependence_tester_1): Likewise dra and drb.
21488 (build_classic_dist_vector): Update calls accordingly.
21489 (subscript_dependence_tester): Likewise.
21490 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21491 DDR_COULD_BE_INDEPENDENT_P.
21492 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21493 comp_alias_ddrs instead of may_alias_ddrs.
21494 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21496 (vect_analyze_data_ref_dependence): Use it if
21497 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21498 distance vectors if that fails.
21499 (dependence_distance_ge_vf): New function.
21500 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21501 LOOP_VINFO_MAY_ALIAS_DDRS.
21503 2017-08-04 Richard Biener <rguenther@suse.de>
21505 PR middle-end/81705
21506 * fold-const.c (fold_binary_loc): Properly restrict
21507 minus_var0 && minus_var1 case when associating undefined overflow
21510 2017-08-04 Tom de Vries <tom@codesourcery.com>
21512 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21514 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21516 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21517 Don't start diagnostic messages with a capital letter.
21518 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21520 (rs6000_invalid_builtin): Likewise.
21521 (rs6000_trampoline_init): Likewise.
21523 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21526 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21527 after setting changeable df flags.
21529 2017-08-03 Richard Biener <rguenther@suse.de>
21531 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21532 up if the use is in USE - X.
21534 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21536 * toplev.c (dumpfile.h): New include.
21537 (internal_error_reentered): New static function. Use it...
21538 (internal_error_function): ...here to handle reentered internal_error.
21540 2017-08-03 Richard Biener <rguenther@suse.de>
21542 PR middle-end/81148
21543 * fold-const.c (split_tree): Add minus_var and minus_con
21544 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21545 here but always use minus_*.
21546 (associate_trees): Assert we never associate with MINUS_EXPR
21547 and NULL first operand. Do not recurse for PLUS_EXPR operands
21548 when associating as MINUS_EXPR either.
21549 (fold_binary_loc): Track minus_var and minus_con.
21551 2017-08-03 Tom de Vries <tom@codesourcery.com>
21554 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21555 ACCEL_COMPILER, apply finish_options on
21556 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21558 2017-08-03 Tom de Vries <tom@codesourcery.com>
21561 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21562 function_entry_patch_area_size > 0.
21564 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21567 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21568 instead of 10??LU, perform unit multiplication in wide_int,
21569 don't change alloc_object_size_limit if the limit is larger
21572 PR tree-optimization/81655
21573 PR tree-optimization/81588
21574 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21575 the case when ranges[i].low and high are 1 for unsigned type with
21578 PR middle-end/81052
21579 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21580 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21582 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21584 * tree-vrp.h: Add include guard.
21586 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21589 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21590 (ud2): New insn pattern.
21591 * config/i386/i386.c (ix86_expand_epilogue):
21592 For naked functions, generate ud2 instead of trap insn.
21594 2017-08-02 Marek Polacek <polacek@redhat.com>
21597 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21599 2017-08-02 Tom de Vries <tom@codesourcery.com>
21600 Cesar Philippidis <cesar@codesourcery.com>
21602 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21603 Add missing edge probabilities.
21605 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21607 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21608 Correct endianness.
21610 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21612 PR middle-end/79499
21613 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21614 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21615 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21616 prologue_seq sequences - if any.
21618 2017-08-02 Richard Biener <rguenther@suse.de>
21620 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21621 via vectors if supported, integer extracts via punning if supported
21622 or otherwise vector extracts.
21624 2017-08-02 Richard Biener <rguenther@suse.de>
21626 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21628 (bitmap_insert_into_set): ... this.
21630 2017-08-02 Richard Biener <rguenther@suse.de>
21632 PR tree-optimization/81633
21634 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21636 PR tree-optimization/71752
21637 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21639 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21641 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21642 (machine_function::call_ms2sysv_pad_out): Remove field.
21643 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21644 (ix86_compute_frame_layout): Likewise.
21646 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21649 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21650 attribute with interrupt attribute.
21652 2017-08-01 Andrew Pinski <apinski@cavium.com>
21654 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21655 BIT_INSERT_EXPR's operand 1
21656 to see if the types precision matches.
21658 2017-08-01 Martin Liska <mliska@suse.cz>
21660 PR middle-end/70140
21661 * builtins.c (expand_builtin_memcpy_args): Remove.
21662 (expand_builtin_memcpy): Call newly added function
21663 expand_builtin_memory_copy_args.
21664 (expand_builtin_memcpy_with_bounds): Likewise.
21665 (expand_builtin_mempcpy): Remove last argument.
21666 (expand_builtin_mempcpy_with_bounds): Likewise.
21667 (expand_builtin_memory_copy_args): New function created from
21668 expand_builtin_mempcpy_args with small modifications.
21669 (expand_builtin_mempcpy_args): Remove.
21670 (expand_builtin_stpcpy): Remove unused argument.
21671 (expand_builtin): Likewise.
21672 (expand_builtin_with_bounds): Likewise.
21674 2017-08-01 Martin Liska <mliska@suse.cz>
21677 Make mempcpy more optimal (PR middle-end/70140).
21679 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21682 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21683 __builtin_vec_cmpne verify both arguments are compatible vectors
21684 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21685 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21686 move computation of aligned to after checking the argument types.
21690 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21691 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21694 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21695 Doug Rupp <rupp@adacore.com>
21696 Olivier Hainque <hainque@adacore.com>
21698 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21699 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21701 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21702 for TARGET_OS_CPP_BUILTIN.
21703 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21704 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21706 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21707 passing required abi options to the assembler for EABI configurations.
21708 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21709 of .text.hot and .text.unlikely sections for kernel modules when
21710 using ARM style exceptions.
21711 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21712 options. Add EXTRA_CC1_SPEC.
21713 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21715 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21717 (ARM_TARGET2_DWARF_FORMAT): Define.
21718 * config/arm/t-vxworks: Adjust multilib control to removal of the
21719 Diab command line options.
21721 2017-08-01 Martin Liska <mliska@suse.cz>
21723 PR gcov-profile/81561
21724 * gcov.c (unblock): Make unblocking safe as we need to preserve
21725 index correspondence of blocks and block_lists.
21727 2017-08-01 Richard Biener <rguenther@suse.de>
21729 PR tree-optimization/81181
21730 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21731 (compute_antic): ... end of iteration here.
21733 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21735 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21736 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21737 (ftree-slp-vectorize): Likewise.
21738 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21739 can no longer be set independent of flag_tree_loop_vectorize.
21740 * omp-general.c (emp_max_vf): Likewise.
21741 * opts.c (enable_fdo_optimizations): Remove references to
21742 flag_tree_vectorize, these are now implicit.
21743 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21744 and leave it for the options machinery.
21746 2017-08-01 Martin Liska <mliska@suse.cz>
21748 PR middle-end/70140
21749 * builtins.c (expand_builtin_memcpy_args): Remove.
21750 (expand_builtin_memcpy): Call newly added function
21751 expand_builtin_memory_copy_args.
21752 (expand_builtin_memcpy_with_bounds): Likewise.
21753 (expand_builtin_mempcpy): Remove last argument.
21754 (expand_builtin_mempcpy_with_bounds): Likewise.
21755 (expand_builtin_memory_copy_args): New function created from
21756 expand_builtin_mempcpy_args with small modifications.
21757 (expand_builtin_mempcpy_args): Remove.
21758 (expand_builtin_stpcpy): Remove unused argument.
21759 (expand_builtin): Likewise.
21760 (expand_builtin_with_bounds): Likewise.
21762 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21765 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21766 print "ds:" only for immediates in generic address space.
21768 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21771 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21772 (ix86_function_ok_for_sibcall): Return false for naked functions.
21774 2017-08-01 Richard Biener <rguenther@suse.de>
21776 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21777 (compute_antic): Seed worklist with exit block predecessors.
21778 * cfganal.c (dfs_find_deadend): For a cycle return the source
21779 of the edge closing it.
21781 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21783 * config/aarch64/aarch64.c
21784 (aarch64_can_const_movi_rtx_p): Move 0 check.
21786 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21788 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21789 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21791 * match.pd: Ditto in address comparison pattern.
21793 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21795 PR tree-optimization/81627
21796 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21797 closed ssa form for store-store chain.
21799 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21801 PR tree-optimization/81620
21802 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21803 for store-store chain.
21805 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21807 PR tree-optimization/81588
21808 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21809 ranges[i].in_p, invert comparison code ccode. For >/>=,
21810 swap rhs1 and rhs2 and comparison code unconditionally,
21811 for </<= don't do that. Don't swap rhs1/rhs2 again if
21812 ranges[i].in_p, instead invert comparison code ccode if
21813 opcode or oe->rank is BIT_IOR_EXPR.
21816 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21817 a direct optab to conversion optab.
21818 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21819 with GET_MODE_INNER as last argument instead of optab_handler.
21820 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21821 vector extraction if possible and optab is available.
21822 * expr.c (store_constructor): Use convert_optab_handler instead
21823 of optab_handler. Use vector initialization from smaller
21824 vectors if possible and optab is available.
21825 * tree-vect-stmts.c (vectorizable_load): Likewise.
21826 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21827 now have two modes.
21828 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21829 of vec_init from half-sized vectors with the same element mode.
21830 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21831 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21832 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21833 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21834 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21835 after mode in gen_vec_extract* calls.
21836 (vec_extract<mode>): Renamed to ...
21837 (vec_extract<mode><ssescalarmodelower>): ... this.
21838 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21839 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21840 element mode after mode in gen_vec_init* calls.
21841 (VEC_INIT_HALF_MODE): New mode iterator.
21842 (vec_init<mode>): Renamed to ...
21843 (vec_init<mode><ssescalarmodelower>): ... this.
21844 (vec_init<mode><ssehalfvecmodelower>): New expander.
21845 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21846 (vec_extractv2sfsf): ... this.
21847 (vec_initv2sf): Renamed to ...
21848 (vec_initv2sfsf): ... this.
21849 (vec_extractv2si): Renamed to ...
21850 (vec_extractv2sisi): ... this.
21851 (vec_initv2si): Renamed to ...
21852 (vec_initv2sisi): ... this.
21853 (vec_extractv4hi): Renamed to ...
21854 (vec_extractv4hihi): ... this.
21855 (vec_initv4hi): Renamed to ...
21856 (vec_initv4hihi): ... this.
21857 (vec_extractv8qi): Renamed to ...
21858 (vec_extractv8qiqi): ... this.
21859 (vec_initv8qi): Renamed to ...
21860 (vec_initv8qiqi): ... this.
21861 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21862 (vec_init<mode>): Renamed to ...
21863 (vec_init<mode><VEC_base_l>): ... this.
21864 (vec_extract<mode>): Renamed to ...
21865 (vec_extract<mode><VEC_base_l>): ... this.
21866 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21867 (vec_initv2sfsf): ... this.
21868 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21869 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21870 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21871 element mode after mode in gen_vec_init* calls.
21872 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21873 (vec_init<mode><Vel>): ... this.
21874 (vec_extract<mode>): Renamed to ...
21875 (vec_extract<mode><Vel>): ... this.
21876 * config/aarch64/iterators.md (Vel): New mode attribute.
21877 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21878 Add element mode after mode in gen_vec_extract* calls.
21879 * config/s390/vector.md (non_vec_l): New mode attribute.
21880 (vec_extract<mode>): Renamed to ...
21881 (vec_extract<mode><non_vec_l>): ... this.
21882 (vec_init<mode>): Renamed to ...
21883 (vec_init<mode><non_vec_l>): ... this.
21884 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21885 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21887 * config/arm/iterators.md (V_elem_l): New mode attribute.
21888 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21889 (vec_extract<mode><V_elem_l>): ... this.
21890 (vec_extractv2di): Renamed to ...
21891 (vec_extractv2didi): ... this.
21892 (vec_init<mode>): Renamed to ...
21893 (vec_init<mode><V_elem_l>): ... this.
21894 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21895 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21896 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21897 Add element mode after gen_vec_extract* calls.
21898 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21899 (vec_init<mode><unitmode>): ... this.
21900 (vec_extract<mode>): Renamed to ...
21901 (vec_extract<mode><unitmode>): ... this.
21902 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21903 (vec_init<mode><unitmode>): ... this.
21904 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21905 (vec_initv2sfsf): ... this.
21906 (vec_extractv2sf): Renamed to ...
21907 (vec_extractv2sfsf): ... this.
21908 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21909 Add element mode after gen_vec_extract* calls.
21910 * config/mips/mips.md (unitmode): New mode iterator.
21911 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21912 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21913 * config/spu/spu.md (inner_l): New mode attribute.
21914 (vec_init<mode>): Renamed to ...
21915 (vec_init<mode><inner_l>): ... this.
21916 (vec_extract<mode>): Renamed to ...
21917 (vec_extract<mode><inner_l>): ... this.
21918 * config/sparc/sparc.md (veltmode): New mode iterator.
21919 (vec_init<VMALL:mode>): Renamed to ...
21920 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21921 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21922 (vec_initv2sisi): ... this.
21923 (vec_initv2sf): Renamed to ...
21924 (vec_initv2sfsf): ... this.
21925 (vec_extractv2sf): Renamed to ...
21926 (vec_extractv2sfsf): ... this.
21927 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21928 (vec_init<mode>): Renamed to ...
21929 (vec_init<mode><VEC_base_l>): ... this.
21930 (vec_extract<mode>): Renamed to ...
21931 (vec_extract<mode><VEC_base_l>): ... this.
21932 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21933 (vec_initv2sfsf): ... this.
21934 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21935 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21936 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21937 gen_vec_init* calls.
21939 2017-08-01 Richard Biener <rguenther@suse.de>
21941 PR tree-optimization/81297
21942 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21943 TREE_OVERFLOW from INTEGER_CSTs.
21945 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21947 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21949 2017-07-31 Carl Love <cel@us.ibm.com>
21951 * config/rs6000/rs6000-c: Add support for built-in functions
21952 vector signed char vec_xl_be (signed long long, signed char *);
21953 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21954 vector signed int vec_xl_be (signed long long, signed int *);
21955 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21956 vector signed long long vec_xl_be (signed long long, signed long long *);
21957 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21958 vector signed short vec_xl_be (signed long long, signed short *);
21959 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21960 vector double vec_xl_be (signed long long, double *);
21961 vector float vec_xl_be (signed long long, float *);
21962 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21963 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21964 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21966 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21967 (altivec_expand_builtin): Add switch statement to call
21968 altivec_expand_xl_be for each builtin.
21969 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21970 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21971 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21972 __builtin_vsx_le_be_v16qi.
21973 * doc/extend.texi: Update the built-in documentation file for the
21974 new built-in functions.
21976 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21979 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21981 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21983 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21985 * config.gcc: Add z14.
21986 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21987 CPU model numbers for z13s and z14.
21988 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21990 * config/s390/s390-opts.h (enum processor_type): Rename
21991 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21992 * config/s390/s390.c (processor_table): Add field for CPU name to
21993 be passed to Binutils.
21994 (s390_asm_output_machine_for_arch): Use the new field in
21995 processor_table for Binutils.
21996 (s390_expand_builtin): Replace arch12 with z14.
21997 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21998 (s390_get_sched_attrmask): Likewise.
21999 (s390_get_unit_mask): Likewise.
22000 * config/s390/s390.opt: Add z14 to processor_type enum.
22002 2017-07-31 Martin Jambor <mjambor@suse.cz>
22005 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22006 regardless of optimization level.
22008 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22009 Martin Liska <mliska@suse.cz>
22011 * predict.def: Remove old comment and adjust probability.
22012 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22013 PREDICT statements.
22015 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22018 * config/i386/i386.c (ix86_function_naked): New function.
22019 (ix86_can_use_return_insn_p): Return false for naked functions.
22020 (ix86_expand_prologue): Skip prologue for naked functions.
22021 (ix86_expand_epilogue): Skip epilogue for naked functions
22022 and emit trap instruction.
22023 (ix86_warn_func_return): New function.
22024 (ix86_attribute_table): Add "naked" attribute specification.
22025 (TARGET_WARN_FUNC_RETURN): Define.
22026 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22028 2017-07-31 Martin Liska <mliska@suse.cz>
22030 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22031 (dump_gimple_bb_header): Always dump BB info.
22032 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22034 2017-07-31 Martin Liska <mliska@suse.cz>
22037 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22038 also with current_function_decl non-null equality.
22040 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22043 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22044 change type to the element type, instead add eltype variable and
22045 use it where we are interested in the element type.
22047 PR tree-optimization/81603
22048 * ipa-polymorphic-call.c
22049 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22050 offset arithmetic in offset_int, bail out if the resulting bit offset
22051 doesn't fit into shwi.
22053 2017-07-31 Martin Liska <mliska@suse.cz>
22055 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22056 (gimplify_save_expr): Fix comment.
22058 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22061 * config/i386/i386.c (ix86_function_arg): Update arguments for
22063 (ix86_compute_frame_layout): Set the initial stack offset to
22064 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22065 INCOMING_FRAME_SP_OFFSET.
22066 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22067 stack before exception handler returns.
22068 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22069 the 'ERROR_CODE' for exception handler.
22071 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22073 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22074 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22075 (ASM_OUTPUT_REG_POP): Ditto.
22076 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22077 instead of asm_fprintf to output pure string.
22079 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22081 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22082 to imported_module_or_decl hook.
22083 (debug_nothing_tree_tree_tree_bool): Remove.
22084 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22085 * debug.c (do_nothing_debug_hooks): Use
22086 debug_nothing_tree_tree_tree_bool_bool instead of
22087 debug_nothing_tree_tree_tree_bool.
22088 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22089 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22090 * sdbout.c (sdb_debug_hooks): Likewise.
22091 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22092 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22094 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22095 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22096 attribute, don't add anything.
22098 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22100 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22101 (fold_build2_stat_loc): Likewise.
22102 (fold_build3_stat_loc): Likewise.
22103 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22104 (fold_build1_loc): Remove macro.
22105 (fold_build2_loc): Likewise.
22106 (fold_build3_loc): Likewise.
22108 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22110 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22111 (gimple_build_debug_bind_source_stat): Likewise.
22112 * gimple.h (gimple_build_debug_bind): Remove macro.
22113 (gimple_build_debug_bind_source): Likewise.
22115 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22117 * bitmap.c (bitmap_alloc): Adjust.
22118 (bitmap_gc_alloc): Likewise.
22119 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22121 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22123 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22124 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22125 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22126 (bitmap_gc_alloc_stat): Likewise.
22127 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22129 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22131 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22132 * rtl.h (shallow_copy_rtx): Remove macro.
22134 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22136 * emit-rtl.c (gen_raw_REG): Adjust.
22137 * gengenrtl.c (gendef): Likewise.
22138 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22139 * rtl.h (rtx_alloc): Remove macro.
22141 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22143 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22144 (build_tree_list_stat): Likewise.
22145 * tree.h (build_tree_list): Remove macro.
22146 (build_tree_list_vec): Likewise.
22148 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22150 * tree.c (make_vector_stat): Remove _stat from name.
22151 (build_vector_stat): Likewise.
22152 * tree.h (make_vector_stat): Remove macro.
22153 (build_vector_stat): Likewise.
22155 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22157 * tree.h (build_var_debug_value): Remove prototype.
22159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22161 * tree.c (tree_cons_stat): Remove _stat from name.
22162 * tree.h (tree_cons): Remove macro.
22164 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22166 * tree.c (build_vl_exp_stat): Remove _stat from name.
22167 * tree.h (build_vl_exp): Remove macro.
22169 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22171 * tree.c (build_decl_stat): Remove _stat from name.
22172 * tree.h (build_decl): Remove macro.
22174 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22176 * gimple.c (gimple_build_with_ops_stat): Adjust.
22177 (gimple_alloc_stat): Remove _stat from name.
22178 * gimple.h (gimple_alloc): Remove macro.
22180 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22182 * tree.c (make_tree_vec_stat): Remove _stat from name.
22183 (grow_tree_vec_stat): Likewise.
22184 * tree.h (make_tree_vec_stat): Adjust prototype.
22185 (grow_tree_vec_stat): Likewise.
22186 (make_tree_vec): Remove macro.
22187 (grow_tree_vec): Likewise.
22189 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22191 * fold-const.c (fold_build1_stat_loc): Adjust.
22192 (fold_build2_stat_loc): Likewise.
22193 (fold_build3_stat_loc): Likewise.
22194 * tree.c (build0_stat): Remove _stat from name.
22195 (build1_stat): Likewise.
22196 (build2_stat): Likewise.
22197 (build3_stat): Likewise.
22198 (build4_stat): Likewise.
22199 (build5_stat): Likewise.
22200 * tree.h (build1_loc): Remove macro, and rename _stat function
22202 (build2_loc): Likewise.
22203 (build3_loc): Likewise.
22204 (build4_loc): Likewise.
22205 (build5_loc): Likewise.
22207 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22209 * tree.c (make_int_cst_stat): Remove _stat from name.
22210 * tree.h (make_int_cst_stat): Adjust prototype.
22211 (make_int_cst): Remove macro.
22213 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22215 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22216 * tree.h (make_tree_binfo_stat): Adjust prototype.
22217 (make_tree_binfo): Remove.
22219 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22221 * tree.c (copy_node_stat): Rename to copy_node.
22222 (build_distinct_type_copy): Adjust.
22223 * tree.h (copy_node_stat): Adjust prototype.
22224 (copy_node): Remove macro.
22226 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22228 * tree.c (make_node_stat): rename to make_node.
22229 (build_tree_list_stat): Adjust.
22230 (build0_stat): Likewise.
22231 (build2_stat): Likewise.
22232 (build3_stat): Likewise.
22233 (build4_stat): Likewise.
22234 (build5_stat): Likewise.
22235 (build_decl_stat): Likewise.
22236 * tree.h (make_node_stat): Adjust prototype.
22237 (make_node): remove macro.
22239 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22241 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22242 (PPC_FEATURE2_SCV): Likewise.
22243 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22245 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22247 * config/aarch64/aarch64.c
22248 (aarch64_internal_mov_immediate): Add new special pattern.
22249 * config/aarch64/aarch64.md (*movdi_aarch64):
22250 Add reg/32bit const mov case.
22252 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22253 Richard Sandiford <richard.sandiford@linaro.org>
22255 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22256 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22257 Add integer and movi cases.
22258 (movi-split-hf-df-sf split, fp16): New.
22259 (enabled): Added TARGET_FP_F16INST.
22260 * config/aarch64/iterators.md (GPF_HF): New.
22261 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22263 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22265 * config/aarch64/aarch64.c
22266 (aarch64_simd_container_mode): Add prototype.
22267 (aarch64_expand_mov_immediate): Add HI support.
22268 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22269 (aarch64_can_const_movi_rtx_p): New.
22270 (aarch64_preferred_reload_class):
22271 Remove restrictions of using FP registers for certain SIMD operations.
22272 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22273 (aarch64_valid_floating_const): Add integer move validation.
22274 (aarch64_simd_imm_scalar_p): Remove.
22275 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22276 (aarch64_legitimate_constant_p): Expand list of supported cases.
22277 * config/aarch64/aarch64-protos.h
22278 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22279 (aarch64_reinterpret_float_as_int): New.
22280 (aarch64_simd_imm_scalar_p): Remove.
22281 * config/aarch64/constraints.md (Uvi): New.
22282 (Dd): Split into Ds and new Dd.
22283 * config/aarch64/aarch64.md (*movsi_aarch64):
22285 (*movdi_aarch64): Add SIMD mov case.
22287 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22289 * tree-predcom.c: (struct chain): Handle store-store chain in which
22290 stores for elimination only store loop invariant values.
22291 (execute_pred_commoning_chain): Ditto.
22292 (prepare_initializers_chain_store_elim): Ditto.
22293 (prepare_finalizers): Ditto.
22294 (is_inv_store_elimination_chain): New function.
22295 (initialize_root_vars_store_elim_1): New function.
22297 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22299 * tree-predcom.c: Revise general description of the pass.
22300 (enum chain_type): New enum type for store elimination.
22301 (struct chain): New field supporting store elimination.
22302 (struct component): Ditto.
22303 (dump_chain): Dump store-stores chain.
22304 (release_chain): Release resources.
22305 (split_data_refs_to_components): Compute and create component
22306 contains only stores for elimination.
22307 (get_chain_last_ref_at): New function.
22308 (make_invariant_chain): Initialization.
22309 (make_rooted_chain): Specify chain type in parameter and record it.
22310 (add_looparound_copies): Skip for store-stores chain.
22311 (determine_roots_comp): Compute type of chain and pass it to
22313 (initialize_root_vars_store_elim_2): New function.
22314 (finalize_eliminated_stores): New function.
22315 (remove_stmt): Handle store for elimination.
22316 (execute_pred_commoning_chain): Execute predictive commoning on
22317 store-store chains.
22318 (determine_unroll_factor): Skip unroll for store-stores chain.
22319 (prepare_initializers_chain_store_elim): New function.
22320 (prepare_initializers_chain): Hanlde store-store chain.
22321 (prepare_finalizers_chain, prepare_finalizers): New function.
22322 (tree_predictive_commoning_loop): Return integer value indicating
22323 if loop is unrolled or lcssa form is corrupted.
22324 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22326 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22328 * tree-predcom.c (initialize_root): Delete.
22329 (execute_pred_commoning_chain): Initialize root vars and replace
22330 reference of non-combined chain directly, rather than call above
22333 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22335 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22336 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22338 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22340 * tree-predcom.c (struct chain): New field init_seq.
22341 (release_chain): Release init_seq.
22342 (prepare_initializers_chain): Record intialization stmts in above
22344 (insert_init_seqs): New function.
22345 (tree_predictive_commoning_loop): Call insert_init_seqs.
22347 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22349 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22351 2017-07-28 Richard Biener <rguenther@suse.de>
22353 * match.pd: Remove superfluous :c.
22354 * genmatch.c (simplify::id): Add member.
22355 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22357 (current_id): New global.
22358 (dt_node::parent): Move from ...
22359 (dt_operand::parent): ... here. Add for_id member.
22360 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22361 (decision_tree::find_node): Relax order requirement when
22362 merging DT_TRUE nodes to ones inbetween the current simplify
22363 and the one we try to merge with. Add diagnostic whenever
22364 we need to enforce pattern order by not merging.
22365 (decision_tree::insert): Set current_id.
22366 (decision_tree::print_node): Dump parent node and for_id.
22367 (parser::last_id): Add member.
22368 (parser::push_simplify): Assign unique id.
22369 (parser::parser): Initialize last_id.
22371 2017-07-28 Martin Liska <mliska@suse.cz>
22374 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22375 gimple_build_debug_bind.
22377 2017-07-28 Richard Biener <rguenther@suse.de>
22379 PR tree-optimization/81502
22380 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22382 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22384 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22385 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22386 for BIT_FIELD_REF args.
22387 * fold-const.c (make_bit_field_ref): Likewise.
22388 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22390 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22393 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22394 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22395 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22396 Or it into SANITIZER_UNDEFINED.
22397 * ubsan.c: Include gimple-fold.h and varasm.h.
22398 (ubsan_expand_ptr_ifn): New function.
22399 (instrument_pointer_overflow): New function.
22400 (maybe_instrument_pointer_overflow): New function.
22401 (instrument_object_size): Formatting fix.
22402 (pass_ubsan::execute): Call instrument_pointer_overflow
22403 and maybe_instrument_pointer_overflow.
22404 * internal-fn.c (expand_UBSAN_PTR): New function.
22405 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22406 * sanitizer.def (__ubsan_handle_pointer_overflow,
22407 __ubsan_handle_pointer_overflow_abort): New builtins.
22408 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22409 * internal-fn.def (UBSAN_PTR): New internal function.
22410 * opts.c (sanitizer_opts): Add pointer-overflow.
22411 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22412 * fold-const.c (build_range_check): Compute pointer range check in
22413 integral type if pointer arithmetics would be needed. Formatting
22416 2017-07-28 Martin Liska <mliska@suse.cz>
22419 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22420 parameters that are of a variable-length.
22422 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22424 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22426 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22427 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22428 (CRT_CALL_STATIC_FUNCTION): Likewise.
22429 (ASM_DEFAULT_SPEC): New define.
22430 (ASM_SPEC32): Likewise.
22431 (ASM_SPEC64): Likewise.
22432 (ASM_SPEC_COMMON): Likewise.
22433 (ASM_SPEC): Likewise.
22434 (INVALID_64BIT): Likewise.
22435 (LINK_OS_DEFAULT_SPEC): Likewise.
22436 (LINK_OS_SPEC32): Likewise.
22437 (LINK_OS_SPEC64): Likewise.
22438 (POWERPC_LINUX): Likewise.
22439 (PTRDIFF_TYPE): Likewise.
22440 (RESTORE_FP_PREFIX): Likewise.
22441 (RESTORE_FP_SUFFIX): Likewise.
22442 (SAVE_FP_PREFIX): Likewise.
22443 (SAVE_FP_SUFFIX): Likewise.
22444 (SIZE_TYPE): Likewise.
22445 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22446 (TARGET_64BIT): Likewise.
22447 (TARGET_64BIT): Likewise.
22448 (TARGET_AIX): Likewise.
22449 (WCHAR_TYPE_SIZE): Likewise.
22450 (WCHAR_TYPE): Undefine.
22451 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22452 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22453 (CPP_OS_RTEMS_SPEC): Delete.
22454 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22455 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22457 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22459 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22461 PR tree-optimization/81578
22462 * tree-parloops.c (build_new_reduction): Bail out if
22463 reduction_code isn't one of the standard OpenMP reductions.
22464 Move the details printing after that decision.
22466 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22468 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22469 related to reload_in_progress.
22470 (splat_input_operand): Likewise.
22471 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22473 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22475 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22476 (TARGET_INSTANTIATE_DECLS): Likewise.
22477 (legitimate_indexed_address_p): Delete reload_in_progress code.
22478 (rs6000_debug_legitimate_address_p): Likewise.
22479 (rs6000_eliminate_indexed_memrefs): Likewise.
22480 (rs6000_emit_le_vsx_store): Likewise.
22481 (rs6000_emit_move_si_sf_subreg): Likewise.
22482 (rs6000_emit_move): Likewise.
22483 (register_to_reg_type): Likewise.
22484 (rs6000_pre_atomic_barrier): Likewise.
22485 (rs6000_machopic_legitimize_pic_address): Likewise.
22486 (rs6000_allocate_stack_temp): Likewise.
22487 (rs6000_address_for_fpconvert): Likewise.
22488 (rs6000_address_for_altivec): Likewise.
22489 (rs6000_secondary_memory_needed_rtx): Delete function.
22490 (rs6000_check_sdmode): Likewise.
22491 (rs6000_alloc_sdmode_stack_slot): Likewise.
22492 (rs6000_instantiate_decls): Likewise.
22493 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22494 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22495 Delete reload_in_progress.
22496 (*vec_reload_and_plus_<mptrsize>): Likewise.
22497 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22498 (vsx_div_v2di): Likewise.
22499 (vsx_udiv_v2di): Likewise.
22501 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22503 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22504 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22505 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22506 (rs6000_debug_reg_global): Delete print of LRA status.
22507 (rs6000_option_override_internal): Delete dead LRA related code.
22508 (rs6000_lra_p): Delete function.
22509 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22511 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22513 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22514 * config/riscv/rtems.h: New file.
22516 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22517 Sudakshina Das <sudi.das@arm.com>
22519 * config/aarch64/aarch64.md
22520 (define_split for and<mode>3nr_compare): Move
22521 non aarch64_logical_operand to a register.
22522 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22523 register immediate operand to a register.
22524 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22526 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22528 PR middle-end/81564
22529 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22531 2017-07-27 Richard Biener <rguenther@suse.de>
22533 PR tree-optimization/81573
22534 PR tree-optimization/81494
22535 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22536 multi defuse cycle case.
22538 2017-07-27 Richard Biener <rguenther@suse.de>
22540 PR tree-optimization/81571
22541 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22544 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22546 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22547 earlier and only if MASK_FPU is set. Adjust formatting.
22549 2017-07-27 Martin Liska <mliska@suse.cz>
22551 * opt-functions.awk: Add validation of value of Init.
22552 * optc-gen.awk: Pass new argument.
22554 2017-07-27 Martin Liska <mliska@suse.cz>
22556 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22557 Fix wrong condition.
22559 2017-07-27 Martin Liska <mliska@suse.cz>
22561 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22562 BBs and edges seen by autoFDO.
22564 2017-07-27 Richard Biener <rguenther@suse.de>
22566 PR tree-optimization/81502
22567 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22568 with incompatible but same sized type.
22569 (execute_update_addresses_taken): Likewise.
22571 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22573 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22574 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22576 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22579 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22580 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22581 Change s_operand to memory_operand.
22583 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22585 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22586 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22587 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22588 Emit instructions rather than returning an expression. Handle TFmode
22589 and KFmode by casting to TImode.
22590 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22591 (rs6000_emit_le_vsx_store): Likewise.
22592 * config/rs6000/vsx.md (VSX_TI): New iterator.
22593 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22594 (*vsx_le_undo_permute_<mode>): Likewise.
22595 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22596 emit the split sequence.
22597 (*vsx_le_perm_store_<mode>): Likewise.
22599 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22601 PR tree-optimization/81555
22602 PR tree-optimization/81556
22603 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22604 if true, force CHANGED for the recursive invocation.
22605 (reassociate_bb): Remember original length of ops array, pass
22606 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22608 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22609 attributes for noipa attribute. For naked attribute use
22610 lookup_attribute first before lookup_attribute_spec.
22611 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22613 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22614 for functions with noipa attribute.
22615 (cgraph_externally_visible_p): Return true for functions with noipa
22617 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22618 for functions with noipa attribute.
22619 * doc/extend.texi: Document noipa function attribute.
22620 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22621 also for functions with noipa attribute.
22622 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22624 2017-07-26 Andrew Pinski <apinski@cavium.com>
22626 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22627 vec_unalign_load_cost and vec_unalign_store_cost.
22629 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22631 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22632 -mvsx-small-integer option.
22633 (ISA_3_0_MASKS_IEEE): Likewise.
22634 (OTHER_VSX_VECTOR_MASKS): Likewise.
22635 (POWERPC_MASKS): Likewise.
22636 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22637 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22638 code, only testing for DImode being allowed in non-VSX floating
22640 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22641 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22643 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22644 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22645 TARGET_P8_VECTOR test.
22646 (rs6000_secondary_reload_simple_move): Likewise.
22647 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22648 since TARGET_P9_VECTOR was already tested.
22649 (rs6000_opt_masks): Remove -mvsx-small-integer.
22650 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22651 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22653 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22654 test for TARGET_VEXTRACTUB was used, and that uses
22656 (p9 extract splitter): Likewise.
22657 (vsx_extract_<mode>_di_p9): Likewise.
22658 (vsx_extract_<mode>_store_p9): Likewise.
22659 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22660 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22661 the elimination of TARGET_VSX_SMALL_INTEGER.
22662 (vsx_extract_<mode>_p8): Likewise.
22663 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22664 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22665 (vsx_set_<mode>_p9): Likewise.
22666 (vsx_set_v4sf_p9): Likewise.
22667 (vsx_set_v4sf_p9_zero): Likewise.
22668 (vsx_insert_extract_v4sf_p9): Likewise.
22669 (vsx_insert_extract_v4sf_p9_2): Likewise.
22670 * config/rs6000/rs6000.md (sign extend splitter): Change
22671 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22672 (floatsi<mode>2_lfiwax_mem): Likewise.
22673 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22674 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22675 since a test for TARGET_P9_VECTOR was used.
22676 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22677 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22678 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22679 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22680 TARGET_P8_VECTOR test.
22681 (fix_trunc<mode>si2_stfiwx): Likewise.
22682 (fix_trunc<mode>si2_internal): Likewise.
22683 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22684 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22686 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22687 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22688 TARGET_P8_VECTOR test.
22689 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22690 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22691 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22693 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22694 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22695 since a test for TARGET_P9_VECTOR was used.
22696 (splitter for loading small constants): Likewise.
22698 2017-07-26 Andrew Pinski <apinski@cavium.com>
22700 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22703 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22706 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22707 (fp_valid_at): Likewise.
22709 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22711 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22712 (qdf24xx_addrcost_table): Likewise.
22713 (cortexa57_tunings): Update to use generic_branch_cost.
22714 (cortexa72_tunings): Likewise.
22715 (cortexa73_tunings): Likewise.
22716 (qdf24xx_tunings): Likewise.
22718 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22720 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22721 (thunderx2t99_branch_cost): Likewise.
22722 (cortexa35_tunings): Update to use generic_branch_cost.
22723 (cortexa53_tunings): Likewise.
22724 (cortexa57_tunings): Likewise.
22725 (cortexa72_tunings): Likewise.
22726 (cortexa73_tunings): Likewise.
22727 (thunderx2t99_tunings): Likewise.
22729 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22731 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22732 (sparc_option_override): Honour MASK_FSMULD.
22733 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22734 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22735 * config/sparc/sparc.opt (mfsmuld): New option.
22736 * doc/invoke.texi (mfsmuld): Document option.
22738 2017-07-26 Marek Polacek <polacek@redhat.com>
22740 PR middle-end/70992
22741 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22743 2017-07-26 Richard Biener <rguenther@suse.de>
22745 * gimple-match-head.c (do_valueize): Return OP if valueize
22747 (get_def): New helper to get at the def stmt of a SSA name
22748 if valueize allows.
22749 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22750 do_valueize to get at the def stmt.
22751 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22753 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22755 PR middle-end/46932
22756 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22758 2017-07-26 Martin Liska <mliska@suse.cz>
22761 * function.c (expand_function_start): Make expansion of
22762 nonlocal_goto_save_area after parm_birth_insn.
22764 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22766 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22767 from all CPU target flags enable members.
22769 2017-07-26 Richard Biener <rguenther@suse.de>
22771 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22772 (decision_tree::gen): Make 'type' const.
22773 (write_predicate): Likewise.
22775 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22777 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22778 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22779 (rs6000_option_override_internal): Likewise.
22780 (rs6000_expand_vector_set): Likewise.
22781 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22782 (TARGET_UPPER_REGS_SF): Likewise.
22783 (TARGET_UPPER_REGS_DI): Likewise.
22784 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22785 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22786 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22787 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22788 (Splitters for DI constants in Altivec registers): Likewise.
22789 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22790 (vsx_set_v4sf_p9): Likewise.
22791 (vsx_set_v4sf_p9_zero): Likewise.
22792 (vsx_insert_extract_v4sf_p9): Likewise.
22793 (vsx_insert_extract_v4sf_p9_2): Likewise.
22795 2017-07-25 Carl Love <cel@us.ibm.com>
22797 * doc/extend.texi: Update the built-in documentation file for the
22798 existing built-in functions
22799 vector signed char vec_cnttz (vector signed char);
22800 vector unsigned char vec_cnttz (vector unsigned char);
22801 vector signed short vec_cnttz (vector signed short);
22802 vector unsigned short vec_cnttz (vector unsigned short);
22803 vector signed int vec_cnttz (vector signed int);
22804 vector unsigned int vec_cnttz (vector unsigned int);
22805 vector signed long long vec_cnttz (vector signed long long);
22806 vector unsigned long long vec_cnttz (vector unsigned long long);
22808 2017-07-25 Andrew Pinski <apinski@cavium.com>
22810 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22811 accesses where the use is for the first operand of a BIT_INSERT.
22813 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22816 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22817 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22819 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22821 * config/i386/gstabs.h: Delete.
22822 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22824 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22826 * config/i386/i386.c (ix86_decompose_address): Do not check for
22827 register RTX when looking at index_reg or base_reg.
22828 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22830 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22832 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22833 to update EH info here.
22835 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22837 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22839 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22841 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22843 2017-07-25 Torsten Duwe <duwe@suse.de>
22845 * common.opt: Introduce -fpatchable-function-entry
22846 command line option, and its variables function_entry_patch_area_size
22847 and function_entry_patch_area_start.
22848 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22849 including a two-value parser.
22850 * target.def (print_patchable_function_entry): New target hook.
22851 * targhooks.h (default_print_patchable_function_entry): New function.
22852 * targhooks.c (default_print_patchable_function_entry): Likewise.
22853 * toplev.c (process_options): Switch off IPA-RA if
22854 patchable function entries are being generated.
22855 * varasm.c (assemble_start_function): Look at the
22856 patchable-function-entry command line switch and current
22857 function attributes and maybe generate NOP instructions by
22858 calling the print_patchable_function_entry hook.
22859 * doc/extend.texi: Document patchable_function_entry attribute.
22860 * doc/invoke.texi: Document -fpatchable_function_entry
22861 command line option.
22862 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22864 * doc/tm.texi: Re-generate.
22866 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22869 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22870 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22872 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22874 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22876 2017-07-25 Richard Biener <rguenther@suse.de>
22878 PR tree-optimization/81455
22879 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22880 not walk in cycles when looking for guards.
22882 2017-07-25 Richard Biener <rguenther@suse.de>
22884 PR tree-optimization/81529
22885 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22886 when optimizing backedge uses.
22888 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22890 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22892 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22893 to dl_section_ref. On AIX, append an expression to subtract
22894 the size of the section length to dl_section_ref.
22896 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22898 * configure.ac: If any of the config.* scripts fail, exit 1.
22899 * configure: Regenerate.
22901 2017-07-25 Richard Biener <rguenther@suse.de>
22903 PR middle-end/81546
22904 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22905 of immediate uses, be more verbose on errors.
22907 2017-07-25 Richard Biener <rguenther@suse.de>
22909 PR tree-optimization/81510
22910 * tree-vect-loop.c (vect_is_simple_reduction): When the
22911 reduction stmt is not inside the loop bail out.
22913 2017-07-25 Richard Biener <rguenther@suse.de>
22915 PR tree-optimization/81303
22916 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22917 profitability check against LOOP_VINFO_NITERSM1.
22919 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22921 * domwalk.c (cmp_bb_postorder): Simplify.
22922 (sort_bbs_postorder): New function. Use it...
22923 (dom_walker::walk): ...here to optimize common cases.
22925 2017-07-25 Martin Liska <mliska@suse.cz>
22928 * ipa-visibility.c (function_and_variable_visibility): Make the
22929 redirection just on target that supports aliasing.
22930 Fix GNU coding style.
22932 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22935 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22937 * config/arm/rtems.h: Likewise.
22938 * config/bfin/rtems.h: Likewise.
22939 * config/i386/rtemself.h: Likewise.
22940 * config/lm32/rtems.h: Likewise.
22941 * config/m32c/rtems.h: Likewise.
22942 * config/m68k/rtemself.h: Likewise.
22943 * config/microblaze/rtems.h: Likewise.
22944 * config/mips/rtems.h: Likewise.
22945 * config/moxie/rtems.h: Likewise.
22946 * config/nios2/rtems.h: Likewise.
22947 * config/powerpcspe/rtems.h: Likewise.
22948 * config/rs6000/rtems.h: Likewise.
22949 * config/rtems.h: Likewise.
22950 * config/sh/rtems.h: Likewise.
22951 * config/sh/rtemself.h: Likewise.
22952 * config/sparc/rtemself.h: Likewise.
22954 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22957 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22958 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22959 * tree-ssa-structalias.c (alias_get_name): Same.
22961 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22964 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22965 instructions if no du chain is found.
22967 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22969 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22971 2017-07-25 Richard Biener <rguenther@suse.de>
22973 PR middle-end/81505
22974 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22977 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22979 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22980 upper-regs options.
22981 (ISA_2_7_MASKS_SERVER): Likewise.
22982 (ISA_3_0_MASKS_IEEE): Likewise.
22983 (OTHER_P8_VECTOR_MASKS): Likewise.
22984 (OTHER_VSX_VECTOR_MASKS): Likewise.
22985 (POWERPC_MASKS): Likewise.
22986 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22987 duplicate list of options.
22988 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22989 explicit -mupper-regs options.
22990 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22991 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22992 alias for -mupper-regs-df.
22993 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22994 (rs6000_init_hard_regno_mode_ok): Likewise.
22995 (rs6000_option_override_internal): Likewise.
22996 (rs6000_opt_masks): Likewise.
22997 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22998 options in terms of whether -mvsx or -mpower8-vector was used.
22999 (TARGET_UPPER_REGS_DI): Likewise.
23000 (TARGET_UPPER_REGS_SF): Likewise.
23001 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23002 -mupper-regs-* options.
23004 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23006 * passes.c (emergency_dump_function): Print some empty lines and a
23007 header before the RTL dump.
23009 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23011 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23013 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23016 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23017 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23019 2017-07-24 Carl Love <cel@us.ibm.com>
23021 * config/rs6000/rs6000-c.c: Add support for built-in functions
23022 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23023 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23024 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23025 vec_extract_fp_from_shortl): Add defines for the two builtins.
23026 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23027 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23029 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23030 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23031 * doc/extend.texi: Update the built-in documentation file for the
23032 new built-in function.
23034 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23037 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23039 * doc/generic.texi: Likewise.
23040 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23041 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23043 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23045 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23046 (aarch64_mls_elt_merge<mode>): Likewise.
23048 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23050 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23051 having __cxa_atexit.
23053 2017-07-23 Michael Collison <michael.collison@arm.com>
23055 * config/arm/arm.c (arm_option_override): Deprecate
23056 use of -mstructure-size-boundary.
23057 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23058 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23060 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23063 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23064 Reduce cost estimate for direct moves.
23066 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23069 * config/i386/i386.c (ix86_option_override_internal): Disable
23070 BMI, BMI2 and TBM instructions for -m16.
23072 2017-07-21 Carl Love <cel@us.ibm.com>
23074 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23075 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23076 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23077 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23078 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23079 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23080 VMULOSW): New enum "unspec" values.
23081 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23082 altivec_vmulosw): New patterns.
23083 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23084 VMULOSW): Add definitions.
23086 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23088 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23089 (qdf24xx): Likewise.
23090 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23091 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23092 (AARCH64_FL_V8_1): Renumber.
23093 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23094 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23095 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23096 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23097 rdma to feature modifiers list.
23099 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23101 PR middle-end/56727
23102 * ipa-visibility (function_and_variable_visibility): Convert
23103 recursive PLT call to direct call if appropriate.
23105 2017-07-21 Andrew Pinski <apinski@cavium.com>
23107 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23108 operand 1 to see if the types precision matches.
23109 * fold-const.c (operand_equal_p): Likewise.
23111 2017-07-21 Richard Biener <rguenther@suse.de>
23113 PR tree-optimization/81303
23114 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23115 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23116 (vect_peeling_hash_get_lowest_cost): Adjust.
23117 (vect_enhance_data_refs_alignment): Likewise. Use
23118 vect_get_peeling_costs_all_drs to compute the penalty for no
23119 peeling to match up costs.
23121 2017-07-21 Richard Biener <rguenther@suse.de>
23123 PR tree-optimization/81500
23124 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23125 we didn't identify a reduction path.
23127 2017-07-21 Tom de Vries <tom@codesourcery.com>
23128 Cesar Philippidis <cesar@codesourcery.com>
23130 PR gcov-profile/81442
23131 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23134 2017-07-21 Tom de Vries <tom@codesourcery.com>
23137 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23139 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23140 nvptx_override_options_after_change.
23142 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23144 * dwarf2out.c (output_file_names): Avoid double testing for
23145 dwarf_version >= 5.
23147 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23149 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23151 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23153 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23155 (try_crossjump_to_edge): Do not punt on partitioned functions.
23157 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23159 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23160 Put all BBs reachable only via paths crossing cold region to cold
23162 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23164 2016-07-21 Richard Biener <rguenther@suse.de>
23166 PR tree-optimization/81303
23167 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23168 into account prologue and epilogue iterations when raising
23169 min_profitable_iters to sth at least covering one vector iteration.
23171 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23173 * config/arm/arm.c (arm_test_cpu_arch_dat):
23176 2017-07-20 Nathan Sidwell <nathan@acm.org>
23178 Remove TYPE_METHODS.
23179 * tree.h (TYPE_METHODS): Delete.
23180 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23181 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23182 (dbxout_type_methods): Scan TYPE_FIELDS.
23183 (dbxout_type): Don't check TYPE_METHODS here.
23184 * function.c (use_register_for_decl): Always ignore register for
23185 class types when not optimizing.
23186 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23187 * tree.c (free_lang_data_in_type): Stitch out member functions and
23188 templates from TYPE_FIELDS.
23189 (build_distinct_type_copy, verify_type_variant,
23190 verify_type): Member fns are on TYPE_FIELDS.
23191 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23192 * tree-pretty-print.c (dump_generic_node): Likewise.
23194 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23197 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23198 V2TImode and V4TImode.
23199 (ix86_expand_vector_extract): Likewise.
23200 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23201 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23202 (ssescalarmode): Handle V4TImode and V2TImode.
23203 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23204 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23205 (VEXTRACTI128_MODE): New mode iterator.
23206 (splitter for *vec_extractv?ti first element): New.
23207 (VEC_INIT_MODE): New mode iterator.
23208 (vec_init<mode>): Consolidate 3 expanders into one using
23209 VEC_INIT_MODE mode iterator.
23211 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23213 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23214 non_spilled_static_chain_regno_p.
23216 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23218 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23220 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23222 * bb-reorder.c (connect_traces): Allow copying of blocks within
23225 2017-07-20 Richard Biener <rguenther@suse.de>
23227 * gimple.h (gimple_phi_result): Add gphi * overload.
23228 (gimple_phi_result_ptr): Likewise.
23229 (gimple_phi_arg): Likewise. Adjust index assert to only
23230 allow actual argument accesses rather than all slots available
23232 (gimple_phi_arg_def): Add gphi * overload.
23233 * tree-phinodes.c (make_phi_node): Initialize only actual
23235 (resize_phi_node): Clear memory not covered by old node,
23236 do not initialize excess argument slots.
23237 (reserve_phi_args_for_new_edge): Initialize new argument slot
23240 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23242 PR tree-optimization/81388
23244 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23246 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23247 by removing computation of may_be_zero.
23249 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23250 Tom de Vries <tom@codesourcery.com>
23252 PR middle-end/81030
23253 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23254 when gimple level profile disagrees with what RTL expander did.
23256 2017-07-20 Richard Biener <rguenther@suse.de>
23258 PR tree-optimization/61171
23259 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23260 (vect_analyze_stmt): Add slp instance parameter.
23261 (vectorizable_reduction): Likewise.
23262 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23263 (vect_is_simple_reduction): Deal with chains not detected
23264 as SLP reduction chain, specifically not properly associated
23265 chains containing a mix of plus/minus.
23266 (get_reduction_op): Remove.
23267 (get_initial_defs_for_reduction): Simplify, pass in whether
23268 this is a reduction chain, pass in the SLP node for the PHIs.
23269 (vect_create_epilog_for_reduction): Get the SLP instance as
23271 (vectorizable_reduction): Get the SLP instance as arg.
23272 During analysis remember the SLP node with the PHIs in the
23273 instance. Simplify getting at the vectorized reduction PHIs.
23274 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23275 through SLP instance.
23276 (vect_slp_analyze_operations): Likewise.
23277 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23278 (vect_transform_stmt): Likewise.
23280 2017-07-20 Tom de Vries <tom@codesourcery.com>
23282 PR tree-optimization/81489
23283 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23284 read of phi arg location to before loop that modifies phi.
23286 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23288 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23291 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23293 PR middle-end/81331
23294 * except.c (execute): Fix ordering issue.
23296 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23298 PR rtl-optimization/81423
23299 * combine.c (make_compound_operation_int): Don't try to optimize
23300 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23302 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23304 PR rtl-optimization/81423
23305 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23306 with a constant that is -1 in the truncated to mode.
23308 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23310 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23311 (determine_unlikely_bbs): ... here.
23312 * predict.h (propagate_unlikely_bbs_forward): Declare.
23313 * cfgexpand.c (pass_expand::execute): Use it.
23314 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23316 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23317 propagate_unlikely_bbs_forward.
23319 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23321 PR middle-end/81331
23322 * except.c (maybe_add_nop_after_section_switch): New function.
23325 2017-07-19 Tom de Vries <tom@codesourcery.com>
23327 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23329 2017-07-19 Tom de Vries <tom@codesourcery.com>
23331 * gimple.h (gimple_phi_arg): Make assert more strict.
23333 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23335 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23336 * config/rs6000/mmintrin.h: New file.
23337 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23339 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23341 PR tree-optimization/81346
23342 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23344 2017-07-19 Tom de Vries <tom@codesourcery.com>
23346 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23348 2017-07-19 Tom de Vries <tom@codesourcery.com>
23350 * config/nvptx/nvptx-modes.def: Add V2DImode.
23351 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23352 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23353 (nvptx_output_mov_insn): Handle lack of mov.b128.
23354 (nvptx_print_operand): Handle 'H' and 'L' codes.
23355 (nvptx_vector_mode_supported): Allow V2DImode.
23356 (nvptx_preferred_simd_mode): New function.
23357 (nvptx_data_alignment): New function.
23358 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23359 nvptx_preferred_simd_mode.
23360 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23362 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23364 2017-07-19 Tom de Vries <tom@codesourcery.com>
23366 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23367 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23368 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23369 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23370 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23371 (mov<VECIM>_insn): New define_insn.
23372 (define_expand "mov<VECIM>): New define_expand.
23374 2017-07-19 Tom de Vries <tom@codesourcery.com>
23376 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23378 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23380 PR tree-optimization/81346
23381 * fold-const.h (fold_div_compare, range_check_type): Declare.
23382 * fold-const.c (range_check_type): New function.
23383 (build_range_check): Use range_check_type.
23384 (fold_div_compare): No longer static, rewritten into
23385 a match.pd helper function.
23386 (fold_comparison): Don't call fold_div_compare here.
23387 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23388 as helper function.
23390 2017-07-19 Nathan Sidwell <nathan@acm.org>
23392 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23393 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23394 * tree.c (find_decls_types_r, verify_type): Use
23395 TYPE_{MIN,MAX}_VALUE_RAW.
23396 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23397 (hash_tree): Likewise.
23398 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23400 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23403 2017-07-18 Tom de Vries <tom@codesourcery.com>
23405 PR middle-end/81464
23406 * omp-expand.c (expand_omp_for_static_chunk): Handle
23407 equal-argument loop exit phi.
23409 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23412 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23413 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23414 operand 2 predicate.
23415 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23416 operand 2 predicate.
23417 (ror,rol -> rorx splitters): Use const_int_operand as
23418 operand 2 predicate.
23420 2017-06-18 Richard Biener <rguenther@suse.de>
23422 PR tree-optimization/81410
23423 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23424 the gap in the ! slp_perm SLP case after each group.
23426 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23428 PR middle-end/81463
23429 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23432 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23434 PR middle-end/81462
23435 * predict.c (set_even_probabilities): Cleanup; do not affect
23436 probabilities that are already known.
23437 (combine_predictions_for_bb): Call even when count is set.
23439 2017-07-18 Nathan Sidwell <nathan@acm.org>
23441 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23444 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23447 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23448 optimization for loop niter analysis.
23450 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23453 * config/avr/avr.c (avr_optimize_casesi): Don't use
23454 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23456 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23458 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23459 body_cost_vec from _vect_peel_extended_info.
23460 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23461 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23464 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23466 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23468 2017-07-18 Richard Biener <rguenther@suse.de>
23470 PR tree-optimization/80620
23471 PR tree-optimization/81403
23472 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23473 info when re-using a VN table entry.
23475 2017-07-18 Richard Biener <rguenther@suse.de>
23477 PR tree-optimization/81418
23478 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23479 vectype_in. Verify that with lane-reducing reduction operations
23480 we have a single def-use cycle.
23482 2017-07-17 Carl Love <cel@us.ibm.com>
23484 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23486 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23487 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23488 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23489 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23490 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23491 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23492 VMULOSW): New enum "unspec" values.
23493 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23494 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23495 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23496 altivec_vmulosw): New patterns.
23497 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23498 VMULOSW): Add definitions.
23500 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23502 * config/alpha/alpha.c: Include predict.h.
23504 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23506 * tree-vrp.c (compare_assert_loc): Fix comparison function
23507 to return predictable results.
23509 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23511 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23513 (subdi3): Likewise.
23514 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23515 * doc/invoke.texi (mexpand-adddi): Update text.
23517 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23519 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23520 that also clobbers the CC register. The old expand code is moved
23522 (*arc_clzsi2): ... here.
23523 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23524 the CC register. The old expand code is moved to ...
23525 (arc_ctzsi2): ... here.
23527 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23529 * config/arc/arc.opt (mindexed-loads): Use initial value
23530 TARGET_INDEXED_LOADS_DEFAULT.
23531 (mauto-modify-reg): Use initial value
23532 TARGET_AUTO_MODIFY_REG_DEFAULT.
23533 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23534 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23535 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23536 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23538 2017-07-17 Martin Liska <mliska@suse.cz>
23541 * opts.c (finish_options): Do not allow -fgnu-tm
23542 w/ -fsanitize={kernel-,}address. Say sorry.
23544 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23547 * tree-loop-distribution.c (classify_partition): Only assert on
23548 numer of iterations.
23549 (merge_dep_scc_partitions): Delete prameter. Update function call.
23550 (distribute_loop): Remove code handling loop with unknown niters.
23551 (pass_loop_distribution::execute): Skip loop with unknown niters.
23553 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23556 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23557 function sort_partitions_by_post_order.
23559 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23561 PR tree-optimization/81374
23562 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23563 the max index of basic blocks, rather than number of basic blocks.
23565 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23567 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23569 (arc_legitimate_pic_operand_p): Likewise.
23570 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23572 (arc_needs_pcl_p): Likewise.
23573 (arc_legitimate_pc_offset_p): Likewise.
23574 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23575 function is also used in constrains.md.
23576 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23577 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23578 PLUS. Only return true/false in known cases, otherwise assert.
23579 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23580 is already called in arc_legitimate_constant_p.
23581 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23583 (LEGITIMATE_PIC_OPERAND_P): Use
23584 arc_raw_symbolic_reference_mentioned_p function.
23585 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23590 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23591 Andrew Burgess <andrew.burgess@embecosm.com>
23593 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23594 (arc_return_address_register): New function.
23595 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23596 (arc_handle_fndecl_attribute): Add naked attribute.
23597 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23598 (TARGET_WARN_FUNC_RETURN): Likewise.
23599 (arc_allocate_stack_slots_for_args): New function.
23600 (arc_warn_func_return): Likewise.
23601 (machine_function): Change type fn_type.
23602 (arc_compute_function_type): Consider new naked function type,
23603 change function return type.
23604 (arc_must_save_register): Adapt to handle new
23605 arc_compute_function_type's return type.
23606 (arc_expand_prologue): Likewise.
23607 (arc_expand_epilogue): Likewise.
23608 (arc_return_address_regs): Delete.
23609 (arc_return_address_register): New function.
23610 (arc_epilogue_uses): Use above function.
23611 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23612 (arc_function_type): Change encoding, add naked type.
23613 (ARC_INTERRUPT_P): Change to handle the new encoding.
23614 (ARC_FAST_INTERRUPT_P): Likewise.
23615 (ARC_NORMAL_P): Define.
23616 (ARC_NAKED_P): Likewise.
23617 (arc_compute_function_type): Delete prototype.
23618 * config/arc/arc.md (in_ret_delay_slot): Use
23619 arc_return_address_register function.
23620 (simple_return): Likewise.
23621 (p_return_i): Likewise.
23623 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23625 PR tree-optimization/81428
23626 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23627 can't be built for those types.
23629 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23631 Remove stuff dead since r239246.
23633 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23634 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23635 (avr_inform_devices): Remove dead stuff.
23637 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23639 * config/arm/arm_neon.h: Fix softp typo.
23641 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23643 PR tree-optimization/81365
23644 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23645 aggregate moves onto bb predecessor edges, make sure there are no
23646 loads that could alias the lhs in between the start of bb and the
23649 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23652 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23653 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23654 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23656 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23658 PR tree-optimization/81396
23659 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23660 (init_symbolic_number): Initialize it to 1.
23661 (perform_symbolic_merge): Add n_ops from both operands into the new
23663 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23664 without base_addr as useless if they need more than one operation.
23665 (bswap_replace): Handle !bswap case for NULL base_addr.
23667 2017-07-17 Tom de Vries <tom@codesourcery.com>
23670 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23673 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23675 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23676 conditional builtin define __FIX_LEON3FT_B2BST.
23678 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23680 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23681 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23684 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23686 PR rtl-optimization/81424
23687 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23688 to remove potential trapping from operands if -fnon-call-exceptions.
23690 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23692 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23693 profile_proability for scalling.
23694 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23696 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23698 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23700 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23702 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23703 fixpoint arithmetics.
23705 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23707 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23708 fixpoint arithmetics.
23710 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23712 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23713 fixpoint arithmetics.
23715 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23717 * profile-count.h (profile_probability::from_reg_br_prob_note,
23718 profile_probability::to_reg_br_prob_note): New functions.
23719 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23720 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23721 * predict.c (probability_reliable_p): Update.
23722 (edge_probability_reliable_p): Update.
23723 (br_prob_note_reliable_p): Update.
23724 (invert_br_probabilities): Update.
23725 (add_reg_br_prob_note): New function.
23726 (combine_predictions_for_insn): Update.
23727 * asan.c (asan_clear_shadow): Update.
23728 * cfgbuild.c (compute_outgoing_frequencies): Update.
23729 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23730 (update_br_prob_note): Update.
23731 (rtl_verify_edges): Update.
23732 (purge_dead_edges): Update.
23733 (fixup_reorder_chain): Update.
23734 * emit-rtl.c (try_split): Update.
23735 * ifcvt.c (cond_exec_process_insns): Update.
23736 (cond_exec_process_if_block): Update.
23737 (dead_or_predicable): Update.
23738 * internal-fn.c (expand_addsub_overflow): Update.
23739 (expand_neg_overflow): Update.
23740 (expand_mul_overflow): Update.
23741 * loop-doloop.c (doloop_modify): Update.
23742 * loop-unroll.c (compare_and_jump_seq): Update.
23743 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23744 * predict.h: Update.
23745 * reorg.c (mostly_true_jump): Update.
23747 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23748 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23749 * config/arc/arc.c: (emit_unlikely_jump): Update.
23750 * config/arm/arm.c: (emit_unlikely_jump): Update.
23751 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23752 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23753 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23754 (ix86_print_operand): Update.
23755 (ix86_split_fp_branch): Update.
23756 (predict_jump): Update.
23757 * config/ia64/ia64.c (ia64_print_operand): Update.
23758 * config/mmix/mmix.c (mmix_print_operand): Update.
23759 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23760 (rs6000_expand_split_stack_prologue): Update.
23761 * config/rs6000/rs6000.c: Update.
23762 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23763 (s390_expand_vec_movstr): Update.
23764 (s390_expand_cs_tdsi): Update.
23765 (s390_expand_split_stack_prologue): Update.
23766 * config/sh/sh.c (sh_print_operand): Update.
23767 (expand_cbranchsi4): Update.
23768 (expand_cbranchdi4): Update.
23769 * config/sparc/sparc.c (output_v9branch): Update.
23770 * config/spu/spu.c (get_branch_target): Update.
23771 (ea_load_store_inline): Update.
23772 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23773 * config/tilepro/tilepro.c: Update.
23775 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23777 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23778 (gimplify_save_expr): Likewise.
23780 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23782 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23784 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23786 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23788 * timevar.def (TV_IPA_FNSUMMARY): Define.
23790 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23792 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23793 to back store errata sensitive sequence from being generated.
23794 (sqrtdf2_fix): Likewise.
23796 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23798 * tree-ssa-threadupdate.c (compute_path_counts,
23799 update_joiner_offpath_counts): Use profile_probability.
23801 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23804 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23806 * config/arm/arm-c.c (arm_cpu_builtins): Define
23807 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23809 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23811 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23812 array entries to represent __ieee128 versions of the
23813 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23814 scalar_extract_sig, and scalar_insert_exp built-in functions.
23815 (altivec_resolve_overloaded_builtin): Add special case handling
23816 for the __builtin_scalar_insert_exp function, as represented by
23817 the P9V_BUILTIN_VEC_VSIEDP constant.
23818 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23819 exponent support for __ieee128 argument.
23820 (VSESQP): Add scalar extract signature support for __ieee128
23822 (VSTDCNQP): Add scalar test negative support for __ieee128
23824 (VSIEQP): Add scalar insert exponent support for __int128 argument
23825 with __ieee128 result.
23826 (VSIEQPF): Add scalar insert exponent support for __ieee128
23827 argument with __ieee128 result.
23828 (VSTDCQP): Add scalar test data class support for __ieee128
23830 (VSTDCNQP): Add overload support for scalar test negative with
23831 __ieee128 argument.
23832 (VSTDCQP): Add overload support for scalar test data class
23833 __ieee128 argument.
23834 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23835 UNSPEC_VSX_SXSIGDP.
23836 (UNSPEC_VSX_SIEXPQP): New constant.
23837 (xsxexpqp): New insn for VSX scalar extract exponent quad
23839 (xsxsigqp): New insn for VSX scalar extract significand quad
23841 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23842 precision with floating point argument.
23843 (xststdcqp): New expand for VSX scalar test data class quad
23845 (xststdcnegqp): New expand for VSX scalar test negative quad
23847 (xststdcqp): New insn to match expansions for VSX scalar test data
23848 class quad precision and VSX scalar test negative quad precision.
23849 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23850 special case operand checking to enforce that second operand of
23851 VSX scalar test data class with quad precision argument is a 7-bit
23853 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23854 prototypes and descriptions of __ieee128 versions of
23855 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23856 scalar_test_data_class, and scalar_test_neg built-in functions.
23858 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23860 PR tree-optimization/81162
23861 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23862 replace a negate with an add.
23864 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23866 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23868 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23870 * config/arm/arm-c.c (arm_cpu_builtins): Define
23871 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23873 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23875 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23876 (armv8-r): Set ARM Cortex-R52 as default CPU.
23877 * config/arm/arm-tables.opt: Regenerate.
23878 * config/arm/arm-tune.md: Regenerate.
23879 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23881 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23882 extension for -mcpu=cortex-r52.
23884 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23886 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23887 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23888 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23889 (fp-armv8): Define it as FP_ARMv8 only.
23890 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23891 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23893 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23894 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23895 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23896 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23897 than TARGET_FPU_ARMV8.
23898 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23899 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23900 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23902 * config/arm/neon.md (neon_vrint): Likewise.
23903 (neon_vcvt): Likewise.
23904 (neon_<fmaxmin_op><mode>): Likewise.
23905 (<fmaxmin><mode>3): Likewise.
23906 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23907 * config/arm/predicates.md (arm_cond_move_operator): Check against
23908 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23910 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23912 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23913 to top of function.
23915 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23917 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23918 loop in comment with memset.
23920 2017-07-14 Martin Liska <mliska@suse.cz>
23922 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23923 * dwarf2out.c (is_java): Remove the function.
23924 (output_pubname): Remove usage of the function.
23925 (lower_bound_default): Remove usage of DW_LANG_Java.
23926 (gen_compile_unit_die): Likewise.
23927 * gcc.c: Remove compiler defaults for .java and .zip files.
23928 * gimple-expr.c (remove_suffix): Change as there's no longer
23929 extension than 4-letter one.
23930 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23931 (gimplify_save_expr): Likewise.
23932 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23933 as it's possible even for other languages than Java.
23934 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23935 * lto-opts.c (lto_write_options): Remove reference to Java.
23936 * opts.c (strip_off_ending): Update file extension handling.
23937 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23938 * tree-eh.c (lower_resx): Likewise.
23939 * tree.c (free_lang_data_in_type): Remove dead code.
23940 (find_decls_types_r): Likewise.
23941 (build_common_builtin_nodes): Remove Java from a comment.
23942 (verify_type): Remove dead code.
23943 * varasm.c (assemble_external): Remove Java from a comment.
23945 2017-07-14 Martin Liska <mliska@suse.cz>
23947 * opts.c (finish_options): Add quotes.
23948 (common_handle_option): Likewise.
23950 2017-07-14 Martin Liska <mliska@suse.cz>
23952 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23953 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23954 Remove N_SO_PASCAL.
23955 * dwarf2out.c (lower_bound_default): Do not handle
23957 (gen_compile_unit_die): Likewise.
23958 * gcc.c: Remove default extension binding for GNU Pascal.
23959 * stmt.c: Remove Pascal language from a comment.
23960 * xcoffout.c: Likewise.
23962 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23965 * diagnostic-show-locus.c (fixit_cmp): New function.
23966 (layout::layout): Sort m_fixit_hints.
23967 (column_range::column_range): Assert that the values are valid.
23968 (struct char_span): New struct.
23969 (correction::overwrite): New method.
23970 (struct source_line): New struct.
23971 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23972 calls in terms of classes source_line and char_span, and
23973 correction::overwrite.
23974 (selftest::test_overlapped_fixit_printing_2): New function.
23975 (selftest::diagnostic_show_locus_c_tests): Call it.
23977 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23979 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23980 early if there is no lhs.
23982 2017-07-13 Martin Liska <mliska@suse.cz>
23984 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23985 (gen_reference_type_die): Likewise.
23986 * stor-layout.c: Remove Pascal-related comment.
23988 2017-07-13 Martin Liska <mliska@suse.cz>
23990 * opts.c (finish_options): Add quotes to error messages.
23991 (parse_sanitizer_options): Likewise.
23993 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23995 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23997 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23999 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24001 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24003 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24005 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24007 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24010 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24011 provides the hardware capability bits, define the macro
24012 __BUILTIN_CPU_SUPPORTS__.
24013 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24014 if GLIBC does not provide the hardware capability bits. Add a
24015 gcc_unreachable call if the built-in cpu function is neither
24016 __builtin_cpu_is nor __builtin_cpu_supports.
24017 (rs6000_get_function_versions_dispatcher): Change the warning
24018 that an old GLIBC is used which does not export the capability
24019 bits to be an error.
24020 * doc/extend.texi (target_clones attribute): Document the
24021 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24022 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24023 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24024 the macros defined by GCC if the newer GLIBC is available.
24026 2017-07-12 Jeff Law <law@redhat.com>
24028 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24029 remaining includes slightly.
24030 * config/riscv/riscv-builtins.c: Include profile-count.h.
24032 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24035 * config/avr/avr.c (avr_set_current_function): In diagnostic
24036 messages: Quote keywords and (parts of) identifiers.
24037 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24040 2017-07-12 Carl Love <cel@us.ibm.com>
24042 * config/rs6000/rs6000-c.c: Add support for built-in functions
24043 vector bool char vec_revb (vector bool char);
24044 vector bool short vec_revb (vector short char);
24045 vector bool int vec_revb (vector bool int);
24046 vector bool long long vec_revb (vector bool long long);
24047 * doc/extend.texi: Update the built-in documentation file for the
24048 new built-in functions.
24050 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24052 * config/s390/s390.md: Remove movcc splitter.
24054 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24056 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24057 load/store on condition.
24059 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24062 * config/avr/avr.c (avr_encode_section_info)
24063 [progmem && !TREE_READONLY]: Error if progmem object needs
24066 2017-07-11 Michael Collison <michael.collison@arm.com>
24068 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24071 2017-07-11 Carl Love <cel@us.ibm.com>
24073 * config/rs6000/rs6000-c.c: Add support for builtins
24074 vector unsigned int vec_parity_lsbb (vector signed int);
24075 vector unsigned int vec_parity_lsbb (vector unsigned int);
24076 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24077 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24078 vector unsigned long long vec_parity_lsbb (vector signed long long);
24079 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24080 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24081 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24082 * doc/extend.texi: Update the built-in documentation file for the
24083 new built-in functions.
24085 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24087 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24088 (layout::m_primary_loc): New field.
24089 (layout::layout): Initialize new field. Move location filtering
24090 logic from here to...
24091 (layout::maybe_add_location_range): ...this new method. Add
24092 support for filtering to just the lines already specified by other
24094 (layout::will_show_line_p): New method.
24095 (gcc_rich_location::add_location_if_nearby): New method.
24096 (selftest::test_add_location_if_nearby): New test function.
24097 (selftest::diagnostic_show_locus_c_tests): Call it.
24098 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24101 2017-07-11 Tom de Vries <tom@codesourcery.com>
24103 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24104 (bb_first_real_insn): New function.
24105 (nvptx_single): Add extra initialization of broadcasted condition
24108 2017-07-11 Nathan Sidwell <nathan@acm.org>
24110 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24112 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24114 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24115 Binutils doc as TEXI will mess them up.
24116 * doc/invoke.texi (AVR Options): Same here.
24118 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24120 * config/sparc/sparc.opt (mfix-ut700): New option.
24121 (mfix-gr712rc): Likewise.
24122 (sparc_fix_b2bst): New variable.
24123 * doc/invoke.texi (SPARC options): Document them.
24124 (ARM options): Fix warnings.
24125 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24126 instructions to prevent sequences that can trigger the store-store
24127 errata for certain LEON3FT processors.
24128 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24129 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24130 * config/sparc/sparc.md (fix_b2bst): New attribute.
24131 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24133 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24136 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24138 (*rsqrtsf2_sse): Ditto.
24140 (div<mode>3): Macroize insn from divdf3 and divsf3
24141 using MODEF mode iterator.
24143 2017-07-10 Martin Sebor <msebor@redhat.com>
24145 PR tree-optimization/80397
24146 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24147 instead of testing for equality to INTEGER_TYPE.
24149 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24151 * config.gcc: Remove uclibc from arc target spec.
24153 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24155 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24157 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24160 * lto-wrapper.c (remove_option): New function.
24161 (merge_and_complain): Merge PIC/PIE options more realistically.
24163 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24165 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24169 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24170 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24171 * config.in: Regenerate.
24172 * configure: Regenerate.
24173 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24174 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24175 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24176 (TARGET_GASISR_PROLOGUES): ...target mask.
24177 * common/config/avr/avr-common.c
24178 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24179 Set -mgas-isr-prologues.
24180 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24181 INSERT_PASS_BEFORE for it.
24182 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24183 * config/avr/avr.c (avr_option_override)
24184 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24185 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24186 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24187 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24188 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24190 (make_avr_pass_pre_proep): New function.
24191 (emit_push_sfr) <treg>: Add argument to function and use it
24192 instead of TMP_REG.
24193 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24194 and set machine->gasisr.yes.
24195 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24196 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24197 __gcc_isr.n_pushed to .L__stack_usage.
24198 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24199 (avr_asm_final_postscan_insn): ...this new static function.
24200 * config/avr/avr.h (machine_function)
24201 <is_no_gccisr, use_L__stack_usage>: New fields.
24202 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24203 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24204 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24205 (gasisr, *gasisr): New expander and insn.
24206 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24207 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24208 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24210 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24212 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24215 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24217 Move jump-tables out of .text again.
24220 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24221 (ASM_OUTPUT_ADDR_VEC): New function.
24222 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24223 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24224 INSN_ADDRESSes as asm comment.
24225 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24226 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24227 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24228 * config/avr/avr.md (*tablejump): Adjust comment.
24229 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24230 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24232 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24233 (avr_output_addr_vec): New proto.
24234 (avr_log_t) <insn_addresses>: New field.
24236 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24239 * config/i386/i386.c (ix86_function_arg_advance): Set
24240 outgoing_args_on_stack to true if there are outgoing arguments
24242 (ix86_function_arg): Likewise.
24243 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24244 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24245 * config/i386/i386.h (machine_function): Add
24246 outgoing_args_on_stack.
24248 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24250 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24251 supporting pthreds.
24252 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24254 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24256 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24257 (REAL_H): Remove $(MACHMODE_H).
24258 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24260 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24261 $(MACHMODE_H) and double-int.h.
24262 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24264 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24267 2017-07-07 Andrew Pinski <apinski@cavium.com>
24269 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24270 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24272 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24274 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24275 Add warning if GCC was not configured to link against a GLIBC that
24276 exports the hardware capability bits.
24277 (make_resolver_func): Make resolver function private and not a
24278 COMDAT function. Create the name with clone_function_name instead
24279 of make_unique_name.
24282 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24283 correct operand in doing the split.
24285 2017-07-07 Carl Love <cel@us.ibm.com>
24287 * config/rs6000/rs6000-c: Add support for built-in function
24288 vector unsigned short vec_pack_to_short_fp32 (vector float,
24290 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24291 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24292 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24293 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24294 (convert_4f32_8i16): Add define_expand.
24295 * doc/extend.texi: Update the built-in documentation file for the
24296 new built-in function.
24298 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24300 * config/sparc/m8.md: New file.
24301 * config/sparc/sparc.md: Include m8.md.
24303 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24305 * config/sparc/sparc.opt: New option -mvis4b.
24306 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24307 (sparc_option_override): Handle VIS4B.
24308 (enum sparc_builtins): Define
24309 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24310 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24311 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24312 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24313 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24314 (check_constant_argument): New function.
24315 (sparc_vis_init_builtins): Define builtins
24316 __builtin_vis_dictunpack{8,16,32},
24317 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24318 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24319 __builtin_vis_fpcmpde{8,16,32}shl and
24320 __builtin_vis_fpcmpur{8,16,32}shl.
24321 (sparc_expand_builtin): Check that the constant operands to
24322 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24323 constant and in range.
24324 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24326 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24327 (SPARC_IMM5_P): Likewise.
24328 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24329 (enabled): Handle vis4b.
24330 (UNSPEC_DICTUNPACK): New unspec.
24331 (UNSPEC_FPCMPSHL): Likewise.
24332 (UNSPEC_FPUCMPSHL): Likewise.
24333 (UNSPEC_FPCMPDESHL): Likewise.
24334 (UNSPEC_FPCMPURSHL): Likewise.
24335 (cpu_feature): New CPU feature `vis4b'.
24336 (dictunpack{8,16,32}): New insns.
24337 (FPCSMODE): New mode iterator.
24338 (fpcscond): New code iterator.
24339 (fpcsucond): Likewise.
24340 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24341 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24342 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24343 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24344 * config/sparc/constraints.md: Define constraints `q' for unsigned
24345 2-bit integer constants and `t' for unsigned 5-bit integer
24347 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24349 (imm5_operand_dictunpack16): Likewise.
24350 (imm5_operand_dictunpack32): Likewise.
24351 (imm2_operand): Likewise.
24352 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24353 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24354 ditunpack* and fpcmp*shl builtins.
24356 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24358 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24359 * config.in: Add HAVE_AS_SPARC6 define.
24360 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24362 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24364 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24365 (CPP_CPU_SPEC): Handle m8.
24366 (ASM_CPU_SPEC): Likewise.
24367 * config/sparc/sparc-opts.h (enum processor_type): Add
24369 * config/sparc/sparc.c (m8_costs): New struct.
24370 (sparc_option_override): Handle TARGET_CPU_m8.
24371 (sparc32_initialize_trampoline): Likewise.
24372 (sparc64_initialize_trampoline): Likewise.
24373 (sparc_issue_rate): Likewise.
24374 (sparc_register_move_cost): Likewise.
24375 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24376 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24377 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24378 (CPP_CPU_SPEC): Handle M8.
24379 (ASM_CPU_SPEC): Likewise.
24380 (AS_M8_FLAG): Define.
24381 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24382 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24383 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24385 * configure: Regenerate.
24386 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24389 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24391 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24393 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24394 ("*movdi_insn_sp32"): Do not set v3pipe.
24395 ("*movsi_insn"): Likewise.
24396 ("*movdi_insn_sp64"): Likewise.
24397 ("*movsf_insn"): Likewise.
24398 ("*movdf_insn_sp32"): Likewise.
24399 ("*movdf_insn_sp64"): Likewise.
24400 ("*zero_extendsidi2_insn_sp64"): Likewise.
24401 ("*sign_extendsidi2_insn"): Likewise.
24402 ("*mov<VM32:mode>_insn"): Likewise.
24403 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24404 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24405 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24406 ("<vlop:code><VL:mode>3"): Likewise.
24407 ("*not_<vlop:code><VL:mode>3"): Likewise.
24408 ("*nand<VL:mode>_vis"): Likewise.
24409 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24410 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24411 ("one_cmpl<VL:mode>2"): Likewise.
24412 ("faligndata<VM64:mode>_vis"): Likewise.
24413 ("alignaddrsi_vis"): Likewise.
24414 ("alignaddrdi_vis"): Likweise.
24415 ("alignaddrlsi_vis"): Likewise.
24416 ("alignaddrldi_vis"): Likewise.
24417 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24418 ("bmaskdi_vis"): Likewise.
24419 ("bmasksi_vis"): Likewise.
24420 ("bshuffle<VM64:mode>_vis"): Likewise.
24421 ("cmask8<P:mode>_vis"): Likewise.
24422 ("cmask16<P:mode>_vis"): Likewise.
24423 ("cmask32<P:mode>_vis"): Likewise.
24424 ("pdistn<P:mode>_vis"): Likewise.
24425 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24427 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24429 * config/sparc/sparc.md ("subtype"): New insn attribute.
24430 ("*wrgsr_sp64"): Set insn subtype.
24431 ("*rdgsr_sp64"): Likewise.
24432 ("alignaddrsi_vis"): Likewise.
24433 ("alignaddrdi_vis"): Likewise.
24434 ("alignaddrlsi_vis"): Likewise.
24435 ("alignaddrldi_vis"): Likewise.
24436 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24437 ("fexpand_vis"): Likewise.
24438 ("fpmerge_vis"): Likewise.
24439 ("faligndata<VM64:mode>_vis"): Likewise.
24440 ("bshuffle<VM64:mode>_vis"): Likewise.
24441 ("cmask8<P:mode>_vis"): Likewise.
24442 ("cmask16<P:mode>_vis"): Likewise.
24443 ("cmask32<P:mode>_vis"): Likewise.
24444 ("fchksm16_vis"): Likewise.
24445 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24446 ("fmean16_vis"): Likewise.
24447 ("fp<plusminus_insn>64_vis"): Likewise.
24448 ("<plusminus_insn>v8qi3"): Likewise.
24449 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24450 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24451 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24452 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24453 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24454 ("*movqi_insn"): Likewise.
24455 ("*movhi_insn"): Likewise.
24456 ("*movsi_insn"): Likewise.
24457 ("movsi_pic_gotdata_op"): Likewise.
24458 ("*movdi_insn_sp32"): Likewise.
24459 ("*movdi_insn_sp64"): Likewise.
24460 ("movdi_pic_gotdata_op"): Likewise.
24461 ("*movsf_insn"): Likewise.
24462 ("*movdf_insn_sp32"): Likewise.
24463 ("*movdf_insn_sp64"): Likewise.
24464 ("*zero_extendhisi2_insn"): Likewise.
24465 ("*zero_extendqihi2_insn"): Likewise.
24466 ("*zero_extendqisi2_insn"): Likewise.
24467 ("*zero_extendqidi2_insn"): Likewise.
24468 ("*zero_extendhidi2_insn"): Likewise.
24469 ("*zero_extendsidi2_insn_sp64"): Likewise.
24470 ("ldfsr"): Likewise.
24471 ("prefetch_64"): Likewise.
24472 ("prefetch_32"): Likewise.
24473 ("tie_ld32"): Likewise.
24474 ("tie_ld64"): Likewise.
24475 ("*tldo_ldub_sp32"): Likewise.
24476 ("*tldo_ldub1_sp32"): Likewise.
24477 ("*tldo_ldub2_sp32"): Likewise.
24478 ("*tldo_ldub_sp64"): Likewise.
24479 ("*tldo_ldub1_sp64"): Likewise.
24480 ("*tldo_ldub2_sp64"): Likewise.
24481 ("*tldo_ldub3_sp64"): Likewise.
24482 ("*tldo_lduh_sp32"): Likewise.
24483 ("*tldo_lduh1_sp32"): Likewise.
24484 ("*tldo_lduh_sp64"): Likewise.
24485 ("*tldo_lduh1_sp64"): Likewise.
24486 ("*tldo_lduh2_sp64"): Likewise.
24487 ("*tldo_lduw_sp32"): Likewise.
24488 ("*tldo_lduw_sp64"): Likewise.
24489 ("*tldo_lduw1_sp64"): Likewise.
24490 ("*tldo_ldx_sp64"): Likewise.
24491 ("*mov<VM32:mode>_insn"): Likewise.
24492 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24493 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24495 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24497 * config/sparc/sparc.md ("type"): New insn type viscmp.
24498 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24500 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24501 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24502 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24503 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24505 ("n7_vis_logical_11cycle"): Likewise.
24506 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24507 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24508 * config/sparc/niagara.md ("niag_vis"): Likewise.
24509 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24510 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24512 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24514 * config/sparc/sparc.md: New instruction type `bmask'.
24515 (bmaskdi_vis): Use the `bmask' type.
24516 (bmasksi_vis): Likewise.
24517 * config/sparc/ultra3.md (us3_array): Likewise.
24518 * config/sparc/niagara7.md (n7_array): Likewise.
24519 * config/sparc/niagara4.md (n4_array): Likewise.
24520 * config/sparc/niagara2.md (niag2_vis): Likewise.
24521 (niag3_vis): Likewise.
24522 * config/sparc/niagara.md (niag_vis): Likewise.
24524 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24526 * ipa-comdats.c: Remove optimize check from gate.
24527 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24528 for functions not optimized.
24529 (ipa_fn_summary_read): Skip optimize check.
24530 (ipa_fn_summary_write): Likewise.
24531 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24533 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24535 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24536 (check_callers): Give up on caller which is not optimized.
24537 (inline_small_functions): Likewise.
24538 (ipa_inline): Do not give up when not optimizing.
24539 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24540 away unoptimizes cdtors.
24541 (whole_program_function_and_variable_visibility): Do
24542 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24543 * ipa.c (process_references): Do not check optimize.
24544 (symbol_table::remove_unreachable_nodes): Update optimize check.
24545 (set_writeonly_bit): Update optimize check.
24546 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24547 (pass_ipa_single_use::gate): Remove.
24549 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24551 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24552 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24553 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24554 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24555 permute_load, permute_store, adjust_extract, adjust_splat,
24556 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24557 replace_swap_with_copy, dump_swap_insn_table,
24558 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24559 recombine_lvx_pattern, recombine_stvx_pattern,
24560 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24561 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24562 to file rs6000-p8swap.c.
24563 * config/rs6000/rs6000-p8swap.c: New file.
24564 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24565 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24566 and rs6000*-*-* targets.
24568 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24570 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24572 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24574 * lto-wrapper.c (merge_and_complain): Do not merge
24575 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24576 fsigned_zeros, ftrapping_math, fwrapv.
24577 (append_compiler_options): Do not track these options.
24578 (append_linker_options): Likewie
24580 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24582 * cgraphunit.c (cgraph_node::finalize_function): When
24583 !flag_toplevel_reorde set no_reorder flag.
24584 (varpool_node::finalize_decl): Likewise.
24585 (symbol_table::compile): Drop no toplevel reorder path.
24587 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24589 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24590 edges; zero probability is not better than uninitialized.
24592 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24594 * asan.h (asan_sanitize_allocas_p): Declare.
24595 * asan.c (asan_sanitize_allocas_p): New function.
24596 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24597 (handle_builtin_alloca): Likewise.
24598 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24599 if !asan_sanitize_allocas_p.
24600 * params.def (asan-instrument-allocas): Add new option.
24601 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24602 * opts.c (common_handle_option): Disable allocas sanitization for
24605 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24607 * asan.c: Include gimple-fold.h.
24608 (get_last_alloca_addr): New function.
24609 (handle_builtin_stackrestore): Likewise.
24610 (handle_builtin_alloca): Likewise.
24611 (asan_emit_allocas_unpoison): Likewise.
24612 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24613 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24614 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24615 (instrument_builtin_call): Pass gimple iterator to
24616 get_mem_refs_of_builtin_call.
24617 (last_alloca_addr): New global.
24618 * asan.h (asan_emit_allocas_unpoison): Declare.
24619 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24620 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24621 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24622 if function calls alloca.
24623 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24624 * gimple-fold.h (replace_call_with_value): Declare.
24625 * internal-fn.c: Include asan.h.
24626 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24627 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24629 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24631 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24632 (C_SELFTEST_FLAGS): New.
24633 (CPP_SELFTEST_FLAGS): New.
24634 (SELFTEST_DEPS): New, from deps of s-selftest.
24635 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24636 (CPP_SELFTEST_DEPS): New.
24637 (selftest): Add dependency on s-selftest-c++.
24638 (s-selftest): Rename to...
24639 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24640 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24641 than SELFTEST_FLAGS.
24642 (selftest-gdb): Rename to...
24643 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24645 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24646 (selftest-valgrind): Rename to...
24647 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24649 (selftest-valgrind): Reintroduce as an alias for
24650 selftest-c-valgrind.
24651 (s-selftest-c++): New.
24652 (selftest-c++-gdb): New.
24653 (selftest-c++-valgrind): New.
24655 2017-07-06 Olivier Hainque <hainque@adacore.com>
24657 * gcc.c (process_command): When deciding if undefined variables
24658 should be ignored when processing specs, accept "gcc -v" as well.
24660 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24662 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24663 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24665 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24667 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24668 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24669 * config/arm/arm-tables.opt: Regenerate.
24670 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24672 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24674 2017-07-06 Carl Love <cel@us.ibm.com>
24676 * ChangeLog: Clean up from mid air collision
24678 2017-07-06 Carl Love <cel@us.ibm.com>
24680 * config/rs6000/rs6000-c.c: Add support for built-in functions
24681 vector signed int vec_subc (vector signed int, vector signed int);
24682 vector signed __int128 vec_subc (vector signed __int128,
24683 vector signed __int128);
24684 vector unsigned __int128 vec_subc (vector unsigned __int128,
24685 vector unsigned __int128);
24686 vector signed int vec_sube (vector signed int, vector signed int,
24687 vector signed int);
24688 vector unsigned int vec_sube (vector unsigned int,
24689 vector unsigned int,
24690 vector unsigned int);
24691 vector signed __int128 vec_sube (vector signed __int128,
24692 vector signed __int128,
24693 vector signed__int128);
24694 vector unsigned __int128 vec_sube (vector unsigned __int128,
24695 vector unsigned __int128,
24696 vector unsigned __int128);
24697 vector signed int vec_subec (vector signed int, vector signed int,
24698 vector signed int);
24699 vector unsigned int vec_subec (vector unsigned int,
24700 vector unsigned int,
24701 vector unsigned int);
24702 vector signed __int128 vec_subec (vector signed __int128,
24703 vector signed __int128,
24704 vector signed__int128);
24705 vector unsigned __int128 vec_subec (vector unsigned __int128,
24706 vector unsigned __int128,
24707 vector unsigned __int128);
24708 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24709 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24710 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24711 BU_ALTIVEC_OVERLOAD_X definitions.
24712 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24713 * doc/extend.texi: Update the built-in documentation file for the new
24714 built-in functions.
24716 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24719 * diagnostic-show-locus.c (layout::layout): Use start and finish
24720 spelling location for the start and finish of each range.
24721 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24722 Add unused aspect param.
24723 * input.c (expand_location_1): Add "aspect" param, and use it
24724 to access the correct part of the location.
24725 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24727 (expand_location_to_spelling_point): Likewise.
24728 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24729 param, and pass it to expand_location_1.
24731 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24733 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24734 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24735 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24736 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24737 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24738 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24739 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24740 _mm_maskz_getmant_ss): New intrinsics.
24741 (__builtin_ia32_getexpss128_mask): Changed to ...
24742 __builtin_ia32_getexpss128_round ... this.
24743 (__builtin_ia32_getexpsd128_mask): Changed to ...
24744 __builtin_ia32_getexpsd128_round ... this.
24745 * config/i386/i386-builtin-types.def
24746 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24747 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24748 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24749 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24750 __builtin_ia32_getmantss_mask_round): New builtins.
24751 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24752 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24753 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24754 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24755 * config/i386/sse.md
24756 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24757 avx512f_sgetexp<mode><mask_scalar_name>
24758 <round_saeonly_scalar_name> ... this.
24759 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24760 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24761 vgetexp<ssescalarmodesuffix>
24762 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24763 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24764 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24765 avx512f_vgetmant<mode><mask_scalar_name>
24766 <round_saeonly_scalar_name> ... this.
24767 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24768 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24769 vgetmant<ssescalarmodesuffix>
24770 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24771 %0<mask_scalar_operand4>, %1, %2
24772 <round_saeonly_scalar_mask_op4>, %3} ... this.
24773 * config/i386/subst.md (mask_scalar_operand4,
24774 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24775 round_saeonly_scalar_nimm_predicate): New subst attributes.
24777 2017-07-06 Julia Koval <julia.koval@intel.com>
24779 * config/i386/i386.c (ix86_erase_embedded_rounding):
24780 Remove code for old rounding pattern.
24782 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24784 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24786 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24788 * doc/sourcebuild.texi (Test Directives, Variants of
24789 dg-require-support): Add documentation for dg-require-stack-check.
24791 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24793 * config/i386/subst.md (mask_scalar, round_scalar,
24794 round_saeonly_scalar): New meta-templates.
24795 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24796 round_scalar_mask_operand3, round_scalar_mask_op3,
24797 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24798 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24799 round_saeonly_scalar_constraint,
24800 round_saeonly_scalar_prefix): New subst attribute.
24801 * config/i386/sse.md
24802 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24803 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24804 <round_scalar_name> ... this.
24805 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24806 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24807 <round_scalar_name> ... this.
24808 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24809 <sse>_vm<code><mode>3<mask_scalar_name>
24810 <round_saeonly_scalar_name> ... this.
24811 (v<plusminus_mnemonic><ssescalarmodesuffix>
24812 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24813 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24814 v<plusminus_mnemonic><ssescalarmodesuffix>
24815 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24816 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24817 (v<multdiv_mnemonic><ssescalarmodesuffix>
24818 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24819 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24820 v<multdiv_mnemonic><ssescalarmodesuffix>
24821 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24822 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24823 (v<maxmin_float><ssescalarmodesuffix>
24824 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24825 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24826 v<maxmin_float><ssescalarmodesuffix>
24827 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24828 %0<mask_scalar_operand3>, %1, %<iptr>2
24829 <round_saeonly_scalar_mask_op3>} ... this.
24831 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24833 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24834 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24836 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24837 Alan Hayward <alan.hayward@arm.com>
24838 David Sherwood <david.sherwood@arm.com>
24840 * combine.c (simplify_if_then_else): Remove "enum" before
24842 * compare-elim.c (can_eliminate_compare): Likewise.
24843 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24845 (aarch64_lookup_simd_builtin_type): Likewise.
24846 (aarch64_simd_builtin_type): Likewise.
24847 (aarch64_init_simd_builtin_types): Likewise.
24848 (aarch64_simd_expand_args): Likewise.
24849 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24851 (aarch64_reverse_mask): Likewise.
24852 (aarch64_simd_emit_reg_reg_move): Likewise.
24853 (aarch64_gen_adjusted_ldpstp): Likewise.
24854 (aarch64_ccmp_mode_to_code): Likewise.
24855 (aarch64_operands_ok_for_ldpstp): Likewise.
24856 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24857 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24859 (aarch64_min_divisions_for_recip_mul): Likewise.
24860 (aarch64_reassociation_width): Likewise.
24861 (aarch64_get_condition_code_1): Likewise.
24862 (aarch64_simd_emit_reg_reg_move): Likewise.
24863 (aarch64_simd_attr_length_rglist): Likewise.
24864 (aarch64_reverse_mask): Likewise.
24865 (aarch64_operands_ok_for_ldpstp): Likewise.
24866 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24867 (aarch64_gen_adjusted_ldpstp): Likewise.
24868 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24870 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24871 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24872 (arm_lookup_simd_builtin_type): Likewise.
24873 (arm_simd_builtin_type): Likewise.
24874 (arm_init_simd_builtin_types): Likewise.
24875 (arm_expand_builtin_args): Likewise.
24876 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24877 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24878 (ft32_setup_incoming_varargs): Likewise.
24879 (ft32_function_arg): Likewise.
24880 (ft32_function_arg_advance): Likewise.
24881 (ft32_pass_by_reference): Likewise.
24882 (ft32_arg_partial_bytes): Likewise.
24883 (ft32_valid_pointer_mode): Likewise.
24884 (ft32_addr_space_pointer_mode): Likewise.
24885 (ft32_addr_space_legitimate_address_p): Likewise.
24886 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24888 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24889 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24890 (iamcu_alignment): Likewise.
24891 (canonicalize_vector_int_perm): Likewise.
24892 (ix86_noce_conversion_profitable_p): Likewise.
24893 (ix86_mpx_bound_mode): Likewise.
24894 (ix86_operands_ok_for_move_multiple): Likewise.
24895 * config/microblaze/microblaze-protos.h
24896 (microblaze_expand_conditional_branch_reg): Likewise.
24897 * config/microblaze/microblaze.c
24898 (microblaze_expand_conditional_branch_reg): Likewise.
24899 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24901 (rs6000_reassociation_width): Likewise.
24902 (rs6000_invalid_binary_op): Likewise.
24903 (fusion_p9_p): Likewise.
24904 (emit_fusion_p9_load): Likewise.
24905 (emit_fusion_p9_store): Likewise.
24906 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24908 (riscv_hard_regno_mode_ok_p): Likewise.
24909 (riscv_address_insns): Likewise.
24910 (riscv_split_symbol): Likewise.
24911 (riscv_legitimize_move): Likewise.
24912 (riscv_function_value): Likewise.
24913 (riscv_hard_regno_nregs): Likewise.
24914 (riscv_expand_builtin): Likewise.
24915 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24916 (riscv_build_integer): Likewise.
24917 (riscv_split_integer): Likewise.
24918 (riscv_legitimate_constant_p): Likewise.
24919 (riscv_cannot_force_const_mem): Likewise.
24920 (riscv_regno_mode_ok_for_base_p): Likewise.
24921 (riscv_valid_base_register_p): Likewise.
24922 (riscv_valid_offset_p): Likewise.
24923 (riscv_valid_lo_sum_p): Likewise.
24924 (riscv_classify_address): Likewise.
24925 (riscv_legitimate_address_p): Likewise.
24926 (riscv_address_insns): Likewise.
24927 (riscv_load_store_insns): Likewise.
24928 (riscv_force_binary): Likewise.
24929 (riscv_split_symbol): Likewise.
24930 (riscv_force_address): Likewise.
24931 (riscv_legitimize_address): Likewise.
24932 (riscv_move_integer): Likewise.
24933 (riscv_legitimize_const_move): Likewise.
24934 (riscv_legitimize_move): Likewise.
24935 (riscv_address_cost): Likewise.
24936 (riscv_subword): Likewise.
24937 (riscv_output_move): Likewise.
24938 (riscv_canonicalize_int_order_test): Likewise.
24939 (riscv_emit_int_order_test): Likewise.
24940 (riscv_function_arg_boundary): Likewise.
24941 (riscv_pass_mode_in_fpr_p): Likewise.
24942 (riscv_pass_fpr_single): Likewise.
24943 (riscv_pass_fpr_pair): Likewise.
24944 (riscv_get_arg_info): Likewise.
24945 (riscv_function_arg): Likewise.
24946 (riscv_function_arg_advance): Likewise.
24947 (riscv_arg_partial_bytes): Likewise.
24948 (riscv_function_value): Likewise.
24949 (riscv_pass_by_reference): Likewise.
24950 (riscv_setup_incoming_varargs): Likewise.
24951 (riscv_print_operand): Likewise.
24952 (riscv_elf_select_rtx_section): Likewise.
24953 (riscv_save_restore_reg): Likewise.
24954 (riscv_for_each_saved_reg): Likewise.
24955 (riscv_register_move_cost): Likewise.
24956 (riscv_hard_regno_mode_ok_p): Likewise.
24957 (riscv_hard_regno_nregs): Likewise.
24958 (riscv_class_max_nregs): Likewise.
24959 (riscv_memory_move_cost): Likewise.
24960 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24961 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24962 (rl78_addr_space_address_mode): Likewise.
24963 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24965 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24966 (rs6000_reassociation_width): Likewise.
24967 (rs6000_invalid_binary_op): Likewise.
24968 (fusion_p9_p): Likewise.
24969 (emit_fusion_p9_load): Likewise.
24970 (emit_fusion_p9_store): Likewise.
24971 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24972 (ok_for_simple_move_operands): Likewise.
24973 (ok_for_simple_move_strict_operands): Likewise.
24974 (ok_for_simple_arith_logic_operands): Likewise.
24975 (visium_legitimize_reload_address): Likewise.
24976 (visium_select_cc_mode): Likewise.
24977 (output_cbranch): Likewise.
24978 (visium_split_double_move): Likewise.
24979 (visium_expand_copysign): Likewise.
24980 (visium_expand_int_cstore): Likewise.
24981 (visium_expand_fp_cstore): Likewise.
24982 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24983 (visium_function_arg): Likewise.
24984 (visium_function_arg_advance): Likewise.
24985 (visium_libcall_value): Likewise.
24986 (visium_setup_incoming_varargs): Likewise.
24987 (visium_legitimate_constant_p): Likewise.
24988 (visium_legitimate_address_p): Likewise.
24989 (visium_legitimize_address): Likewise.
24990 (visium_secondary_reload): Likewise.
24991 (visium_register_move_cost): Likewise.
24992 (visium_memory_move_cost): Likewise.
24993 (prepare_move_operands): Likewise.
24994 (ok_for_simple_move_operands): Likewise.
24995 (ok_for_simple_move_strict_operands): Likewise.
24996 (ok_for_simple_arith_logic_operands): Likewise.
24997 (visium_function_value_1): Likewise.
24998 (rtx_ok_for_offset_p): Likewise.
24999 (visium_legitimize_reload_address): Likewise.
25000 (visium_split_double_move): Likewise.
25001 (visium_expand_copysign): Likewise.
25002 (visium_expand_int_cstore): Likewise.
25003 (visium_expand_fp_cstore): Likewise.
25004 (visium_split_cstore): Likewise.
25005 (visium_select_cc_mode): Likewise.
25006 (visium_split_cbranch): Likewise.
25007 (output_cbranch): Likewise.
25008 (visium_print_operand_address): Likewise.
25009 * expmed.c (flip_storage_order): Likewise.
25010 * expmed.h (emit_cstore): Likewise.
25011 (flip_storage_order): Likewise.
25012 * genrecog.c (validate_pattern): Likewise.
25013 * hsa-gen.c (gen_hsa_addr): Likewise.
25014 * internal-fn.c (expand_arith_overflow): Likewise.
25015 * ira-color.c (allocno_copy_cost_saving): Likewise.
25016 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25017 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25018 (process_invariant_for_inheritance): Likewise.
25019 * lra-eliminations.c (move_plus_up): Likewise.
25020 * omp-low.c (lower_oacc_reductions): Likewise.
25021 * simplify-rtx.c (simplify_subreg): Likewise.
25022 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25023 (TARGET_CHKP_BOUND_MODE): Likewise..
25024 * targhooks.c (default_chkp_bound_mode): Likewise.
25025 (default_setup_incoming_vararg_bounds): Likewise.
25026 * targhooks.h (default_chkp_bound_mode): Likewise.
25027 (default_setup_incoming_vararg_bounds): Likewise.
25028 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25029 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25030 (have_whole_vector_shift): Likewise.
25031 * tree-vect-stmts.c (vectorizable_load): Likewise.
25032 * doc/tm.texi: Regenerate.
25034 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25036 Graceful degrade if Binutils PR21472 is not available.
25039 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25040 .rodata in flash test fails.
25041 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25042 * confgure: Regenerate.
25043 * config.in: Regenerate.
25044 * config/avr/avr.c (avr_asm_named_section)
25045 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25046 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25047 (avr_asm_init_sections): Same.
25049 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25051 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25052 (fma<VH:mode>4_intrinsic): Likewise.
25053 (*fmsub<VCVTF:mode>4): Likewise.
25054 (*fmsub<VH:mode>4_intrinsic): Likewise.
25056 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25059 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25060 Don't depend on "optimize > 0".
25061 (out_movhi_r_mr, out_movqi_mr_r): Same.
25062 (out_movhi_mr_r, out_movqi_r_mr): Same.
25063 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25064 io_address_operand on "optimize > 0".
25066 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25068 * tree-loop-distribution.c: Add general explanantion on the pass.
25069 (generate_loops_for_partition): Mark distributed loop.
25070 (pg_add_dependence_edges): New parameter. Handle alias data
25071 dependence specially and record it in the parameter if asked.
25072 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25073 (init_partition_graph_vertices, add_partition_graph_edge): New.
25074 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25075 (free_partition_graph_vdata, build_partition_graph): New.
25076 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25077 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25078 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25079 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25080 (version_for_distribution_p, finalize_partitions): New.
25081 (distribute_loop): Handle alias data dependence specially. Factor
25082 out loop fusion code as functions and call these functions.
25084 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25086 * tree-loop-distribution.c (classify_partition): New parameter and
25087 better handle reduction statement.
25088 (rdg_build_partitions): Revise comment.
25089 (distribute_loop): Compute statements in all partitions and pass it
25090 to classify_partition.
25092 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25094 * tree-loop-distribution.c (enum partition_type): New.
25095 (struct partition): New field type.
25096 (partition_merge_into): Add parameter. Update partition type.
25097 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25098 (build_rdg_partition_for_vertex): Compute partition type.
25099 (rdg_build_partitions): Dump partition type.
25100 (distribute_loop): Update calls to partition_merge_into.
25102 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25104 * tree-loop-distribution.c (struct ddr_hasher): New.
25105 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25107 (classify_partition): Call get_data_dependence.
25108 (pg_add_dependence_edges): Ditto.
25109 (distribute_loop): Release data dependence hash table.
25111 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25113 * tree-loop-distribution.c (ref_base_address): Delete.
25114 (similar_memory_accesses): Rename ...
25115 (share_memory_accesses): ... to this. Check if partitions access
25116 the same memory reference.
25117 (distribute_loop): Call share_memory_accesses.
25119 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25121 * tree-loop-distribution.c (struct partition): New field recording
25122 its data reference.
25123 (partition_alloc, partition_free): Init and release data refs.
25124 (partition_merge_into): Merge data refs.
25125 (build_rdg_partition_for_vertex): Collect data refs for partition.
25126 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25128 (distribute_loop): Remve data refs from vertice data of partition
25131 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25133 * tree-loop-distribution.c (params.h): Include header file.
25134 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25135 (datarefs_vec): New global var.
25136 (create_rdg_vertices): Use datarefs_vec directly.
25137 (free_rdg): Don't free data references.
25138 (build_rdg): Update use. Don't free data references.
25139 (distribute_loop): Compute global variable for data references.
25140 Bail out if there are too many data references.
25142 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25144 * tree-loop-distribution.c (loop_nest): New global var.
25145 (build_rdg): Use loop directly, rather than loop nest.
25146 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25148 (distribute_loop): Compute global variable loop nest. Update use.
25150 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25152 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25153 (partition_merge_into): New parameter. Dump reason for fusion.
25154 (distribute_loop): Update use of partition_merge_into.
25156 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25158 * tree-loop-distribution.c (bb_top_order_index): New.
25159 (bb_top_order_index_size, bb_top_order_cmp): New.
25160 (stmts_from_loop): Use topological order.
25161 (pass_loop_distribution::execute): Compute and release topological
25162 order for basic blocks.
25164 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25166 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25169 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25171 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25172 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25173 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25174 * internal-fn.def (LOOP_DIST_ALIAS): New.
25175 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25176 (fold_loop_internal_call): ... this.
25177 (vect_loop_dist_alias_call): New function.
25178 (set_uid_loop_bbs): Call fold_loop_internal_call.
25179 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25182 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25185 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25186 Require dead FLAGS_REG at the beginning of a peephole.
25188 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25191 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25192 arguments in the call to __builtin_ia32_sbb_u32.
25193 (_subborrow_u64): Swap _X and _Y arguments in the call to
25194 __builtin_ia32_sbb_u64.
25196 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25199 * tree-vrp.c (compare_assert_loc): Turn into a function template
25200 with stable template parameter. Only test if a->e is NULL,
25201 !a->e == !b->e has been verified already. Use e == NULL or
25202 e != NULL instead of e or ! e tests. If stable is true, don't use
25203 iterative_hash_expr, on the other side allow a or b or both NULL
25204 and sort the NULLs last.
25205 (process_assert_insertions): Sort using compare_assert_loc<false>
25206 instead of compare_assert_loc, later sort using
25207 compare_assert_loc<true> before calling process_assert_insertions_for
25208 in a loop. Use break instead of continue once seen NULL pointer.
25210 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25212 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25213 Cortex-R7 and Cortex-R8 processors.
25215 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25217 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25218 uninitialized while src is not.
25220 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25222 * common/config/arm/arm-common.c: Adjust include path for
25224 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25225 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25226 (arm-cpu-data.h): Likewise.
25227 (arm-cpu-cdata.h): Likewise.
25228 * config/arm/arm-cpu.h: Delete.
25229 * config/arm/arm-cpu-cdata.h: Delete.
25230 * config/arm/arm-cpu-data.h: Delete.
25232 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25234 * config/arm/arm-cpus.in (cortex-a55): New.
25235 (cortex-a75): Likewise.
25236 (cortex-a75.cortex-a55): Likewise.
25237 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25239 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25240 * config/arm/arm-cpu-cdata.h: Regenerate.
25241 * config/arm/arm-cpu-data.h: Regenerate.
25242 * config/arm/arm-cpu.h: Regenerate.
25243 * config/arm/arm-tables.opt: Regenerate.
25244 * config/arm/arm-tune.md: Regenerate.
25246 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25248 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25250 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25252 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25255 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25257 PR tree-optimization/81292
25258 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25259 full_string_p, also call adjust_related_strinfos if the adjustment
25260 is simple, otherwise invalidate related strinfos.
25262 2017-07-04 Martin Liska <mliska@suse.cz>
25265 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25266 newly created variable as DECL_IGNORED_P.
25268 2017-07-04 Martin Liska <mliska@suse.cz>
25271 * ipa-inline.c (inline_small_functions):
25272 Use xstrdup_for_dump.
25274 2017-07-04 Tom de Vries <tom@codesourcery.com>
25276 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25278 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25281 * config/darwin.c (darwin_function_switched_text_sections):
25284 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25286 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25288 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25290 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25292 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25294 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25295 min_profitable_iters, and th as inclusive lower bounds.
25296 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25297 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25298 for min_profitable_iters and min_profitable_estimate.
25299 (vect_transform_loop): Treat th as an inclusive lower bound.
25300 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25302 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25305 * config/darwin.c (darwin_function_switched_text_sections):
25306 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25307 in two pieces, and suppress the use of buf.
25309 2017-07-03 Nathan Sidwell <nathan@acm.org>
25311 * hash-table.h (hash_table_mod1): Fix indentation.
25313 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25315 PR middle-end/81290
25316 * predict.c (force_edge_cold): Be more careful about propagation
25318 * profile-count.h (profile_probability::guessed,
25319 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25321 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25323 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25325 * doc/invoke.texi (rcpc architecture extension): Document it.
25327 2017-07-03 Richard Biener <rguenther@suse.de>
25329 PR tree-optimization/60510
25330 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25331 the scalar reduction PHI and use it.
25332 (vectorizable_reduction): Properly guard the single_defuse_cycle
25333 path for non-SLP reduction chains where we cannot use it.
25334 Rework reduc_def/index and vector type deduction. Rework
25335 vector operand gathering during reduction op code-gen.
25336 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25337 chains dissolve the chain and leave it to non-SLP reduction
25340 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25342 * tree-data-ref.h (dr_alignment): Declare.
25343 * tree-data-ref.c (dr_alignment): New function.
25344 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25345 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25347 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25349 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25351 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25352 and base_misalignment fields.
25353 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25354 * tree-data-ref.c: Include builtins.h.
25355 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25356 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25357 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25358 * tree-vect-data-refs.c: Include tree-cfg.h.
25359 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25360 fields instead of calculating an alignment here.
25361 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25362 innermost_loop_behavior fields.
25364 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25366 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25368 (DR_STEP_ALIGNMENT): New macro.
25369 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25370 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25371 (create_data_ref): Print it.
25372 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25373 to tell whether the step preserves vector (mis)alignment.
25374 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25375 Move the check for an integer step and generalise to all INTEGER_CST.
25376 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25377 Print the outer step alignment.
25379 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25381 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25382 with offset_alignment.
25383 (DR_ALIGNED_TO): Delete.
25384 (DR_OFFSET_ALIGNMENT): New macro.
25385 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25386 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25387 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25388 (create_data_ref): Likewise.
25389 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25390 (vect_analyze_data_refs): Likewise.
25391 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25392 creating dummy innermost behavior.
25394 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25396 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25397 with a "innermost_loop_behavior *" and refeence tree.
25398 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25399 (create_data_ref): Update call accordingly.
25400 * tree-predcom.c (find_looparound_phi): Likewise.
25402 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25404 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25405 fields with dr_wrt_vec_loop.
25406 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25407 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25408 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25409 (vect_dr_behavior): New function.
25410 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25411 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25412 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25413 track whether the step preserves the misalignment.
25414 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25415 Use vect_dr_behavior.
25416 (vect_setup_realignment): Update call accordingly.
25417 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25418 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25419 call to vect_create_addr_base_for_vector_ref.
25420 (vect_create_cond_for_align_checks): Likewise.
25421 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25422 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25423 (vect_recog_mask_conversion_pattern): Likewise.
25424 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25425 (new_stmt_vec_info): Remove redundant zeroing.
25427 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25429 * common/config/arm/arm-common.c (arm_be8_option): New function.
25430 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25431 (ISA_ARMv6): Add isa_bit_be8.
25432 * config/arm/arm.h (arm_be8_option): Add prototype.
25433 (BE8_SPEC_FUNCTION): New define.
25434 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25435 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25436 (mlittle-endian): Similarly.
25437 (mbe8, mbe32): New options.
25438 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25439 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25441 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25443 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25445 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25447 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25448 (cleanup_tree_cfg_bb): Use it.
25449 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25451 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25453 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25456 * loop-doloop.c (add_test): Update profile.
25458 2017-07-03 Martin Liska <mliska@suse.cz>
25461 * sanopt.c (rewrite_usage_of_param): New function.
25462 (sanitize_rewrite_addressable_params): Likewise.
25463 (pass_sanopt::execute): Call rewrite_usage_of_param.
25465 2017-07-03 Richard Biener <rguenther@suse.de>
25467 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25468 back to using VIEW_CONVERT_EXPR.
25470 2017-07-03 Martin Liska <mliska@suse.cz>
25473 * doc/extend.texi: Document when a resolver function is
25474 generated for target_clones.
25476 2017-07-03 Martin Liska <mliska@suse.cz>
25478 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25479 and shadow memory of auto variables which are subject of
25480 use-after-scope sanitization.
25481 (asan_expand_mark_ifn): Add do set only when is_poison.
25483 2016-07-03 Richard Biener <rguenther@suse.de>
25485 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25487 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25488 (vectorizable_reduction): Perform reduction PHI creation when
25489 visiting a reduction PHI and adjust and simplify code generation
25490 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25491 (vect_transform_loop): Visit reduction PHIs.
25492 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25493 defs into the SLP tree.
25494 (vect_build_slp_tree): Reduction defs terminate the recursion.
25495 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25497 (vect_get_vec_defs_for_stmt_copy): Export.
25498 (vect_get_vec_defs): Likewise.
25499 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25501 (vect_get_vec_defs_for_stmt_copy): Declare.
25502 (vect_get_vec_defs): Likewise.
25504 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25506 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25507 parameter with a "loop" parameter and use it instead of the
25508 loop containing DR_STMT. Don't check simple_iv when doing
25509 BB analysis. Describe the two analysis modes in the comment.
25511 2017-07-03 Tom de Vries <tom@codesourcery.com>
25513 PR tree-optimization/69468
25514 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25515 (find_same_succ_bb): Handle ignore_edge_flags.
25517 2017-07-03 Tom de Vries <tom@codesourcery.com>
25519 PR tree-optimization/81192
25520 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25522 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25524 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25526 2017-07-03 Tom de Vries <tom@codesourcery.com>
25528 PR tree-optimization/81192
25529 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25530 BB_SAME_SUCC (bb) == NULL.
25532 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25534 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25537 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25539 * dumpfile.c: Include profile-count.h
25540 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25542 (insert_cond_bb): Update profile.
25543 * tree-cfg.h (insert_cond_bb): Update prototype.
25544 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25545 * tree-dump.c: Do not include tree-cfg.
25547 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25549 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25551 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25553 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25556 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25558 * tree-complex.c (expand_complex_div_wide): update profile.
25560 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25561 Alan Hayward <alan.hayward@arm.com>
25562 David Sherwood <david.sherwood@arm.com>
25564 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25565 (CORETYPES_H): New define.
25566 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25567 (insn-modes-inline.h, s-modes-inline-h): New rules.
25568 (generated_files): Add insn-modes-inline.h.
25569 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25570 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25571 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25572 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25573 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25574 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25575 (build/gencodes.o, build/genconditions.o): Likewise.
25576 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25577 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25578 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25579 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25580 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25581 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25582 * coretypes.h: Include everything up to real.h for generators.
25583 Include insn-modes.h first. Include wide-int-print.h after
25584 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25585 * machmode.h: Don't include insn-modes.h here.
25586 * function-tests.c: Remove includes of signop.h, machmode.h,
25587 double-int.h and wide-int.h.
25589 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25591 * optc-save-gen.awk: Likewise.
25592 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25593 * godump.c: Remove include of wide-int-print.h.
25594 * pretty-print.h: Likewise.
25595 * wide-int-print.cc: Likewise.
25596 * wide-int.cc: Likewise.
25597 * hash-map-tests.c: Remove include of signop.h.
25598 * hash-set-tests.c: Likewise.
25599 * rtl-tests.c: Likewise.
25600 * mkconfig.sh: Remove include of machmode.h.
25601 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25603 (emit_insn_modes_inline_h): ...this new function. Emit the code
25604 into an insn-modes-inline.h header file, adding appropriate
25605 include guards and end comments.
25606 (emit_insn_modes_c_header): Remove include of machmode.h.
25607 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25609 (main): Handle -i flag and call emit_insn_modes_inline_h when
25612 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25614 * tree-ssa-strlen.c (strinfo): Rename the length field to
25615 nonzero_chars. Add a full_string_p field.
25616 (compare_nonzero_chars, zero_length_string_p): New functions.
25617 (get_addr_stridx): Add an offset_out parameter.
25618 Use compare_nonzero_chars.
25619 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25620 (new_strinfo): Update after above changes to strinfo.
25621 (set_endptr_and_length): Set full_string_p.
25622 (get_string_length): Update after above changes to strinfo.
25623 (unshare_strinfo): Update call to new_strinfo.
25624 (maybe_invalidate): Likewise.
25625 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25626 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25627 as a uhwi instead of an shwi. Update after above changes to
25628 strinfo and new_strinfo.
25629 (zero_length_string): Assert that chainsi contains full strings.
25630 Use zero_length_string_p. Update call to new_strinfo.
25631 (adjust_related_strinfos): Update after above changes to strinfo.
25632 Copy full_string_p from origsi.
25633 (adjust_last_stmt): Use zero_length_string_p.
25634 (handle_builtin_strlen): Update after above changes to strinfo and
25635 new_strinfo. Install the lhs as the string length if the previous
25636 entry didn't describe a full string.
25637 (handle_builtin_strchr): Update after above changes to strinfo
25639 (handle_builtin_strcpy): Likewise.
25640 (handle_builtin_strcat): Likewise.
25641 (handle_builtin_malloc): Likewise.
25642 (handle_pointer_plus): Likewise.
25643 (handle_builtin_memcpy): Likewise. Track nonzero characters
25644 that aren't necessarily followed by a nul terminator.
25645 (handle_char_store): Likewise.
25647 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25649 PR tree-optimization/80769
25650 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25651 for malloc and calloc. Document the new invariant that all related
25652 strinfos have delayed lengths or none do.
25653 (verify_related_strinfos): Move earlier in file.
25654 (set_endptr_and_length): New function, split out from...
25655 (get_string_length): ...here. Also set the lengths of related
25657 (zero_length_string): Assert that chainsi has known (rather than
25659 (adjust_related_strinfos): Likewise.
25661 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25663 PR tree-optimization/81136
25664 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25665 assert that two references with the same misalignment have the same
25666 compile-time misalignment if those compile-time misalignments
25669 2017-07-01 Andi Kleen <ak@linux.intel.com>
25671 * print-tree.c (print_node): Print all attributes.
25673 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25675 * cfg.c (scale_bbs_frequencies): New function.
25676 * cfg.h (scale_bbs_frequencies): Declare it.
25677 * cfgloopanal.c (single_likely_exit): Cleanup.
25678 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25680 (scale_loop_profile): Likewise.
25681 (loop_version): Likewise.
25682 (create_empty_loop_on_edge): Update.
25683 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25684 scale_loop_frequencies, scale_loop_profile, loopify,
25685 loop_version): Update prototypes.
25686 * modulo-sched.c (sms_schedule): Update.
25687 * predict.c (unlikely_executed_edge_p): Also check probability.
25688 (probably_never_executed_edge_p): Fix typo.
25689 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25690 * tree-parloops.c (gen_parallel_loop): Update.
25691 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25692 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25693 * tree-ssa-loop-split.c (split_loop): Update.
25694 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25695 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25696 (vect_loop_versioning): Update.
25697 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25699 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25701 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25703 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25705 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25706 to keep profile consistent.
25708 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25710 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25711 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25712 * profile-count.h (max_safe_multiplier): Make unsigned.
25713 (profile_count::guessed_zero): New.
25715 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25717 * bb-reorder.c (fix_up_crossing_landing_pad,
25718 fix_crossing_conditional_branches): Use make_single_succ_edge
25719 to keep profile consistent.
25721 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25723 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25726 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25729 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25730 the right scopes, make sure cond_jump isn't preserved between multiple
25731 iterations. Search for fallthru edge whenever there are 3+ edges and
25732 use find_fallthru_edge for it.
25734 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25736 Patch by Alexander Monakov <amonakov@ispras.ru>
25737 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25738 probabilities consistently.
25740 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25742 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25743 emit_cmp_and_jump_insns.
25745 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25748 * tree-inline.c (expand_call_inline): Combine profile statuses.
25750 2017-06-30 Andrew Pinski <apinski@cavium.com>
25752 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25753 fold_stmt returned true.
25755 2017-06-30 Nathan Sidwell <nathan@acm.org>
25757 * ggc.h (empty_string): Delete.
25758 * cfgexpand.c (expand_asm_stmt): Use plain "".
25759 * optabs.c (expand_asm_memory_barrier): Likewise.
25760 * stringpool.c (empty_string): Delete.
25761 (digit_vector, digit_string): Delete.
25762 (ggc_alloc_string): Use plain "", don't optimize single digit
25763 strings. Use ggc_alloc_atomic.
25765 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25767 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25768 comparison set and one other set, use the cost of the non-comparison
25771 2017-06-30 Nathan Sidwell <nathan@acm.org>
25773 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25776 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25778 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25779 loops. Remove now unneeded calls to gimple_switch_set_label() that
25780 just set removed labels to NULL_TREE.
25782 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25784 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25785 (set_range_info): ...here. Only call set_range_info_raw if domain
25787 (set_nonzero_bits): Call set_range_info_raw.
25788 * tree-ssanames.h (set_range_info_raw): New.
25790 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25793 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25794 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25795 of nonimmediate_operand and <store_mask_constraint> instead of m
25796 for the input operand. For V8FI iterator, always split if input
25797 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25798 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25799 <store_mask_predicate> instead of register_operand and
25800 <store_mask_constraint> instead of v for the input operand. Make
25801 sure both operands aren't MEMs for if not <mask_applied>.
25803 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25805 * lto-wrapper.c (copy_file) Close both file descriptors before
25808 2017-06-30 Martin Liska <mliska@suse.cz>
25811 * multiple_target.c (create_dispatcher_calls): Make ifunc
25812 also for function that don't have calls or are not referenced.
25814 2017-06-30 Richard Biener <rguenther@suse.de>
25816 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25817 analyze the first scalar stmt. Move vector type computation
25818 for the BB case here from ...
25819 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25820 live operation processing in the SLP case properly.
25822 2017-06-30 Richard Biener <rguenther@suse.de>
25824 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25826 2017-06-30 Martin Liska <mliska@suse.cz>
25829 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25830 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25832 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25834 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25837 2017-06-30 Richard Biener <rguenther@suse.de>
25839 PR tree-optimization/81249
25840 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25841 condition reduction result to original scalar type.
25843 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25845 * profile-count.h (enum profile_quality): Fix typos and whitespace
25848 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25850 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25851 type for branch probabilities.
25853 2017-06-29 Julian Brown <julian@codesourcery.com>
25854 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25856 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25857 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25858 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25859 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25861 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25863 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25864 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25865 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25866 CC usage from generic code to here.
25867 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25868 CC usage into the target macros.
25870 2017-06-29 Maya Rashish <coypu@sdf.org>
25872 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25875 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25877 * arm/arm-builtins.c: Include profile-count.h
25878 * except.c (sjlj_emit_function_enter): Use
25879 profile_probability::unlikely.
25881 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25883 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25884 and tocrel_offset be pointer args rather than implicitly using
25886 (legitimate_constant_pool_address_p, rs6000_emit_move,
25887 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25888 tocrel_offset and use in toc_relative_expr_p call.
25889 (print_operand, print_operand_address): Use static tocrel_base_oac
25890 and tocrel_offset_oac.
25891 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25894 2017-06-29 Maya Rashish <coypu@sdf.org>
25896 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25898 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25900 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25901 objects, take into account only the alignment of 'op0' and 'mode1' if
25904 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25906 * ccmp.c (ccmp_tree_comparison_p): New function.
25907 (ccmp_candidate_p): Update to use above function.
25908 (get_compare_parts): New function.
25909 (expand_ccmp_next): Update to use new functions.
25910 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25912 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25913 take mode as argument.
25914 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25915 * expr.c (expand_expr_real_1): Pass mode as argument.
25917 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25919 * combine.c (combine_instructions): Print insns to dump_file, together
25922 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25924 * asan.c (asan_emit_stack_protection): Update.
25925 (create_cond_insert_point): Update.
25926 * auto-profile.c (afdo_propagate_circuit): Update.
25927 * basic-block.h (struct edge_def): Turn probability to
25928 profile_probability.
25929 (EDGE_FREQUENCY): Update.
25930 * bb-reorder.c (find_traces_1_round): Update.
25931 (better_edge_p): Update.
25932 (sanitize_hot_paths): Update.
25933 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25934 (make_single_succ_edge): Update.
25935 (check_bb_profile): Update.
25936 (dump_edge_info): Update.
25937 (update_bb_profile_for_threading): Update.
25938 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25939 probabilitycount to 0.
25940 * cfgbuild.c (compute_outgoing_frequencies): Update.
25941 * cfgcleanup.c (try_forward_edges): Update.
25942 (outgoing_edges_match): Update.
25943 (try_crossjump_to_edge): Update.
25944 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25945 (expand_gimple_tailcall): Update.
25946 (construct_init_block): Use make_single_succ_edge.
25947 (construct_exit_block): Use make_single_succ_edge.
25948 * cfghooks.c (verify_flow_info): Update.
25949 (redirect_edge_succ_nodup): Update.
25950 (split_edge): Update.
25951 (account_profile_record): Update.
25952 * cfgloopanal.c (single_likely_exit): Update.
25953 * cfgloopmanip.c (scale_loop_profile): Update.
25954 (set_zero_probability): Remove.
25955 (duplicate_loop_to_header_edge): Update.
25956 * cfgloopmanip.h (loop_version): Update prototype.
25957 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25958 (force_nonfallthru_and_redirect): Update.
25959 (update_br_prob_note): Update.
25960 (rtl_verify_edges): Update.
25961 (purge_dead_edges): Update.
25962 (rtl_lv_add_condition_to_bb): Update.
25963 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25964 * cgraphunit.c (init_lowered_empty_function): Update.
25965 (cgraph_node::expand_thunk): Update.
25966 * cilk-common.c: Include profile-count.h
25967 * dojump.c (inv): Remove.
25968 (jumpifnot): Update.
25969 (jumpifnot_1): Update.
25970 (do_jump_1): Update.
25972 (do_jump_by_parts_greater_rtx): Update.
25973 (do_compare_rtx_and_jump): Update.
25974 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25975 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25976 * dwarf2cfi.c: Include profile-count.h
25977 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25978 (sjlj_emit_dispatch_table): Likewise.
25979 * explow.c: Include profile-count.h
25980 * expmed.c (emit_store_flag_force): Update.
25981 (do_cmp_and_jump): Update.
25982 * expr.c (compare_by_pieces_d::generate): Update.
25983 (compare_by_pieces_d::finish_mode): Update.
25984 (emit_block_move_via_loop): Update.
25985 (store_expr_with_bounds): Update.
25986 (store_constructor): Update.
25987 (expand_expr_real_2): Update.
25988 (expand_expr_real_1): Update.
25989 * expr.h (try_casesi, try_tablejump): Update prototypes.
25990 * gimple-pretty-print.c (dump_probability): Update.
25991 (dump_profile): New.
25992 (dump_gimple_label): Update.
25993 (dump_gimple_bb_header): Update.
25994 * graph.c (draw_cfg_node_succ_edges): Update.
25995 * hsa-gen.c (convert_switch_statements): Update.
25996 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25997 (find_if_case_1): Update.
25998 (find_if_case_2): Update.
25999 * internal-fn.c (expand_arith_overflow_result_store): Update.
26000 (expand_addsub_overflow): Update.
26001 (expand_neg_overflow): Update.
26002 (expand_mul_overflow): Update.
26003 (expand_vector_ubsan_overflow): Update.
26004 * ipa-cp.c (good_cloning_opportunity_p): Update.
26005 * ipa-split.c (split_function): Use make_single_succ_edge.
26006 * ipa-utils.c (ipa_merge_profiles): Update.
26007 * loop-doloop.c (add_test): Update.
26008 (doloop_modify): Update.
26009 * loop-unroll.c (compare_and_jump_seq): Update.
26010 (unroll_loop_runtime_iterations): Update.
26011 * lra-constraints.c (lra_inheritance): Update.
26012 * lto-streamer-in.c (input_cfg): Update.
26013 * lto-streamer-out.c (output_cfg): Update.
26014 * mcf.c (adjust_cfg_counts): Update.
26015 * modulo-sched.c (sms_schedule): Update.
26016 * omp-expand.c (expand_omp_for_init_counts): Update.
26017 (extract_omp_for_update_vars): Update.
26018 (expand_omp_ordered_sink): Update.
26019 (expand_omp_for_ordered_loops): Update.
26020 (expand_omp_for_generic): Update.
26021 (expand_omp_for_static_nochunk): Update.
26022 (expand_omp_for_static_chunk): Update.
26023 (expand_cilk_for): Update.
26024 (expand_omp_simd): Update.
26025 (expand_omp_taskloop_for_outer): Update.
26026 (expand_omp_taskloop_for_inner): Update.
26027 * omp-simd-clone.c (simd_clone_adjust): Update.
26028 * optabs.c (expand_doubleword_shift): Update.
26029 (expand_abs): Update.
26030 (emit_cmp_and_jump_insn_1): Update.
26031 (expand_compare_and_swap_loop): Update.
26032 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26033 * predict.c (predictable_edge_p): Update.
26034 (edge_probability_reliable_p): Update.
26035 (set_even_probabilities): Update.
26036 (combine_predictions_for_insn): Update.
26037 (combine_predictions_for_bb): Update.
26038 (propagate_freq): Update.
26039 (estimate_bb_frequencies): Update.
26040 (force_edge_cold): Update.
26041 * profile-count.c (profile_count::dump): Add missing space into dump.
26042 (profile_count::debug): Add newline.
26043 (profile_count::differs_from_p): Explicitly convert to unsigned.
26044 (profile_count::stream_in): Update.
26045 (profile_probability::dump): New member function.
26046 (profile_probability::debug): New member function.
26047 (profile_probability::differs_from_p): New member function.
26048 (profile_probability::differs_lot_from_p): New member function.
26049 (profile_probability::stream_in): New member function.
26050 (profile_probability::stream_out): New member function.
26051 * profile-count.h (profile_count_quality): Rename to ...
26052 (profile_quality): ... this one.
26053 (profile_probability): New.
26054 (profile_count): Update.
26055 * profile.c (compute_branch_probabilities): Update.
26056 * recog.c (peep2_attempt): Update.
26057 * sched-ebb.c (schedule_ebbs): Update.
26058 * sched-rgn.c (find_single_block_region): Update.
26059 (compute_dom_prob_ps): Update.
26060 (schedule_region): Update.
26061 * sel-sched-ir.c (compute_succs_info): Update.
26062 * stmt.c (struct case_node): Update.
26063 (do_jump_if_equal): Update.
26064 (get_outgoing_edge_probs): Update.
26065 (conditional_probability): Update.
26066 (emit_case_dispatch_table): Update.
26067 (expand_case): Update.
26068 (expand_sjlj_dispatch_table): Update.
26069 (emit_case_nodes): Update.
26070 * targhooks.c: Update.
26071 * tracer.c (better_p): Update.
26072 (find_best_successor): Update.
26073 * trans-mem.c (expand_transaction): Update.
26074 * tree-call-cdce.c: Update.
26075 * tree-cfg.c (gimple_split_edge): Upate.
26076 (move_sese_region_to_fn): Upate.
26077 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26078 * tree-eh.c (lower_resx): Upate.
26079 (cleanup_empty_eh_move_lp): Upate.
26080 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26081 * tree-inline.c (copy_edges_for_bb): Update.
26082 (copy_cfg_body): Update.
26083 * tree-parloops.c (gen_parallel_loop): Update.
26084 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26085 (gimple_gen_time_profiler): Update.
26086 * tree-ssa-dce.c (remove_dead_stmt): Update.
26087 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26088 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26089 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26090 (unloop_loops): Update.
26091 (try_peel_loop): Update.
26092 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26093 * tree-ssa-loop-split.c (connect_loops): Update.
26094 (split_loop): Update.
26095 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26096 (hoist_guard): Update.
26097 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26098 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26099 (value_replacement): Update.
26100 * tree-ssa-reassoc.c (branch_fixup): Update.
26101 * tree-ssa-tail-merge.c (replace_block_by): Update.
26102 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26103 (create_edge_and_update_destination_phis): Update.
26104 (compute_path_counts): Update.
26105 (recompute_probabilities): Update.
26106 (update_joiner_offpath_counts): Update.
26107 (freqs_to_counts_path): Update.
26108 (duplicate_thread_path): Update.
26109 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26110 (struct switch_conv_info): Update.
26111 (gen_inbound_check): Update.
26112 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26113 (vect_do_peeling): Update.
26114 (vect_loop_versioning): Update.
26115 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26116 (optimize_mask_stores): Update.
26117 * ubsan.c (ubsan_expand_null_ifn): Update.
26118 * value-prof.c (gimple_divmod_fixed_value): Update.
26119 (gimple_divmod_fixed_value_transform): Update.
26120 (gimple_mod_pow2): Update.
26121 (gimple_mod_pow2_value_transform): Update.
26122 (gimple_mod_subtract): Update.
26123 (gimple_mod_subtract_transform): Update.
26124 (gimple_ic): Update.
26125 (gimple_stringop_fixed_value): Update.
26126 (gimple_stringops_transform): Update.
26127 * value-prof.h: Update.
26129 2017-06-29 Carl Love <cel@us.ibm.com>
26131 * config/rs6000/rs6000-c.c: Add support for built-in functions
26132 vector signed int vec_signed (vector float);
26133 vector signed long long vec_signed (vector double);
26134 vector signed int vec_signed2 (vector double, vector double);
26135 vector signed int vec_signede (vector double);
26136 vector signed int vec_signedo (vector double);
26137 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26138 instruction generator.
26139 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26140 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26141 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26143 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26144 vunsignede_v2df): Add define_expands.
26145 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26146 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26147 VEC_UNSIGNEDO): Add definitions.
26148 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26149 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26150 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26151 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26152 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26153 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26154 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26155 vec_unsignede, vec_unsignedo): Add builtin defines.
26156 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26158 * doc/extend.texi: Update the built-in documentation file for the
26159 new built-in functions.
26161 2017-06-29 Richard Biener <rguenther@suse.de>
26163 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26164 reduction chains to LOOP_VINFO_REDUCTIONS.
26165 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26166 SLP reductions after processing reduction chains.
26168 2017-06-29 Nathan Sidwell <nathan@acm.org>
26170 * builtins.c (fold_builtin_FUNCTION): Use
26171 lang_hooks.decl_printable_name.
26173 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26175 PR middle-end/81194
26176 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26177 with only one label.
26178 * stmt.c (expand_case): Assert NCASES is greater than one.
26180 2017-06-29 Richard Biener <rguenther@suse.de>
26182 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26184 (group_case_labels): Likewise.
26185 (find_taken_edge): Push sanity checking on val to workers...
26186 (find_taken_edge_cond_expr): ... here
26187 (find_taken_edge_switch_expr): ... and here, handle cases
26188 with just a default label.
26189 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26190 (group_case_labels): Likewise.
26191 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26192 group_case_labels does anything cleanup the CFG again.
26194 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26196 PR tree-optimization/81196
26197 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26198 exit condition comparing two IVs.
26200 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26202 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26203 profile to the dummy entry at the end of the list of architectures.
26204 * config/arm/arm-cpu-cdata.h: Regenerated.
26206 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26207 Michael Collison <michael.collison@arm.com>
26210 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26212 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26213 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26214 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26215 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26216 of shift when the shift amount is masked with constant equal to
26217 the size of the mode.
26218 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26221 2017-06-29 Martin Liska <mliska@suse.cz>
26223 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26225 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26228 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26231 (edge_growth_cache_entry::edge_growth_cache_entry): New
26233 (reset_edge_growth_cache): Update to use constructor.
26235 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26237 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26238 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26239 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26241 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26243 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26244 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26246 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26248 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26249 (*-linux-uclibc*): Add t-uclibc tmake_file.
26250 * config/t-musl: New.
26251 * config/t-uclibc: New.
26253 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26255 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26257 (gen_comm_data): Emit architectural setting of arch_prof.
26258 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26260 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26261 (armv8-m.base, armv8-m.main): Likewise.
26262 * arm-protos.h (arm_build_target): Add profile field.
26263 (arch_option): Likewise.
26264 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26266 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26267 arm_active_target.profile.
26269 2017-06-28 Richard Biener <rguenther@suse.de>
26271 PR middle-end/81227
26272 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26273 TYPE_OVERFLOW_WRAPS.
26274 * match.pd (negate_expr_p): Likewise.
26275 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26276 fold_build2, not fold_binary.
26278 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26280 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26281 Convert memory address to Pmode.
26282 (aarch64_print_operand): Assert MEM operands are always Pmode.
26284 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26287 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26288 Remove redundant if.
26289 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26290 * config/arm/aarch-common-protos.h
26291 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26292 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26294 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26297 * multiple_target.c (create_dispatcher_calls): Set the default
26298 clone to be static, not public.
26300 2017-06-28 Richard Biener <rguenther@suse.de>
26302 * tree-vect-loop.c (vectorizable_reduction): Move special
26303 cond reduction IV var creation ...
26304 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26305 parameter. Use STMT_VINFO_VECTYPE.
26306 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26309 2017-06-28 Martin Liska <mliska@suse.cz>
26312 * ipa-visibility.c (non_local_p): Handle visibility.
26314 2017-06-28 Martin Liska <mliska@suse.cz>
26317 * common.opt: Add IntegerRange to various options.
26318 * opt-functions.awk (integer_range_info): New function.
26319 * optc-gen.awk: Add integer_range_info to cl_options struct.
26320 * opts-common.c (decode_cmdline_option): Handle
26321 CL_ERR_INT_RANGE_ARG.
26322 (cmdline_handle_error): Likewise.
26323 * opts.c (print_filtered_help): Show valid interval in
26324 when --help is provided.
26325 * opts.h (struct cl_option): Add range_min and range_max fields.
26326 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26328 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26330 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26331 (x * C EQ/NE y * C): New transformation.
26333 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26335 * genmultilib (combination_space): Accept '+' in option names.
26337 2017-06-28 Martin Liska <mliska@suse.cz>
26340 * asan.c (instrument_derefs): Bail out inner references
26341 that are hard register variables.
26343 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26346 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26347 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26349 2017-06-28 Richard Biener <rguenther@suse.de>
26351 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26352 (vect_get_slp_defs): Adjust.
26353 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26355 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26357 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26358 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26359 (vectorizable_reduction): Adjust.
26360 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26362 (vect_get_slp_defs): Likewise.
26363 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26364 (vectorizable_bswap): Adjust.
26365 (vectorizable_call): Likewise.
26366 (vectorizable_conversion): Likewise.
26367 (vectorizable_assignment): Likewise.
26368 (vectorizable_shift): Likewise.
26369 (vectorizable_operation): Likewise.
26370 (vectorizable_store): Likewise.
26371 (vectorizable_condition): Likewise.
26372 (vectorizable_comparison): Likewise.
26374 2017-06-28 Michael Collison <michael.collison@arm.com>
26377 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26379 (arm_gen_movmemqi): Removed unused variable 'i'.
26380 Convert 'for' loop into 'while' loop.
26381 (arm_expand_prologue): Remove last unnecessary set of insn.
26382 (thumb_pop): Remove unused variable 'pushed_words'.
26383 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26385 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26387 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26388 * config/s390/s390-protos.h: Add prototype of
26389 s390_rel_address_ok_p.
26390 * config/s390/s390.c (s390_got_symbol): New function.
26391 (s390_rel_address_ok_p): New function.
26392 (legitimize_pic_address): Use s390_rel_address_ok_p.
26393 (s390_load_got): Use s390_got_symbol.
26394 (s390_option_override): Issue error if
26395 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26396 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26398 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26400 2017-06-27 Andrew Pinski <apinski@cavium.com>
26402 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26403 (X * copysign (1.0, X)): New pattern.
26404 (X * copysign (1.0, -X)): New pattern.
26405 (copysign (-1.0, CST)): New pattern.
26407 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26409 * genmultilib (combination_space): Remove variable.
26410 Validate reuse rules against regular expression for any sequence
26411 of multilib options in any order.
26413 2017-06-27 Michael Collison <michael.collison@arm.com>
26415 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26416 call aarch64_split_simd_combine.
26417 * (aarch64_combine_internal<mode>): Delete pattern.
26418 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26419 Allow register and subreg operands.
26421 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26423 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26424 specific need, just fallback on defaults.
26425 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26427 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26428 Olivier Hainque <hainque@adacore.com>
26430 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26432 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26433 targets. Pick a default if no particular attempt applied.
26434 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26437 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26439 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26440 (x86_64-wrs-vxworks7): Likewise.
26442 2017-06-27 Marek Polacek <polacek@redhat.com>
26445 * ubsan.c (instrument_null): Check get_base_address's result for null.
26447 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26449 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26451 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26453 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26454 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26455 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26456 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26457 New function types.
26458 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26459 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26460 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26461 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26462 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26463 BUILT_IN_FEUPDATEENV): New builtins.
26464 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26465 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26466 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26467 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26469 (builtin_structptr_types): Adjust size.
26470 * tree.c (builtin_structptr_types): Add four entries.
26472 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26473 Olivier Hainque <hainque@adacore.com>
26475 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26476 (TLS_SYM): New local macro, forcing reference to __tls__ on
26477 link command lines for VxWorks 7 RTPs, triggering initialization
26479 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26480 OS features TLS support, true for RTPs on VxWorks 7.
26481 * config/vxworks.c (vxworks_override_options): Setup emutls
26484 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26486 * predict.c (test_prediction_value_range): Use -1U instead of -1
26487 to avoid narrowing conversion warning.
26488 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26489 to avoid narrowing conversion warning.
26490 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26492 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26494 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26496 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26497 64bit configurations.
26498 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26499 (SIZE_TYPE): Likewise.
26500 * config/vxworks.c (vxworks_emutls_var_fields): Use
26501 long_unsigned_type_node instead of unsigned_type_node as the offset
26502 field type, which is "pointer" mode in emutls.c.
26504 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26507 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26509 PR middle-end/81207
26510 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26511 gimple_vuse copying separately from gimple_vdef copying.
26513 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26515 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26518 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26519 Olivier Hainque <hainque@adacore.com>
26521 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26522 for all vxworks7 targets.
26523 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26524 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26525 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26526 variations for VX6/VX7 and 32/64bits later on in ...
26527 (VXWORKS_LIB_SPEC): Leverage new macros.
26528 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26529 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26531 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26533 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26534 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26536 2017-06-26 Carl Love <cel@us.ibm.com>
26538 * config/rs6000/rs6000-c.c: Add support for built-in functions
26539 vector bool char vec_reve (vector bool char);
26540 vector signed char vec_reve (vector signed char);
26541 vector unsigned char vec_reve (vector unsigned char);
26542 vector bool int vec_reve (vector bool int);
26543 vector signed int vec_reve (vector signed int);
26544 vector unsigned int vec_reve (vector unsigned int);
26545 vector bool long long vec_reve (vector bool long long);
26546 vector signed long long vec_reve (vector signed long long);
26547 vector unsigned long long vec_reve (vector unsigned long long);
26548 vector bool short vec_reve (vector bool short);
26549 vector signed short vec_reve (vector signed short);
26550 vector double vec_reve (vector double);
26551 vector float vec_reve (vector float);
26552 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26553 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26554 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26555 (altivec_vreve): New pattern.
26556 * config/rs6000/altivec.h (vec_reve): New define.
26557 * doc/extend.texi (vec_rev): Update the built-in documentation file
26558 for the new built-in functions.
26560 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26562 PR tree-optimization/71815
26563 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26565 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26567 (slsr_process_phi): Likewise.
26568 (replace_uncond_cands_and_profitable_phis): Don't replace a
26569 multiply candidate with a stride of 1 (copy or cast).
26570 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26572 (lowest_cost_path): Likewise.
26573 (total_savings): Likewise.
26575 2017-06-26 Richard Biener <rguenther@suse.de>
26578 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26579 Use def_builtin_pure for all gather builtins.
26581 2017-06-26 Richard Biener <rguenther@suse.de>
26583 PR tree-optimization/81203
26584 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26585 non-dominating BBs.
26587 2017-06-26 Marek Polacek <polacek@redhat.com>
26590 * doc/invoke.texi: Document -Wmultistatement-macros.
26592 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26594 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26595 arm_neon_ok_no_float_abi effective target.
26597 2017-06-26 Richard Biener <rguenther@suse.de>
26599 PR tree-optimization/80928
26600 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26601 (copy_bbs): Set BB_DUPLICATED flag early.
26602 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26604 (execute_on_shrinking_pred): Likewise.
26605 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26606 BB_DUPLICATED blocks.
26607 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26608 iterate over all PHIs considering removal of *gsi.
26610 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26612 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26615 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26617 * config/rs6000/rs6000-string.c: (expand_block_clear,
26618 do_load_for_compare, select_block_compare_mode,
26619 compute_current_alignment, expand_block_compare,
26620 expand_strncmp_align_check, expand_strn_compare,
26621 expand_block_move, rs6000_output_load_multiple)
26622 Move functions related to string/block move/compare
26623 to a separate file.
26624 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26625 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26626 for this function which is now used in two files.
26627 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26628 * config.gcc: Add rs6000-string.o to extra_objs for
26629 targets powerpc*-*-* and rs6000*-*-*.
26631 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26634 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26635 32-bit, since indexed is not valid for DImode.
26636 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26637 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26638 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26639 (define_peephole2 for Altivec d-form store): Likewise.
26642 * multiple_target.c (create_dispatcher_calls): Only create the
26643 dispatcher call if the function is the default clone of a
26644 versioned function.
26646 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26648 PR middle-end/80902
26649 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26650 a call, force the call to not be a tail call.
26652 2017-06-23 Jeff Law <law@redhat.com>
26654 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26657 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26659 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26660 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26661 handling for early expansion of vector shifts (sl,sr,sra,rl).
26662 (builtin_function_type): Add vector shift right instructions
26663 to the unsigned argument list.
26665 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26667 rtl-optimizatoin/79286
26668 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26669 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26670 trap. PIC register plus a const unspec without offset can never trap.
26672 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26674 * tree.h (builtin_structptr_type): New type.
26675 (builtin_structptr_types): Declare new array.
26676 * tree.c (builtin_structptr_types): New array.
26677 (free_lang_data, build_common_tree_nodes): Use it.
26679 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26682 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26685 2017-06-22 Matt Turner <mattst88@gmail.com>
26687 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26688 Lake models to skylake case. Assume skylake for unknown
26689 models with clflushopt.
26691 2017-06-22 Jeff Law <law@redhat.com>
26693 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26694 frame sizes that do not satisfy aarch64_uimm12_shift.
26696 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26698 * profile-count.h (apply_probability,
26699 apply_scale, probability_in): Fix checks for zero.
26701 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26703 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26704 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26706 2016-06-22 Richard Biener <rguenther@suse.de>
26708 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26709 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26710 REDUC_MAX_EXPR support.
26711 (vectorizable_reduction): Likewise.
26712 (vect_create_epilog_for_reduction): Likewise.
26714 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26716 * match.pd (A / (1 << B) -> A >> B): New.
26717 * generic-match-head.c: Include optabs-tree.h.
26718 * gimple-match-head.c: Likewise.
26719 * optabs-tree.h (target_supports_op_p): New.
26720 * optabs-tree.c (target_supports_op_p): New.
26722 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26724 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26725 $gcc_cv_ld --help output.
26726 (gcc_cv_ld_demangle): Likewise.
26727 (gcc_cv_ld_eh_frame_hdr): Likewise.
26728 (gcc_cv_ld_pie): Likewise.
26729 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26730 (gcc_cv_ld_buildid): Likewise.
26731 (gcc_cv_ld_sysroot): Likewise.
26732 (ld_bndplt_support): Likewise.
26733 (ld_pushpopstate_support): Likewise.
26734 * configure: Regenerate.
26735 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26737 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26740 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26741 operands indexes to avoid gap between operands and match_dups.
26743 2017-06-21 Andrew Pinski <apinski@cavium.com>
26745 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26746 Increment Arith_shift and Arith_shift_reg by 1.
26747 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26749 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26750 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26751 (aarch64_strip_extend): Add new argument and test for it.
26752 (aarch64_cheap_mult_shift_p): New function.
26753 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26754 add a cost if it is true.
26755 Update calls to aarch64_strip_extend.
26756 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26758 2017-06-21 Andrew Pinski <apinski@cavium.com>
26760 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26762 (thunderxt88): Likewise.
26763 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26764 (thunderx_prefetch_tune): New variable.
26765 (thunderx2t99_prefetch_tune): Update for the correct values.
26766 (thunderxt88_tunings): New variable.
26767 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26768 generic_prefetch_tune.
26769 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26771 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26773 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26774 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26775 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26776 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26777 (aarch64_atomic_cas<mode>, GPI): Likewise.
26779 2017-06-21 Martin Liska <mliska@suse.cz>
26781 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26782 statements on cold and hot labels.
26783 * predict.c (tree_estimate_probability_bb): Remove the
26784 prediction from this place.
26786 2017-06-21 Martin Liska <mliska@suse.cz>
26788 PR tree-optimization/79489
26789 * gimplify.c (maybe_add_early_return_predict_stmt): New
26791 (gimplify_return_expr): Call the function.
26792 * predict.c (tree_estimate_probability_bb): Remove handling
26794 * predict.def: Update comment about early return predictor.
26795 * gimple-predict.h (is_gimple_predict): New function.
26796 * predict.def: Change default value of early return to 66.
26797 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26799 * passes.def: Put pass_strip_predict_hints to the beginning of
26802 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26804 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26805 FUNCTION_DECL declarations.
26806 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26808 (dwaf2out_decl): Likewise.
26809 * godump.c (go_early_global_decl): Skip call to the real debug hook
26810 for FUNCTION_DECL declarations.
26811 * passes.c (rest_of_decl_compilation): Skip call to the
26812 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26813 -fdump-go-spec is passed.
26815 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26817 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26818 Reorder for compactness.
26819 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26820 (def_builtin_pure, def_builtin_pure2): New functions.
26821 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26823 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26825 * match.pd (nop_convert): New predicate.
26826 ((A +- CST1) +- CST2): Allow some NOP conversions.
26828 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26831 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26832 with ctors/dtors if GOVD_SHARED is set.
26834 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26836 * config/aarch64/aarch64.md (movti_aarch64):
26837 Emit mov rather than orr.
26838 (movtf_aarch64): Likewise.
26839 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26840 Emit mov rather than orr.
26842 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26844 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26845 Swap alternatives, make integer dup more expensive.
26847 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26849 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26850 Return true for non-tls symbols.
26852 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26854 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26855 (cortex-a75): Likewise.
26856 (cortex-a75.cortex-a55): Likewise.
26857 * config/aarch64/aarch64-tune.md: Regenerate.
26858 * doc/invoke.texi (-mtune): Document new values for -mtune.
26860 2017-06-21 Tom de Vries <tom@codesourcery.com>
26862 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26863 stack_size feature.
26864 (Effective-Target Keywords, Other attributes): Suggest using
26865 dg-add-options stack_size feature to get stack limit in stack_size
26866 effective target documentation.
26868 2017-06-21 Julian Brown <julian@codesourcery.com>
26869 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26871 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26872 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26873 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26875 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26876 attribute type list for neon_multiply.
26877 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26878 attribute type list for neon_multiply.
26879 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26880 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26881 attribute type list for neon_multiply.
26882 * config/arm/types.md (crypto_pmull): Add.
26883 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26884 attribute type list.
26886 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26888 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26891 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26893 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26894 to make sure not to dereference a NULL cost_classes_ptr pointer.
26896 2017-06-20 Carl Love <cel@us.ibm.com>
26898 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26899 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26900 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26901 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26902 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26903 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26904 VMULOSW): New enum "unspec" values.
26905 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26906 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26907 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26908 altivec_vmulosw): New patterns.
26909 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26910 VMULOSW): Add definitions.
26912 2017-06-20 Julia Koval <julia.koval@intel.com>
26914 * config/i386/i386.c: Fix rounding expand for new pattern.
26915 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26917 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26919 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26920 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26922 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26924 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26927 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26929 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26930 architecture, then by implementer ID.
26931 * config/aarch64/aarch64-tune.md: Regenerate.
26933 2017-06-20 Richard Biener <rguenther@suse.de>
26935 PR middle-end/81097
26936 * fold-const.c (split_tree): Fold to type before negating.
26938 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26940 * diagnostic-show-locus.c
26941 (selftest::test_fixit_deletion_affecting_newline): New function.
26942 (selftest::diagnostic_show_locus_c_tests): Call it.
26944 2017-06-20 Andreas Schwab <schwab@suse.de>
26947 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26950 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26952 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26953 __ARM_FEATURE_COPROC according to support.
26955 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26957 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26958 Rewritten to avoid overflow for > 32-bit pointers.
26961 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26962 by removing enum keyword.
26963 (ubsan_type_descriptor): Likewise. Formatting fix.
26966 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26967 splitter): Require TARGET_SSE2 in the condition.
26969 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26972 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26973 for doing vector set of SFmode on ISA 3.0.
26974 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26975 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26977 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26978 SFmode value into a V4SF variable that was extracted from another
26979 V4SF variable without converting the element to double precision
26980 and back to single precision vector format.
26981 (vsx_insert_extract_v4sf_p9_2): Likewise.
26983 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26985 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26986 in UWHI to avoid undefined overflow.
26989 * ubsan.h (enum ubsan_encode_value_phase): New.
26990 (ubsan_encode_value): Change second argument to
26991 enum ubsan_encode_value_phase with default value of
26992 UBSAN_ENCODE_VALUE_GENERIC.
26993 * ubsan.c (ubsan_encode_value): Change second argument to
26994 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26995 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26996 create_tmp_var_raw instead of create_tmp_var and use a
26998 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26999 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27000 ubsan_encode_value callers.
27003 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27004 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27005 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27007 2017-06-19 Richard Biener <rguenther@suse.de>
27009 PR middle-end/81118
27010 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27011 estimates if we changed anything.
27013 2017-06-19 Richard Biener <rguenther@suse.de>
27015 PR tree-optimization/80887
27016 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27017 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27018 simplified lookups, then reset mprts_hook.
27019 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27021 (try_to_simplify): Likewise.
27023 2017-06-19 Martin Liska <mliska@suse.cz>
27026 * gimplify.c (gimplify_switch_expr):
27027 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27029 2017-06-19 Martin Liska <mliska@suse.cz>
27031 * doc/install.texi: Document that PGO runs in 4 stages.
27033 2017-06-19 Martin Liska <mliska@suse.cz>
27036 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27037 to dispatcher function name.
27038 * multiple_target.c (replace_function_decl): New function.
27039 (create_dispatcher_calls): Redirect both edges and references.
27041 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27043 * profile-count.c (profile_count::dump): Dump quality.
27044 (profile_count::differs_from_p): Update for unsigned val.
27045 * profile-count.h (profile_count_quality): New enum.
27046 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27048 2017-06-19 Richard Biener <rguenther@suse.de>
27050 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27051 struct function as arg.
27052 (estimate_numbers_of_iterations): Export overload with loop arg.
27053 (free_numbers_of_iterations_estimates_loop): Use an overload of
27054 free_numbers_of_iterations_estimates instead.
27055 * tree-cfg.c (remove_bb): Adjust.
27056 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27057 * tree-parloops.c (gen_parallel_loop): Likewise.
27058 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27060 (tree_unroll_loops_completely): Likewise.
27061 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27062 Use an overload instead and export.
27063 (estimated_loop_iterations): Adjust.
27064 (max_loop_iterations): Likewise.
27065 (likely_max_loop_iterations): Likewise.
27066 (estimate_numbers_of_iterations): Take struct function as arg
27068 (loop_exits_before_overflow): Adjust.
27069 (free_numbers_of_iterations_estimates_loop): Use an overload.
27070 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27071 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27073 2017-06-19 Richard Biener <rguenther@suse.de>
27076 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27077 RANGE_EXPR conservatively.
27079 2017-06-16 Carl Love <cel@us.ibm.com>
27081 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27082 definitions for vec_float, vec_float2, vec_floato,
27083 vec_floate built-ins.
27084 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27085 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27087 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27088 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27089 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27090 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27091 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27092 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27093 vec_floato): Add builtin defines.
27094 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27095 Update the built-in documentation file for the new built-in
27098 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27100 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27101 (mthumb): Mark as the negative of -marm.
27103 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27105 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27107 (ARM Options, -mtune): Document that this accepts the same
27108 extension options as -mcpu.
27109 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27113 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27114 permitted extensions.
27116 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27118 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27119 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27120 (armv8-m.main): Add option +nodsp.
27121 * config/arm/arm-cpu-cdata.h: Regenerated.
27123 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27125 * config/arm/t-fuchsia: New file.
27126 * config.gcc (arm*-*-fuchsia*): Use it.
27128 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27130 * config/arm/t-symbian: Rewrite for new option infrastructure.
27132 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27134 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27135 (MULTILIB_REQUIRED): Likewise.
27137 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27139 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27140 (MULTILIB_RESUE): Likewise.
27141 (MULTILIB_MATCHES): Likewise.
27142 (MULTLIB_REQUIRED): Likewise.
27144 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27146 * config/arm/t-rtems: Rewrite for new option framework.
27148 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27150 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27151 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27152 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27153 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27154 * config/arm/t-multilib: ... here.
27155 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27156 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27157 armv7-a and armv8*-a when A-profile libraries have not been built.
27158 * config/arm/t-rmprofile: Rewrite.
27160 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27162 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27163 with a backslash. Remove the backslash after substituting unescaped
27165 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27167 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27169 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27170 the driver to pass through the default setting of -mfloat-abi.
27171 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27173 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27174 (all_feat_combs): New rule.
27175 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27177 * config/arm/t-aprofile: Rewrite.
27179 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27181 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27182 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27183 fpu is not specified by the user/command-line.
27184 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27185 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27186 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27187 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27188 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27189 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27191 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27193 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27194 * config/arm/t-arm-elf: Rewritten.
27196 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27198 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27199 have some floating-point instructions.
27200 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27201 (TARGET_MAYBE_HARD_FLOAT): New macro.
27202 * config/arm/arm-builtins.c (arm_init_builtins): Use
27203 TARGET_MAYBE_HARD_FLOAT.
27204 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27206 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27208 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27209 (configargs.h): Include it.
27210 (arm_print_hint_for_fpu_option): New function.
27211 (arm_parse_fpu_option): New function.
27212 (candidate_extension): New class.
27213 (arm_canon_for_multilib): New function.
27214 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27215 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27216 (ARCH_CANONICAL_SPECS): New macro.
27217 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27219 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27221 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27222 are set after handling multilib fragments. Set target_cpu_default2
27225 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27227 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27229 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27230 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27232 (check_arch): Likewise.
27233 * config/arm/arm.c (arm_configure_build_target): Handle
27234 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27235 options in the default.
27237 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27239 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27240 when an option is an alias of another.
27241 * config/arm/parsecpu.awk (optalias): New parser token.
27242 (gen_comm_data): Mark non-alias options as such. Emit entries
27243 for extension aliases.
27244 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27245 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27246 (armv6kz, armv6zk, armv6t2): Likewise.
27247 (armv7): Make vfpv3-d16 an alias.
27248 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27250 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27251 Sort in canonical order.
27252 (armv8-a): Sort in canonical order.
27253 (armv8.1-a, armv8.2-a): Likewise.
27254 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27256 (cortex-a9): Sort in canonical order.
27257 * config/arm/arm.c (selftests.h): Include it.
27258 (arm_test_cpu_arch_data): New function.
27259 (arm_run_self_tests): New function.
27260 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27261 (targetm): Move declaration to the end of the file.
27262 * arm-cpu-cdata.h: Regenerated.
27264 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27266 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27267 call to target_mode_check describing the type of option passed.
27268 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27269 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27270 arm_parse_cpu_option_name to match parameters against list of
27272 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27273 arm_arch_core_flags data structure.
27274 * config/arm/arm-cpu_cdata.h: Regenerated.
27276 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27278 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27280 (arm_print_hint_for_cpu_option): Likewise.
27281 (arm_print_hint_for_arch_option): Likewise.
27282 (arm_parse_cpu_option_name): Likewise.
27283 (arm_parse_arch_option_name): Likewise.
27284 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27285 of entries in the all_fpus list.
27286 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27287 (arm_parse_cpu_option_name): Declare.
27288 (arm_parse_arch_option_name): Declare.
27289 (arm_parse_option_features): Declare.
27290 (arm_intialize_isa): Declare.
27291 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27293 (gen_comm_data): ... here. Make definitions non-static.
27294 * config/arm/arm-cpu-data.h: Regenerated.
27295 * config/arm/arm-cpu-cdata.h: Regenerated.
27297 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27299 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27300 (cpu_arch_extension): New structure.
27301 (cpu_arch_option, arch_option, cpu_option): New structures.
27302 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27303 architecture types.
27304 (gen_data): Generate new format data tables.
27305 * config/arm/arm.c (cpu_tune): New structure.
27306 (cpu_option, processors): Delete.
27307 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27308 (arm_print_hint_for_cpu_option): ... this and ...
27309 (arm_print_hint_for_arch_option): ... this.
27310 (arm_parse_arch_cpu_name): Delete. Replace with ...
27311 (arm_parse_cpu_option_name): ... this and ...
27312 (arm_parse_arch_option_name): ... this.
27313 (arm_unrecognized_feature): Change type of target parameter to
27315 (arm_parse_arch_cpu_features): Delete. Replace with ...
27316 (arm_parse_option_features): ... this.
27317 (arm_configure_build_target): Rework to use new configuration data
27319 (arm_print_tune_info): Rework for new configuration data tables.
27320 * config/arm/arm-cpu-data.h: Regenerated.
27321 * config/arm/arm-cpu.h: Regenerated.
27323 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27325 * Makefile.in (OBJS): Move sbitmap.o from here ...
27326 (OBJS-libcommon): ... to here.
27328 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27330 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27331 (ISA_ALL_CRYPTO): New macro.
27332 (ISA_ALL_SIMD): New macro
27333 (ISA_ALL_FP): New macro.
27334 * config/arm/arm.c (fpu_bitlist): Update initializer.
27335 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27337 (arm9e): Add fpu. Add option for nofp
27338 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27339 (arm926ej-s, arm1026ej-s): Likewise.
27340 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27341 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27342 neon-fp16, neon-vfpv4, nofp and nosimd.
27343 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27344 (cortex-a8): Add fpu. Add option for nofp.
27345 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27346 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27347 (cortex-r4f): Add fpu.
27348 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27349 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27351 (cortex-r8): Likewise.
27352 (cortex-m4): Add fpu. Add option for nofp.
27353 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27354 (cortex-a17.cortex-a7): Likewise.
27355 (cortex-a32): Add fpu. Add options for crypto and nofp.
27356 (cortex-a35, cortex-a53): Likewise.
27357 (cortex-a57): Add fpu. Add option for crypto.
27358 (cortex-a72, cortex-a73): Likewise.
27359 (exynos-m1): Likewise.
27360 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27361 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27362 (cortex-m33): Add fpu. Add option for nofp.
27363 * config/arm/arm-cpu-cdata.h: Regenerated
27364 * config/arm/arm-cpu-data.h: Regenerated.
27366 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27368 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27369 (armv5te, armv5tej): Likewise.
27370 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27371 (armv7): Add options fp and vfpv3-d16.
27372 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27373 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27375 (armv7ve): Likewise.
27376 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27377 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27378 (armv8-a): Add nocrypto option.
27379 (armv8.1-a, armv8.2-a): Likewise.
27380 (armv8-m.main): add options fp, fp.dp and nofp.
27382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27384 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27386 (armv8-a+crc): Delete.
27387 (armv8.1-a): Add options simd, crypto and nofp.
27388 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27389 (armv8.2-a+fp16): Delete.
27390 (armv8-m.main): Add option dsp.
27391 (armv8-m.main+dsp): Delete.
27392 (cortex-a8): Add fpu. Add nofp option.
27393 (cortex-a9): Add fpu. Add nofp and nosimd options.
27394 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27395 link to main cpu and architecture data structures.
27396 (gen_comm_data): Only put isa attributes from the main architecture
27398 (option): New statement for architecture and CPU entries.
27399 * arm.c (struct cpu_option): New structure.
27400 (struct processors): Add entry for options.
27401 (arm_unrecognized_feature): New function.
27402 (arm_parse_arch_cpu_name): Ignore any characters after the first
27404 (arm_parse_arch_cpu_feature): New function.
27405 (arm_configure_build_target): Separate out any CPU and architecture
27406 features and parse separately. Don't error out if -mfpu=auto is
27407 used with only an architecture string.
27408 (arm_print_asm_arch_directives): New function.
27409 (arm_file_start): Call it.
27410 * config/arm/arm-cpu-cdata.h: Regenerated.
27411 * config/arm/arm-cpu-data.h: Likewise.
27412 * config/arm/arm-tables.opt: Likewise.
27414 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27416 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27417 assembler when it is not -mfpu=auto.
27419 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27421 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27422 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27423 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27424 (ASM_CPU_SPEC): Rewrite.
27425 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27426 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27427 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27428 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27429 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27430 copied string is NUL-terminated. Also strip any characters prefixed
27432 (arm_rewrite_selected_arch): New function.
27433 (arm_rewrite_march): New function.
27435 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27437 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27438 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27439 (march, mcpu, mtune): Convert to string-based options.
27440 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27441 (arm_parse_arch_cpu_name): New function.
27442 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27443 identify selected architecture or CPU.
27444 (arm_option_save): New function.
27445 (TARGET_OPTION_SAVE): Redefine.
27446 (arm_option_restore): Restore string options.
27447 (arm_option_print): Print string options.
27449 2017-06-16 Martin Sebor <msebor@redhat.com>
27451 PR tree-optimization/80933
27452 PR tree-optimization/80934
27453 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27454 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27455 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27456 (gimple_fold_builtin): Call them.
27458 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27460 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27461 as unlikely; update profile.
27463 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27465 * predict.c (force_edge_cold): Handle declaring edges impossible
27468 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27470 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27472 (try_unroll_loop_completely): Fix reporting.
27474 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27476 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27478 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27481 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27482 if given a non-constant argument for an intrinsic which requires a
27485 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27487 * profile.c (compare_freqs): New function.
27488 (branch_prob): Sort edge list.
27489 (find_spanning_tree): Assume that the list is priority sorted.
27491 2017-06-16 Richard Biener <rguenther@suse.de>
27493 PR tree-optimization/81090
27494 * passes.def (pass_record_bounds): Remove.
27495 * tree-pass.h (make_pass_record_bounds): Likewise.
27496 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27497 make_pass_record_bounds): Likewise.
27498 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27499 not free niter estimates at the beginning but at the end.
27500 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27502 2017-06-16 Richard Biener <rguenther@suse.de>
27504 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27505 initializer to workaround ICE in host GCC 4.8.
27507 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27509 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27511 (clone_inlined_nodes): Update.
27513 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27515 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27516 prefetch settings, and enable prefetching by default at -O3.
27518 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27520 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27521 Set flag_prefetch_loop_arrays according to tuning data.
27523 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27525 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27526 New tune structure.
27527 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27528 [Unrelated to main purpose of the patch] Place the pointer field last
27529 to enable type checking errors when tune structure are wrongly merged.
27530 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27531 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27532 (thunderx2t99_prefetch_tune): New tune constants.
27533 (tune_params *_tunings): Update all tunings (no functional change).
27534 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27535 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27536 from tunings structures.
27538 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27541 * ubsan.c (instrument_null): Add T argument, use it instead
27542 of computing it based on IS_LHS.
27543 (instrument_object_size): Likewise.
27544 (pass_ubsan::execute): Adjust instrument_null and
27545 instrument_object_size callers to pass gimple_get_lhs or
27546 gimple_assign_rhs1 result to it. Use instrument_null instead of
27547 calling get_base_address and instrument_mem_ref. Handle
27548 aggregate call arguments for object-size sanitization.
27550 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27552 PR tree-optimization/81089
27553 * tree-vrp.c (is_masked_range_test): Validate operands of
27556 2017-06-15 Martin Sebor <msebor@redhat.com>
27559 * dumpfile.c (dump_register): Avoid calling memset to initialize
27560 a class with a default ctor.
27561 * gcc.c (struct compiler): Remove const qualification.
27562 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27563 * hash-table.h: Ditto.
27564 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27566 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27567 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27569 * params.h (struct param_info): Make struct members non-const.
27570 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27571 with default initialization.
27572 * vec.h (vec_copy_construct, vec_default_construct): New helper
27574 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27575 with vec_copy_construct.
27576 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27577 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27578 * doc/invoke.texi (-Wclass-memaccess): Document.
27580 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27582 * emit-rtl.h (is_leaf): Update comment about local
27583 register allocator.
27585 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27588 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27589 for a variable to have a section before checking if the section has a
27591 Set section to.persistent if persistent attribute is set.
27592 Warn if .persistent attribute is used on an automatic variable.
27594 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27596 PR rtl-optimization/80474
27597 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27599 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27601 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27604 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27606 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27607 (casesi): Emit bounds checking as RTL.
27608 (casesi_internal_mips16_<mode>): Remove bounds checking.
27610 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27612 * config/xtensa/xtensa.c (xtensa_option_override): Append
27613 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27614 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27615 xtensa_doloop_hooks): Define unconditionally.
27616 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27618 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27619 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27620 for it in xtensa_option_override.
27621 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27622 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27624 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27626 * doc/cppopts.texi: Document '-' special value to -MF.
27628 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27630 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27631 (cortex_a53_fconst): Likewise.
27632 (cortex_a53_fpmul): Likewise.
27633 (cortex_a53_f_load_64): Likewise.
27634 (cortex_a53_f_load_many): Likewise.
27635 (cortex_a53_advsimd_alu): Likewise.
27636 (cortex_a53_advsimd_alu_q): Likewise.
27637 (cortex_a53_advsimd_mul): Likewise.
27638 (cortex_a53_advsimd_mul_q): Likewise.
27639 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27640 Add missing fmul, r2f_cvt and fconst cases.
27642 2017-06-14 Richard Biener <rguenther@suse.de>
27644 PR middle-end/81088
27645 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27647 (fold_binary_loc): When associating do not treat pre-existing
27648 TREE_OVERFLOW on literal constants as a reason to allow
27649 TREE_OVERFLOW on associated literal constants.
27651 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27653 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27654 (MASK_FEATURES): New macro.
27655 * config/sparc/sparc.c (sparc_option_override): Remove the special
27656 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27658 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27660 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27661 a division of 0 if non-call exceptions are enabled.
27663 2017-06-14 Andrew Pinski <apinski@cavium.com>
27664 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27667 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27668 Improve vector initialization code gen for only variable case.
27670 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27672 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27674 2017-06-14 Richard Biener <rguenther@suse.de>
27676 PR tree-optimization/81083
27677 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27680 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27682 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27683 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27684 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27685 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27686 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27687 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27689 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27691 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27692 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27694 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27696 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27698 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27700 * config/rs6000/t-rtems: Don't handle SPE.
27702 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27704 * config/rs6000/t-linux: Don't handle SPE.
27706 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27708 * config/rs6000/eabispe.h: Delete file.
27710 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27712 * config/rs6000/t-spe: Delete file.
27714 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27716 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27717 (rs6000_legitimate_offset_address_p): Return false for anything in
27718 V2SImode or V2SFmode.
27720 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27722 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27723 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27724 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27726 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27727 (rs6000_legitimate_offset_address_p): Ditto.
27728 (rs6000_emit_move): Ditto.
27729 (rs6000_init_builtins): Remove V4HI_type_node.
27731 2017-06-13 Martin Liska <mliska@suse.cz>
27734 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27735 (gate_asan): Likewise.
27736 * asan.h (asan_no_sanitize_address_p): Remove the function.
27737 (sanitize_flags_p): New function.
27738 * builtins.def: Fix coding style.
27739 * common.opt: Use renamed enum value.
27740 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27741 * doc/extend.texi: Document no_sanitize attribute.
27742 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27743 to SANITIZE_UNDEFINED_NONDEFAULT.
27744 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27745 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27746 Use sanitize_flags_p.
27747 * gimplify.c (gimplify_function_tree): Likewise.
27748 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27749 * opts.c (parse_no_sanitize_attribute): New function.
27750 (common_handle_option): Use renamed enum value.
27751 * opts.h (parse_no_sanitize_attribute): Declare.
27752 * tree.c (sanitize_flags_p): New function.
27753 * tree.h: Declared here.
27754 * tsan.c: Use sanitize_flags_p.
27755 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27756 (instrument_mem_ref): Likewise.
27757 (instrument_bool_enum_load): Likewise.
27758 (do_ubsan_in_current_function): Remove the function.
27759 (pass_ubsan::execute): Use sanitize_flags_p.
27760 * ubsan.h: Remove do_ubsan_in_current_function
27761 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27762 (dump_function_to_file): Use it here.
27764 2017-06-13 Martin Jambor <mjambor@suse.cz>
27766 PR tree-optimization/80803
27767 PR tree-optimization/81063
27768 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27769 (propagate_subaccesses_across_link): Enqueue subtree whenever
27770 necessary instead of relying on the caller.
27772 2017-06-13 Martin Jambor <mjambor@suse.cz>
27774 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27775 that have a first_link.
27776 (sort_and_splice_var_accesses): Do not check first_link before
27778 (subtree_mark_written_and_enqueue): Likewise.
27779 (propagate_all_subaccesses): Likewise and do not stop at first
27780 parent with a first_link.
27782 2017-06-13 Martin Jambor <mjambor@suse.cz>
27784 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27787 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27789 * match.pd: New pattern.
27791 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27793 * tree-vrp.c (is_masked_range_test): New function.
27794 (register_edge_assert_for): Determine ranges for
27797 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27799 PR tree-optimization/67328
27800 * fold-const.c (maskable_range_p): New function.
27801 (build_range_check): Generate bittests if possible.
27803 2017-06-13 Martin Liska <mliska@suse.cz>
27805 * gimple-pretty-print.c (dump_probability): Add new argument.
27806 (dump_edge_probability): Dump both probability and count.
27807 (dump_gimple_label): Likewise.
27808 (dump_gimple_bb_header): Likewise.
27810 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27813 * config/avr/avr-devices.c: Fix indentation.
27814 * config/avr/gen-avr-mmcu-specs.c: Dito.
27816 2017-06-13 Richard Biener <rguenther@suse.de>
27818 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27819 instead get vector type from stmt_info.
27820 (vectorizable_reduction): Adjust. Remove dead code.
27822 2017-06-13 Richard Biener <rguenther@suse.de>
27824 PR middle-end/81065
27825 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27826 case of C * (x * C2 + C3).
27827 (fold_addr_of_array_ref_difference): Properly fold index difference.
27829 2017-06-12 David S. Miller <davem@davemloft.net>
27832 * config/sparc/sparc.md (return expander): Emit frame blockage if
27833 function uses alloca.
27835 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27837 * combine.c (make_field_assignment): Check len rather than the mode
27838 precision when calling force_to_mode.
27840 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27842 Support multilibs and devices that see flash in RAM address range.
27845 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27846 (avr_mcu_t) <flash_pm_offset>: New field.
27847 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27848 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27849 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27850 (AVR_TINY_PM_OFFSET): Remove macro.
27851 * config/avr/avr.opt (-mshort-calls): New option.
27852 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27853 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27854 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27855 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27856 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27857 instead of avr_arch->have_jmp_call.
27858 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27859 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27860 avr_arch->flash_pm_offset to define.
27861 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27862 new field flash_pm_offset. Add entry for avrxmega3.
27863 (avr_texinfo): Add entry for avrxmega3.
27864 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27865 attiny212, attiny214,
27866 attiny412, attiny414, attiny416, attiny417,
27867 attiny814, attiny816, attiny817,
27868 attiny1614, attiny1616, attiny1617,
27869 attiny3214, attiny3216, attiny3217.
27870 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27871 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27872 (avr_print_operand_address) [AVR_TINY]: Same.
27873 (avr_asm_init_sections) <readonly_data_section>: Only patch
27874 callback if avr_arch->flash_pm_offset = 0.
27875 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27876 for rodata if avr_arch->flash_pm_offset != 0.
27877 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27878 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27879 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27880 (m_options): Append opt_rcall.
27881 (m_dirnames): Append dir_rcall.
27882 * config/avr/t-multilib: Regenerate.
27884 * configure.ac [target=avr]: Check whether avrxmega3 default
27885 linker description file works as needed.
27886 * configure: Regenerate.
27887 * doc/avr-mmcu.texi: Regenerate.
27888 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27889 <__AVR_ARCH__>: Document avrxmega3 and 103.
27890 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27891 <__AVR_SHORT_CALLS__>: Document it.
27892 <__AVR_PM_BASE_ADDRESS__>: Document it.
27893 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27894 (AVR Variable Attributes) <progmem>: Document this is
27895 not needed for avrxmega3.
27896 (AVR Named Address Spaces) <__flash>: Dito.
27898 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27900 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27902 2017-06-12 Doug Rupp <rupp@adacore.com>
27904 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27905 Append vxworks-stdint.h to the tm_file list.
27906 * config/vxworks-stdint.h: New file.
27908 2017-06-12 Martin Liska <mliska@suse.cz>
27910 PR tree-optimization/81041
27911 * tree-profile.c (gimple_gen_ic_func_profiler):
27912 Create an extra BB in profile-generate
27913 (gimple_gen_time_profiler): Likewise.
27915 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27917 PR tree-optimization/81003
27918 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27919 (update_range_test): Use it instead of force_gimple_operand_gsi.
27921 2017-06-12 Richard Biener <rguenther@suse.de>
27923 PR tree-optimization/81053
27924 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27925 with backedge value not defined in loop. Simplify def stmt
27928 2017-06-11 Tom de Vries <tom@codesourcery.com>
27931 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27933 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27934 nvptx_cannot_force_const_mem.
27936 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27938 * opts.c (finish_options): Move test for flag_split_stack after
27939 it has been initialized.
27941 2017-06-11 Jason Merrill <jason@redhat.com>
27943 * tree.h (id_equal): New.
27944 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27945 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27946 instead of strcmp of IDENTIFIER_POINTER.
27948 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27950 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27951 (mark_all_inlined_calls_cdtor): Fix formating.
27952 (inline_transform): Rescale profile before inlining.
27954 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27956 * cgraph.h (cgraph_edge::clone): Update prototype.
27957 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27958 (cgraph_node::create_clone): Update.
27959 (cgraph_node::create_version_clone): Update.
27960 * tree-inline.c (copy_bb): Update.
27961 (expand_call_inline): Update.
27963 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27965 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27966 factored out from ...
27967 (rs6000_emit_prologue): ... here.
27969 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27971 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27972 factored out from ...
27973 (rs6000_emit_prologue): ... here.
27975 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27977 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27979 (handle_missing_profiles): Fix computation of tp_first_run.
27980 (counts_to_freqs): Do not touch freqs when count is 0.
27982 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27984 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27987 2017-06-10 Tom de Vries <tom@codesourcery.com>
27989 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27990 attributes): Document signal effective target.
27992 2017-06-10 Tom de Vries <tom@codesourcery.com>
27994 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27995 Document effective target stack_size.
27997 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27999 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28000 to the edit_context if they can be auto-applied.
28002 2017-06-9 Ian Lance Taylor <iant@golang.org>
28004 * opts.c (finish_options): If -fsplit-stack, disable implicit
28005 -forder-blocks-and-partition.
28006 * doc/invoke.texi (Optimize Options): Document that when using
28007 -fsplit-stack -forder-blocks-and-partition is not implicitly
28010 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28012 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28013 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28014 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28015 * builtins.def (abort, trap, unreachable): Declare cold.
28016 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28017 * tree-core.h (ECF_COLD): New.
28018 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28019 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28021 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28023 * predict.c (unlikely_executed_stmt_p): Cleanup.
28025 2017-06-09 Richard Biener <rguenther@suse.de>
28027 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28028 model if the ref is always written to.
28030 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28032 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28034 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28036 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28039 2017-06-09 Tom de Vries <tom@codesourcery.com>
28042 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28043 "target cannot support label values" when encountering LABEL_REF.
28045 2017-06-09 Martin Liska <mliska@suse.cz>
28047 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28048 (gimple_gen_ic_func_profiler): Emit direct comparison
28049 of __gcov_indirect_call_callee with NULL.
28050 (gimple_gen_time_profiler): Change probability from
28051 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28053 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28055 * profile.c (edge_gcov_counts): Turn to pointer.
28056 (compute_branch_probabilities, compute_branch_probabilities): Update.
28057 (branch_prob): Do not clear edge_gcov_count.
28058 * profile.h (edge_gcov_counts): Turn to pointer.
28059 (edge_gcov_count): Update.
28061 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28063 * gimple.h (gimple_check_failed): Mark cold.
28065 2017-06-09 Richard Biener <rguenther@suse.de>
28067 PR tree-optimization/66623
28068 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28069 refactor check_reduction into two parts, properly computing
28070 whether we have to check reduction validity for outer loop
28073 2017-06-09 Richard Biener <rguenther@suse.de>
28075 PR tree-optimization/79483
28076 * graphite-scop-detection.c (order): New global.
28077 (get_order): Compute bb to order mapping that satisfies code
28078 generation constraints.
28079 (cmp_pbbs): New helper.
28080 (build_scops): Start domwalk at entry block, sort generated
28083 2017-06-09 Richard Biener <rguenther@suse.de>
28085 PR middle-end/81007
28086 * ipa-polymorphic-call.c
28087 (ipa_polymorphic_call_context::restrict_to_inner_class):
28088 Skip FIELD_DECLs with error_mark_node type.
28089 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28092 2017-06-09 Martin Liska <mliska@suse.cz>
28094 * predict.c (struct branch_predictor): New struct.
28095 (test_prediction_value_range): New test.
28096 (predict_c_tests): New function.
28097 * selftest-run-tests.c (selftest::run_tests): Run the function.
28098 * selftest.h: Declare new tests.
28100 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28103 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28104 gen_add3_insn did not fail.
28105 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28106 r0, construct that number in a temporary reg and add that reg to r0.
28107 If asked to put the result in r0 as well, fail.
28109 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28111 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28112 for early expansion of vec_eqv.
28114 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28116 PR middle-end/81005
28117 * ubsan.c (instrument_null): Avoid pointless code temporary.
28118 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28120 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28124 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28127 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28130 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28132 * predict.c (unlikely_executed_edge_p): Move ahead.
28133 (probably_never_executed_edge_p): Use it.
28135 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28137 PR middle-end/79988
28138 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28139 gimple_call_builtin_p call.
28141 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28143 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28144 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28145 rtl_check_failed_type2, rtl_check_failed_code1,
28146 rtl_check_failed_code2, rtl_check_failed_code_mode,
28147 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28148 rtvec_check_failed_bounds, rtl_check_failed_flag,
28149 _fatal_insn_not_found, _fatal_insn): Likewise.
28150 * tree.h (tree_contains_struct_check_failed,
28151 tree_check_failed, tree_not_check_failed,
28152 tree_class_check_failed, tree_range_check_failed,
28153 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28154 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28155 tree_operand_check_failed, omp_clause_check_failed,
28156 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28159 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28161 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28162 flag_branch_probabilities.
28163 * ipa-inline.c (edge_badness): Likewise.
28164 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28165 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28166 * predict.c (maybe_hot_frequency_p): Likewise.
28167 (probably_never_executed): Likewise.
28168 * sched-ebb.c (schedule_ebbs): Likewise.
28169 * sched-rgn.c (find_single_block_region): Likewise.
28170 * tracer.c (tail_duplicate): Likewise.
28172 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28174 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28175 longer requires x_flag_profile_use.
28177 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28179 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28180 instead of flag_reorder_blocks_and_partition.
28181 * dbxout.c (dbxout_function_end): Likewise.
28182 * dwarf2out.c (gen_subprogram_die): Likewise.
28183 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28184 * hw-doloop.c (reorg_loops): Likewise.
28185 * varasm.c (assemble_start_function,
28186 assemble_end_function): Likewise.
28187 (decide_function_section): Do not check for
28188 flag_reorder_blocks_and_partition.
28190 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28192 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28194 (chkp_get_hard_register_fake_addr_expr): Ditto.
28195 (chkp_build_addr_expr): Add check for hard reg case.
28196 (chkp_parse_array_and_component_ref): Ditto.
28197 (chkp_find_bounds_1): Ditto.
28198 (chkp_process_stmt): Don't generate bounds store for
28201 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28203 * predict.c (maybe_hot_bb_p): Do not check profile status.
28204 (maybe_hot_edge_p): Likewise.
28205 (probably_never_executed): Check for zero counts even if profile
28207 (unlikely_executed_edge_p): New function.
28208 (unlikely_executed_stmt_p): New function.
28209 (unlikely_executed_bb_p): New function.
28210 (set_even_probabilities): Use unlikely predicates.
28211 (combine_predictions_for_bb): Likewise.
28212 (predict_paths_for_bb): Likewise.
28213 (predict_paths_leading_to_edge): Likewise.
28214 (determine_unlikely_bbs): New function.
28215 (estimate_bb_frequencies): Use it.
28216 (compute_function_frequency): Use zero counts even if profile is
28218 * profile-count.h: Fix typo.
28220 2017-08-08 Julia Koval <julia.koval@intel.com>
28222 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28223 _mm512_mask_cvtsepi16_storeu_epi8,
28224 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28225 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28226 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28227 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28228 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28229 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28230 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28231 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28232 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28233 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28234 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28235 __builtin_ia32_pmovuswb256mem_mask,
28236 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28237 __builtin_ia32_pmovwb512mem_mask): New builtins.
28239 2017-08-08 Julia Koval <julia.koval@intel.com>
28241 PR target/73350,80862
28242 * config/i386/subst.md (round): Fix round pattern.
28243 * config/i386/i386.c (ix86_erase_embedded_rounding):
28244 Fix erasing rounding for the fixed pattern.
28246 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28248 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28250 2017-06-08 Martin Liska <mliska@suse.cz>
28252 PR gcov-profile/80911
28253 * gcov.c (block_info::block_info): New constructor.
28255 2017-06-07 Carl Love <cel@us.ibm.com>
28257 * config/rs6000/rs6000-c: The return type of the following
28258 built-in functions was implemented as int not long long. Fix sign
28259 of return value for the unsigned version of vec_mulo and vec_mule.
28260 vector unsigned long long vec_bperm (vector unsigned long long,
28261 vector unsigned char)
28262 vector signed long long vec_mule (vector signed int,
28264 vector unsigned long long vec_mule (vector unsigned int,
28265 vector unsigned int)
28266 vector signed long long vec_mulo (vector signed int,
28268 vector unsigned long long vec_mulo (vector unsigned int,
28269 vector unsigned int)
28270 * doc/extend.texi: Fix the documentation for the built-in
28273 2017-06-07 Carl Love <cel@us.ibm.com>
28276 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28279 2017-06-07 Carl Love <cel@us.ibm.com>
28281 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28282 support, Generate doublehv for signed int/float for BE case only.
28284 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28286 * doc/invoke.texi (mcx16): Rewrite.
28288 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28290 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28291 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28292 *mov<mode>_softfloat, and an anonymous splitter): Use
28293 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28295 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28297 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28299 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28300 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28301 (rs6000_debug_reg_global): Adjust.
28302 (rs6000_init_hard_regno_mode_ok): Adjust.
28303 (rs6000_dbx_register_number): Adjust.
28304 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28305 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28306 Remove SPE_ACC and SPEFSCR.
28307 (REG_ALLOC_ORDER): Ditto.
28308 (FRAME_POINTER_REGNUM): Change to 111.
28309 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28310 (REG_CLASS_NAMES): Ditto.
28311 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28312 (REGISTER_NAMES): Ditto.
28313 (ADDITIONAL_REG_NAMES): Ditto.
28314 (rs6000_reg_names): Ditto.
28315 * config/rs6000/rs6000.md: Renumber some register number
28318 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28320 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28322 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28323 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28325 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28326 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28327 Delete the SPE high registers.
28328 (REG_ALLOC_ORDER): Ditto.
28329 (enum reg_class): Remove SPE_HIGH_REGS.
28330 (REG_CLASS_NAMES): Ditto.
28331 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28332 (REGISTER_NAMES): Ditto.
28333 (rs6000_reg_names): Ditto.
28334 * doc/tm.texi.in: Remove SPE as example.
28335 * doc/tm.texi: Regenerate.
28337 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28339 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28340 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28341 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28342 * config/rs6000/rs6000.md (type): Remove "brinc".
28344 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28346 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28347 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28348 * config/rs6000/linuxspe.h: Delete file.
28349 * config/rs6000/rs6000.md: Don't include spe.md.
28350 * config/rs6000/spe.h: Delete file.
28351 * config/rs6000/spe.md: Delete file.
28352 * config/rs6000/t-rs6000: Remove spe.md.
28354 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28356 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28357 (reg_or_none500mem_operand): Delete.
28358 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28359 instead of reg_or_none500mem_operand.
28361 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28363 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28364 handling of SPE flags.
28365 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28367 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28369 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28371 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28372 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28373 paired_divv2sf3): Similar.
28374 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28375 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28376 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28378 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28379 Rename the paired_* instruction patterns.
28380 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28382 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28383 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28384 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28385 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28386 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28387 PAIRED_VECTOR_MODE.
28388 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28389 (spe_func_has_64bit_regs_p): Delete.
28390 (spe_expand_predicate_builtin): Delete.
28391 (spe_expand_evsel_builtin): Delete.
28392 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28393 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28394 (invalid_e500_subreg): Delete.
28395 (rs6000_legitimize_address): Always force_reg op2 as well, for
28396 paired single memory accesses.
28397 (rs6000_member_type_forces_blk): Delete.
28398 (rs6000_spe_function_arg): Delete.
28399 (rs6000_expand_unop_builtin): Delete SPE handling.
28400 (rs6000_expand_binop_builtin): Ditto.
28401 (spe_expand_stv_builtin): Delete.
28402 (bdesc_2arg_spe): Delete.
28403 (spe_expand_builtin): Delete.
28404 (spe_expand_predicate_builtin): Delete.
28405 (spe_expand_evsel_builtin): Delete.
28406 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28407 (spe_init_builtins): Delete.
28408 (spe_func_has_64bit_regs_p): Delete.
28409 (savres_routine_name): Delete "info" parameter. Adjust callers.
28410 (rs6000_emit_stack_reset): Ditto.
28411 (rs6000_dwarf_register_span): Delete.
28412 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28413 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28414 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28416 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28418 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28419 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28420 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28421 mulv2sf3, divv2sf3): Delete expanders.
28423 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28425 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28427 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28429 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28430 * config/rs6000/rs6000.c: Ditto.
28432 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28434 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28435 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28436 comparison_operator.
28438 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28440 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28441 * config/rs6000/rs6000.opt: Ditto.
28442 * config/rs6000/t-rtems: Ditto.
28444 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28446 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28447 TARGET_E500_SINGLE by 0, simplify.
28448 * config/rs6000/rs6000.c: Ditto.
28449 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28450 (spe_build_register_parallel): Delete.
28451 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28452 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28453 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28454 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28455 (E500_CONVERT): Delete.
28456 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28458 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28460 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28461 * config/rs6000/dfp.md: Ditto.
28462 (negdd2, *negdd2_fpr): Merge.
28463 (absdd2, *absdd2_fpr): Merge.
28464 (negtd2, *negtd2_fpr): Merge.
28465 (abstd2, *abstd2_fpr): Merge.
28466 * config/rs6000/e500.h: Delete file.
28467 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28468 TARGET_FPRS by 1 and simplify.
28469 * config/rs6000/rs6000-c.c: Ditto.
28470 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28471 TARGET_DF_SPE by 0.
28472 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28474 * config/rs6000/rs6000.md: Ditto.
28475 (floatdidf2, *floatdidf2_fpr): Merge.
28476 (move_from_CR_gt_bit): Delete.
28477 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28478 (E500_CR_IOR_COMPARE): Delete.
28479 (All patterns that require !TARGET_FPRS): Delete.
28480 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28482 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28484 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28486 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28488 * graphds.c (add_edge): Intitialize edge's attached data.
28489 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28490 pointer parameter. Call pointed function on each edge during
28491 graph traversing. Skip traversing the edge when the function
28493 (graphds_dfs, graphds_scc): Ditto.
28494 (for_each_edge): New parameter. Pass the new parameter to callback
28496 * graphds.h (skip_edge_callback): New function pointer type.
28497 (graphds_dfs, graphds_scc): New function pointer parameter.
28498 (graphds_edge_callback, for_each_edge): New parameter.
28500 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28502 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28503 out code checking if runtime alias check is possible to below ...
28504 Call the new function.
28505 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28506 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28508 2017-06-07 Marek Polacek <polacek@redhat.com>
28511 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28512 TYPE_OVERFLOW_WRAPS check.
28514 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28516 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28517 if versioning is required.
28518 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28519 peeling with the check for versioning.
28521 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28523 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28524 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28525 Set true to new parameter if new ssa variable is defined.
28526 (vect_gen_vector_loop_niters): Refactor. Set range information
28527 for the new vector loop bound variable.
28528 (vect_do_peeling): Ditto.
28530 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28532 * tree-affine.c (ssa.h): Include header file.
28533 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28534 has wrapping overflow behavior.
28536 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28538 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28540 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28542 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28543 (tree_to_aff_combination): ... here.
28545 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28547 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28548 reg_pressure model function.
28549 (ivopts_global_cost_for_size): Delete.
28550 (determine_set_costs, iv_ca_recount_cost): Call new model function
28551 ivopts_estimate_reg_pressure.
28553 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28555 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28556 expensive than udiv. Remove floating point cases from mod.
28558 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28560 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28561 Increase idiv cost.
28563 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28565 * config/aarch64/aarch64.md
28566 (copysignsf3): Fix mask generation.
28568 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28570 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28572 (class dump_manager): Add register_dumps method.
28573 * dumpfile.c: Include langhooks.h.
28574 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28575 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28576 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28577 (dump_manager::dump_register): Start with 512 entries instead of 32.
28578 (dump_manager::register_dumps): New method.
28579 * toplev.c (general_init): Instead of invoking register_dumps
28580 langhook, invoke register_dumps method on the dump manager.
28581 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28584 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28586 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28587 Say that # requires an associated define_split to exist, and that
28588 the define_split must be suitable for use after register allocation.
28590 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28592 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28593 (compute_outgoing_frequencies): Also initialize zero counts.
28594 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28595 around loops; preserve more of profile when nothing changes.
28597 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28599 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28601 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28602 * config/arm/arm-cpu-cdata.h: Regenerate.
28603 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28604 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28605 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28606 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28607 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28609 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28611 * config/arm/t-rmprofile: Likewise.
28612 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28614 2017-06-06 David S. Miller <davem@davemloft.net>
28617 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28618 blockage if function uses alloca.
28620 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28622 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28623 New "uid" fields to hold pretty-print IDs of group and ref.
28624 Memory references are now identified as <group_id>:<ref_id>
28625 instead of using [random] addresses.
28626 (dump_mem_details): Simplify, no functional change.
28627 (dump_mem_ref): Simplify and make output more concise.
28628 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28629 (find_or_create_group): Initialize group uid.
28630 (record_ref): Initialize ref uid. Improve debug output.
28631 (prune_group_by_reuse, should_issue_prefetch_p,)
28632 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28633 (mark_nontemporal_store, determine_loop_nest_reuse):
28634 Improve debug output.
28636 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28638 * dbgcnt.def (prefetch): New debug counter.
28639 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28640 (schedule_prefetches): Stop issueing prefetches if debug counter
28643 2017-06-06 Tom de Vries <tom@codesourcery.com>
28645 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28646 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28648 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28650 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28651 Use aarch64_reg_or_zero predicate for operand 4.
28652 (aarch64_compare_and_swap<mode> define_insn_and_split):
28653 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28654 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28656 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28658 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28659 (arm_compute_save_core_reg_mask): This.
28660 (thumb1_compute_save_reg_mask): Rename into ...
28661 (thumb1_compute_save_core_reg_mask): This.
28662 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28663 (arm_compute_frame_layout): Likewise.
28665 2017-06-06 Richard Biener <rguenther@suse.de>
28667 PR tree-optimization/80974
28668 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28669 keep or clear leaders SSA info.
28671 2017-06-06 Tom de Vries <tom@codesourcery.com>
28673 * config/nvptx/nvptx.c (split_mode_p): New function.
28674 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28676 2017-06-06 Tom de Vries <tom@codesourcery.com>
28678 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28680 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28683 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28686 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28688 * shrink-wrap.c (handle_simple_exit): Update profile.
28689 (try_shrink_wrapping): Upate profile.
28691 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28693 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28694 (tree_guess_outgoing_edge_probabilities): New.
28695 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28696 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28698 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28700 * ipa-split.c (split_function): Initialize return bb profile.
28702 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28704 * profile.c (compute_branch_probabilities): Also initialize
28705 EXIT_BLOCK profile.
28707 2017-06-06 Richard Biener <rguenther@suse.de>
28709 PR tree-optimization/80928
28710 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28711 (vect_analyze_loop_operations): Properly guard analysis for
28713 (vect_transform_loop): Likewise.
28714 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28715 (vect_model_induction_cost): Do not cost for pure SLP.
28716 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28717 of induction in inner loop vectorization.
28718 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28719 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28720 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28722 (vect_analyze_slp_cost_1): Cost induction.
28723 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28724 (vect_get_slp_vect_defs): Likewise.
28725 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28726 (vect_transform_stmt): Handle SLP reductions.
28727 * tree-vectorizer.h (vectorizable_induction): Adjust.
28729 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28731 * config/rs6000/rs6000.c (make_resolver_func): Update
28732 init_lowered_empty_function call.
28734 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28736 * doc/invoke.texi: Document the -fprofile-abs-path option.
28737 * common.opt (fprofile-abs-path): New option.
28738 * gcov-io.h (gcov_write_filename): Declare.
28739 * gcov-io.c (gcov_write_filename): New function.
28740 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28741 * profile.c (output_location): Likewise.
28743 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28745 * shring-wrap.c: Revert accidental commit.
28747 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28749 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28751 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28753 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28755 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28756 profile in callgraph edge.
28757 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28758 (apply_scale): Likewise.
28759 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28760 Also scale profile when inlining function with zero profile.
28761 (initialize_cfun): Update exit block profile even when it is zero.
28762 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28763 when profile is read.
28765 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28767 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28768 (CLONE_*): New constants to define the processors we can generate
28769 code for with the target_clone attribute.
28770 (rs6000_clone_map): New array to identify which clone processors
28771 the current program is running on.
28772 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28773 target_clone attribute.
28774 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28775 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28776 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28777 (cpu_expand_builtin): Add support for target_clone attribute.
28778 (rs6000_valid_attribute_p): Allow "default" attribute.
28779 (get_decl_name): New debug function to simplify printing the
28780 current function name in debugging statements.
28781 (rs6000_clone_priority): New functions to support the target_clone
28782 attribute, and be able to generate code to switch between ISA 2.05
28783 through ISA 3.0 (power6 through power9).
28784 (rs6000_compare_version_priority): Likewise.
28785 (rs6000_get_function_versions_dispatcher): Likewise.
28786 (make_resolver_func): Likewise.
28787 (add_condition_to_bb): Likewise.
28788 (dispatch_function_versions): Likewise.
28789 (rs6000_generate_version_dispatcher_body): Likewise.
28790 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28791 (fusion_gpr_load_p): Fix a spacing issue.
28792 * doc/extend.texi (Common Function Attributes): Document that the
28793 PowerPC supports the target_clone attribute.
28795 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28797 * config/arm/arm.h: explain F symbol found in description of ARM
28798 register allocation in its legend.
28800 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28802 * config/mips/frame-header-opt.c: Include profile-count.h.
28803 * config/riscv/riscv.c: Include profile-count.h
28805 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28807 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28809 (sm_set_flag_if_changed): Add bbs field.
28810 (execute_sm_if_changed_flag_set): Pass BBS.
28811 (execute_sm): Update.
28813 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28815 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28818 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28820 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28821 (peephole2): New peephole2 to emit the above.
28822 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28824 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28826 * config/aarch64/aarch64.c (define_peephole2 above
28827 *sub_<shift>_<mode>): New peephole.
28829 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28831 * config/i386/i386.c (make_resolver_func): Update.
28832 * Makefile.in: Add profile-count.h and profile-count.o
28833 * auto-profile.c (afdo_indirect_call): Update to new API.
28834 (afdo_set_bb_count): Update.
28835 (afdo_propagate_edge): Update.
28836 (afdo_propagate_circuit): Update.
28837 (afdo_calculate_branch_prob): Update.
28838 (afdo_annotate_cfg): Update.
28839 * basic-block.h: Include profile-count.h
28840 (struct edge_def): Turn count to profile_count.
28841 (struct basic_block_def): Likewie.
28842 (REG_BR_PROB_BASE): Move to profile-count.h
28843 (RDIV): Move to profile-count.h
28844 * bb-reorder.c (max_entry_count): Turn to profile_count.
28845 (find_traces): Update.
28846 (rotate_loop):Update.
28847 (connect_traces):Update.
28848 (sanitize_hot_paths):Update.
28849 * bt-load.c (migrate_btr_defs): Update.
28850 * cfg.c (RDIV): Remove.
28851 (init_flow): Use alloc_block.
28852 (alloc_block): Uninitialize count.
28853 (unchecked_make_edge): Uninitialize count.
28854 (check_bb_profile): Update.
28855 (dump_edge_info): Update.
28856 (dump_bb_info): Update.
28857 (update_bb_profile_for_threading): Update.
28858 (scale_bbs_frequencies_int): Update.
28859 (scale_bbs_frequencies_gcov_type): Update.
28860 (scale_bbs_frequencies_profile_count): New.
28861 * cfg.h (update_bb_profile_for_threading): Update.
28862 (scale_bbs_frequencies_profile_count): Declare.
28863 * cfgbuild.c (compute_outgoing_frequencies): Update.
28864 (find_many_sub_basic_blocks): Update.
28865 * cfgcleanup.c (try_forward_edges): Update.
28866 (try_crossjump_to_edge): Update.
28867 * cfgexpand.c (expand_gimple_tailcall): Update.
28868 (construct_exit_block): Update.
28869 * cfghooks.c (verify_flow_info): Update.
28870 (dump_bb_for_graph): Update.
28871 (split_edge): Update.
28872 (make_forwarder_block): Update.
28873 (duplicate_block): Update.
28874 (account_profile_record): Update.
28875 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28876 (get_estimated_loop_iterations): Update.
28877 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28878 (single_likely_exit): Update.
28879 * cfgloopmanip.c (scale_loop_profile): Update.
28881 (set_zero_probability): Update.
28882 (lv_adjust_loop_entry_edge): Update.
28883 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28884 (purge_dead_edges): Update.
28885 (rtl_account_profile_record): Update.
28886 * cgraph.c (cgraph_node::create): Uninitialize count.
28887 (symbol_table::create_edge): Uninitialize count.
28888 (cgraph_update_edges_for_call_stmt_node): Update.
28889 (cgraph_edge::dump_edge_flags): Update.
28890 (cgraph_node::dump): Update.
28891 (cgraph_edge::maybe_hot_p): Update.
28892 * cgraph.h: Include profile-count.h
28893 (create_clone), create_edge, create_indirect_edge): Update.
28894 (cgraph_node): Turn count to profile_count.
28895 (cgraph_edge0: Likewise.
28896 (make_speculative, clone): Update.
28897 (create_edge): Update.
28898 (init_lowered_empty_function): Update.
28899 * cgraphclones.c (cgraph_edge::clone): Update.
28900 (duplicate_thunk_for_node): Update.
28901 (cgraph_node::create_clone): Update.
28902 * cgraphunit.c (cgraph_node::analyze): Update.
28903 (cgraph_node::expand_thunk): Update.
28904 * final.c (dump_basic_block_info): Update.
28905 * gimple-streamer-in.c (input_bb): Update.
28906 * gimple-streamer-out.c (output_bb): Update.
28907 * graphite.c (print_global_statistics): Update.
28908 (print_graphite_scop_statistics): Update.
28909 * hsa-brig.c: Include basic-block.h.
28910 * hsa-dump.c: Include basic-block.h.
28911 * hsa-gen.c (T sum_slice): Update.
28912 (convert_switch_statements):Update.
28913 * hsa-regalloc.c: Include basic-block.h.
28914 * ipa-chkp.c (chkp_produce_thunks): Update.
28915 * ipa-cp.c (struct caller_statistics): Update.
28916 (init_caller_stats): Update.
28917 (gather_caller_stats): Update.
28918 (ipcp_cloning_candidate_p): Update.
28919 (good_cloning_opportunity_p): Update.
28920 (get_info_about_necessary_edges): Update.
28921 (dump_profile_updates): Update.
28922 (update_profiling_info): Update.
28923 (update_specialized_profile): Update.
28924 (perhaps_add_new_callers): Update.
28925 (decide_about_value): Update.
28926 (ipa_cp_c_finalize): Update.
28927 * ipa-devirt.c (struct odr_type_warn_count): Update.
28928 (struct decl_warn_count): Update.
28929 (struct final_warning_record): Update.
28930 (possible_polymorphic_call_targets): Update.
28931 (ipa_devirt): Update.
28932 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28933 * ipa-icf.c (sem_function::merge): Update.
28934 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28935 * ipa-inline.c (compute_uninlined_call_time): Update.
28936 (compute_inlined_call_time): Update.
28937 (want_inline_small_function_p): Update.
28938 (want_inline_self_recursive_call_p): Update.
28939 (edge_badness): Update.
28940 (lookup_recursive_calls): Update.
28941 (recursive_inlining): Update.
28942 (inline_small_functions): Update.
28943 (dump_overall_stats): Update.
28944 (dump_inline_stats): Update.
28945 * ipa-profile.c (ipa_profile_generate_summary): Update.
28946 (ipa_propagate_frequency): Update.
28947 (ipa_profile): Update.
28948 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28949 * ipa-utils.c (ipa_merge_profiles): Update.
28950 * loop-doloop.c (doloop_modify): Update.
28951 * loop-unroll.c (report_unroll): Update.
28952 (unroll_loop_runtime_iterations): Update.
28953 * lto-cgraph.c (lto_output_edge): Update.
28954 (lto_output_node): Update.
28955 (input_node): Update.
28956 (input_edge): Update.
28957 (merge_profile_summaries): Update.
28958 * lto-streamer-in.c (input_cfg): Update.
28959 * lto-streamer-out.c (output_cfg): Update.
28960 * mcf.c (create_fixup_graph): Update.
28961 (adjust_cfg_counts): Update.
28962 (sum_edge_counts): Update.
28963 * modulo-sched.c (sms_schedule): Update.
28964 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28965 * predict.c (maybe_hot_count_p): Update.
28966 (probably_never_executed): Update.
28967 (dump_prediction): Update.
28968 (combine_predictions_for_bb): Update.
28969 (propagate_freq): Update.
28970 (handle_missing_profiles): Update.
28971 (counts_to_freqs): Update.
28972 (rebuild_frequencies): Update.
28973 (force_edge_cold): Update.
28974 * predict.h: Include profile-count.h
28975 (maybe_hot_count_p, counts_to_freqs): UPdate.
28976 * print-rtl-function.c: Do not include cfg.h
28977 * print-rtl.c: Include basic-block.h
28978 * profile-count.c: New file.
28979 * profile-count.h: New file.
28980 * profile.c (is_edge_inconsistent): Update.
28981 (correct_negative_edge_counts): Update.
28982 (is_inconsistent): Update.
28983 (set_bb_counts): Update.
28984 (read_profile_edge_counts): Update.
28985 (compute_frequency_overlap): Update.
28986 (compute_branch_probabilities): Update; Initialize and deinitialize
28988 (branch_prob): Update.
28989 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28990 (edge_gcov_count): New.
28991 (bb_gcov_count): New.
28992 * shrink-wrap.c (try_shrink_wrapping): Update.
28993 * tracer.c (better_p): Update.
28994 * trans-mem.c (expand_transaction): Update.
28995 (ipa_tm_insert_irr_call): Update.
28996 (ipa_tm_insert_gettmclone_call): Update.
28997 * tree-call-cdce.c: Update.
28998 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28999 (gimple_duplicate_sese_tail): Update.
29000 (gimple_account_profile_record): Update.
29001 (execute_fixup_cfg): Update.
29002 * tree-inline.c (copy_bb): Update.
29003 (copy_edges_for_bb): Update.
29004 (initialize_cfun): Update.
29005 (freqs_to_counts): Update.
29006 (copy_cfg_body): Update.
29007 (expand_call_inline): Update.
29008 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29009 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29010 (try_unroll_loop_completely): Update.
29011 (try_peel_loop): Update.
29012 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29013 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29014 * tree-ssa-loop-split.c (connect_loops): Update.
29015 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29016 * tree-ssa-reassoc.c (branch_fixup): Update.
29017 * tree-ssa-tail-merge.c (replace_block_by): Update.
29018 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29019 (compute_path_counts): Update.
29020 (update_profile): Update.
29021 (recompute_probabilities): Update.
29022 (update_joiner_offpath_counts): Update.
29023 (estimated_freqs_path): Update.
29024 (freqs_to_counts_path): Update.
29025 (clear_counts_path): Update.
29026 (ssa_fix_duplicate_block_edges): Update.
29027 (duplicate_thread_path): Update.
29028 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29029 (struct switch_conv_info): Update.
29030 * tree-tailcall.c (decrease_profile): Update.
29031 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29032 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29033 * value-prof.c (check_counter): Update.
29034 (gimple_divmod_fixed_value): Update.
29035 (gimple_mod_pow2): Update.
29036 (gimple_mod_subtract): Update.
29037 (gimple_ic_transform): Update.
29038 (gimple_stringop_fixed_value): Update.
29039 * value-prof.h (gimple_ic): Update.
29041 2017-06-02 Carl Love <cel@us.ibm.com>
29043 * config/rs6000/rs6000-c: Add support for built-in functions
29044 vector double vec_doublee (vector signed int);
29045 vector double vec_doublee (vector unsigned int);
29046 vector double vec_doublee (vector float);
29047 vector double vec_doubleh (vector signed int);
29048 vector double vec_doubleh (vector unsigned int);
29049 vector double vec_doubleh (vector float);
29050 vector double vec_doublel (vector signed int);
29051 vector double vec_doublel (vector unsigned int);
29052 vector double vec_doublel (vector float);
29053 vector double vec_doubleo (vector signed int);
29054 vector double vec_doubleo (vector unsigned int);
29055 vector double vec_doubleo (vector float);.
29056 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29057 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29059 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29060 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29061 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29063 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29064 vec_doublel, vec_doubleh.
29065 * doc/extend.texi: Update the built-in documentation file for the
29066 new built-in functions.
29068 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29071 * ipa-inline-analysis.c (free_growth_caches): Set
29072 edge_removal_hook_holder to NULL after removing it.
29074 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29076 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29077 comparision with zero.
29079 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29080 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29081 for early expansion of vec_min and vec_max builtins.
29082 (builtin_function_type): Add min/max unsigned variants to those
29083 identified as having unsigned arguments.
29085 2017-06-02 Olivier Hainque <hainque@adacore.com>
29087 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29089 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29091 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29092 Use VALL_F16 iterator rather than VALL.
29094 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29096 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29097 Emit CBNZ inside loop when doing a strong exchange and comparing
29098 against zero. Generate the CC flags after the loop.
29100 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29102 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29103 (dl_section_ref): New.
29104 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29105 On AIX, append an expression to subtract the size of the
29106 section length to dl_section_ref.
29108 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29110 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29111 for early expansion of vector absolute builtins.
29113 2017-06-02 Richard Biener <rguenther@suse.de>
29115 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29116 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29118 2017-06-02 Richard Biener <rguenther@suse.de>
29120 PR tree-optimization/80948
29121 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29122 stmt order as well.
29124 2017-06-02 Richard Biener <rguenther@suse.de>
29126 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29128 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29129 for inductions as relevant.
29131 2017-06-02 Richard Biener <rguenther@suse.de>
29133 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29134 (vectorizable_induction): ... this. Remove dead code.
29136 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29138 * builtins. (expand_builtin_alloca): Remove second parameter and
29139 infer its value from the first parameter instead.
29140 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29142 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29144 PR rtl-optimization/80903
29145 * loop-doloop.c (add_test): Unshare sequence.
29147 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29149 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29151 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29153 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29155 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29156 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29157 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29158 (xlogue_layout::get_stub_rtx): Make static.
29159 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29160 (xlogue_layout::compute_stub_managed_regs): Rename to...
29161 (xlogue_layout::count_stub_managed_regs): ...this.
29162 (xlogue_layout::is_stub_managed_reg): New function.
29163 (xlogue_layout::m_stub_names): Rename to...
29164 (xlogue_layout::s_stub_names): ...this, make static.
29165 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29166 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29167 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29168 xlogue_layout::s_stub_names): Instantiate statics.
29169 (stub_managed_regs): Remove.
29170 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29171 (disable_call_ms2sysv_xlogues): Rename to...
29172 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29173 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29175 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29176 change after reload_completed.
29177 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29179 (ix86_expand_prologue): Likewise.
29180 (ix86_expand_epilogue): Likewise.
29181 (ix86_expand_split_stack_prologue): Likewise.
29182 (ix86_compute_frame_layout): Remove frame parameter ...
29183 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29184 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29186 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29187 (ix86_frame): Move from here ...
29188 * config/i386/i386.h (ix86_frame): ... to here.
29189 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29190 complete ix86_frame data structure instead. Remove some_ld_name.
29192 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29194 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29195 symbols that hold a DECL_VALUE_EXPR.
29197 2017-06-01 Martin Jambor <mjambor@suse.cz>
29199 PR tree-optimization/80898
29200 * tree-sra.c (process_subtree_disqualification): Removed.
29201 (disqualify_candidate): Do not acll
29202 process_subtree_disqualification.
29203 (subtree_mark_written_and_enqueue): New function.
29204 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29205 RHS has been disqualified and re-queue LHS if necessary. Apart
29206 from that, ignore disqualified RHS.
29208 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29210 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29211 address fetch for z10 or later.
29213 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29215 * config/arc/arc.md (tst_movb): Add guard when splitting.
29217 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29219 * config/arc/arc.c (arc_can_eliminate): Test against
29220 arc_frame_pointer_needed.
29222 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29224 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29225 to prevent store reordering.
29226 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29227 (type): Add block type.
29228 (stack_tie): Define special instruction to be used in
29231 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29233 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29234 constraint. It is not valid for the pattern.
29235 (noncommutative_binary_comparison): Likewise.
29237 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29239 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29242 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29244 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29245 be used by the reg-alloc.
29247 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29249 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29250 reg-alloc when having mul64 or mul32x16 instructions.
29251 (mulsidi3): Likewise.
29252 (umulsidi3): Likewise.
29253 (mulsi32x16): New pattern.
29254 (mulsi64): Likewise.
29255 (mulsidi64): Likewise.
29256 (umulsidi64): Likewise.
29257 (MUL32x16_REG): Define.
29258 (mul64_600): Use MUL32x16_REG.
29259 (mac64_600): Likewise.
29260 (umul64_600): Likewise.
29261 (umac64_600): Likewise.
29263 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29265 * config/arc/arc.md (mulsi3_700): Make it commutative.
29267 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29269 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29271 (*sign_extendsidi2_insn): Likewise for movstosw.
29273 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29275 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29276 the type of the input discriminant value. Convert the
29277 discriminant value of signedness vary.
29279 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29281 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29282 Add to -Wall section.
29284 2017-06-01 Richard Biener <rguenther@suse.de>
29286 PR middle-end/66313
29287 * fold-const.c (fold_plusminus_mult_expr): If the factored
29288 factor may be zero use a wrapping type for the inner operation.
29289 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29290 and handle moved defs.
29291 (process_assignment): Properly guard the unary op case. Return a
29292 tri-state indicating that moving the stmt before the call may allow
29293 to continue. Pass through to_move.
29294 (find_tail_calls): Handle moving unrelated defs before
29297 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29300 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29301 splitter result in the canonical way.
29303 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29305 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29306 also for 32bit target. Update insn attributes.
29307 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29309 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29311 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29312 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29313 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29314 (_mm_maskz_min_ss): New intrinsics.
29316 2017-05-31 Martin Liska <mliska@suse.cz>
29318 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29319 Change comment style to one we normally use.
29320 (vectorizable_reduction): Likewise.
29321 (vectorizable_induction): Likewise.
29322 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29323 (vectorizable_call): Likewise.
29324 (vectorizable_simd_clone_call): Likewise.
29325 (vectorizable_conversion): Likewise.
29326 (vectorizable_assignment): Likewise.
29327 (vectorizable_shift): Likewise.
29328 (vectorizable_operation): Likewise.
29329 (vectorizable_store): Likewise.
29330 (vectorizable_load): Likewise.
29331 * tree-vectorizer.h: Likewise.
29333 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29335 * passes.c (emergency_dump_function): New.
29336 * tree-pass.h (emergency_dump_function): Declare.
29337 * plugin.c (plugins_internal_error_function): Remove.
29338 * plugin.h (plugins_internal_error_function): Remove declaration.
29339 * toplev.c (internal_error_function): New static function. Use it...
29340 (general_init): ...here.
29342 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29344 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29345 (arc_rtx_costs): Add costs for new patterns.
29346 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29347 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29349 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29351 * tree-ssa-strlen.c (get_next_strinfo): New function.
29352 (get_stridx_plus_constant): Use it.
29353 (zero_length_string): Likewise.
29354 (adjust_related_strinfos): Likewise.
29355 (adjust_last_stmt): Likewise.
29357 2017-05-31 Richard Biener <rguenther@suse.de>
29360 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29361 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29363 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29365 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29366 loop_vinfo argument and use of dependence distance vectors.
29367 Check instead whether the two references differ only in their
29368 initial value and assume that they have the same alignment if the
29369 difference is a multiple of the vector alignment.
29370 (vect_analyze_data_refs_alignment): Update call accordingly.
29372 2017-05-31 Martin Liska <mliska@suse.cz>
29375 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29377 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29379 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29380 (create_intersect_range_checks): Move from ...
29381 * tree-data-ref.c (create_intersect_range_checks_index)
29382 (create_intersect_range_checks): ... to here.
29383 (create_runtime_alias_checks): New function factored from ...
29384 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29385 here. Call above function.
29386 * tree-data-ref.h (create_runtime_alias_checks): New function.
29388 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29390 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29391 segment length for dr_b and compute it in wide_int.
29393 2017-05-31 Richard Biener <rguenther@suse.de>
29395 PR tree-optimization/80906
29396 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29397 and pass through iv_map.
29398 (copy_bb_and_scalar_dependences): Adjust.
29399 (translate_pending_phi_nodes): Likewise.
29400 (copy_loop_close_phi_args): Handle code-generating IVs instead
29403 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29405 * diagnostic-color.c (color_dict): Add "type-diff".
29406 (parse_gcc_colors): Update comment.
29407 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29408 -fdiagnostics-show-template-tree and -fno-elide-type.
29409 (GCC_COLORS): Add type-diff to example.
29411 (-fdiagnostics-show-template-tree): New.
29412 (-fno-elide-type): New.
29413 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29414 the pp_format_decoder callback. Call any m_format_postprocessor's
29416 (pretty_printer::pretty_printer): Initialize
29417 m_format_postprocessor.
29418 (pretty_printer::~pretty_printer): Delete any
29419 m_format_postprocessor.
29420 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29421 (class format_postprocessor): New class.
29422 (struct pretty_printer::format_decoder): Document the new parameters.
29423 (struct pretty_printer::m_format_postprocessor): New field.
29424 * tree-diagnostic.c (default_tree_printer): Update for new
29425 bool and const char ** params.
29426 * tree-diagnostic.h (default_tree_printer): Likewise.
29428 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29430 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29431 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29432 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29433 rs6000_gen_cell_microcode code.
29434 (rs6000_final_prescan_insn): Delete.
29435 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29436 "warn-cell-microcode" entries.
29437 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29438 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29439 throughout. Change cc_reg_not_micro_cr0_operand to
29440 cc_reg_not_cr0_operand throughout.
29441 (*extendhi<mode>2_noload): Delete.
29442 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29443 (mwarn-cell-microcode): Delete.
29444 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29445 -mgen-cell-microcode and -mwarn-cell-microcode.
29447 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29450 * config/i386/constraints.md (Yd): New constraint.
29452 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29453 and (?Yd, r) alternatives. Update insn attributes.
29454 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29455 and (?*Yd, r) alternatives. Update insn attributes.
29456 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29458 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29460 * gimplify.c (gimplify_modify_expr): Don't create a
29461 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29464 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29466 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29468 2017-05-30 Richard Biener <rguenther@suse.de>
29470 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29471 and reduc_def fields.
29472 (STMT_VINFO_REDUC_TYPE): New define.
29473 (STMT_VINFO_REDUC_DEF): Likewise.
29474 (vect_force_simple_reduction): Adjust prototype.
29475 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29476 (vect_is_simple_reduction): Remove check_reduction argument.
29477 (vect_force_simple_reduction): Adjust and set
29478 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29479 (vectorizable_reduction): Do not re-do reduction analysis
29480 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29481 * tree-parloops.c (gather_scalar_reductions): Adjust.
29483 2017-05-30 Richard Biener <rguenther@suse.de>
29485 PR middle-end/80901
29486 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29489 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29491 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29492 Introduce unknown_misalignment parameter and remove vf.
29493 (vect_peeling_hash_get_lowest_cost):
29494 Pass unknown_misalignment parameter.
29495 (vect_enhance_data_refs_alignment):
29496 Fix unsupportable data ref treatment.
29498 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29500 * tree-vect-data-refs.c (vect_get_data_access_cost):
29501 Workaround for SLP handling.
29502 (vect_enhance_data_refs_alignment):
29503 Compute costs for doing no peeling at all, compare to the best
29504 peeling costs so far and avoid peeling if cheaper.
29506 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29508 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29509 Return peeling info and set costs to zero for unlimited cost
29511 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29512 with unknown misalignment. Compute and costs for unknown
29513 misalignment, compare them to the costs for known misalignment
29514 and choose the cheapest for peeling.
29516 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29518 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29519 (vect_get_peeling_costs_all_drs): Create function.
29520 (vect_peeling_hash_get_lowest_cost):
29521 Use vect_get_peeling_costs_all_drs.
29522 (vect_peeling_supportable): Create function.
29523 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29525 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29527 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29528 DR_HAS_NEGATIVE_STEP.
29529 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29530 (vect_enhance_data_refs_alignment): Use.
29531 (vect_duplicate_ssa_name_ptr_info): Use.
29532 * tree-vectorizer.h (dr_misalignment): Use.
29533 (known_alignment_for_access_p): Use.
29535 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29538 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29540 (has_section_name): New function.
29542 2017-05-30 Martin Liska <mliska@suse.cz>
29545 * auto-profile.c (get_function_decl_from_block): Fix
29548 2017-05-30 Richard Biener <rguenther@suse.de>
29550 PR middle-end/80876
29551 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29553 2017-05-30 Martin Liska <mliska@suse.cz>
29555 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29556 * dumpfile.h (struct dump_file_info): Remove ctors.
29558 2017-05-30 Martin Liska <mliska@suse.cz>
29560 * predict.def: Fix GNU coding style.
29562 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29564 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29565 Mark 'to' argument with ATTRIBUTE_UNUSED.
29567 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29569 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29570 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29571 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29574 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29576 * doc/install.texi (Options specification): Restore entry of
29577 --enable-sjlj-exceptions.
29579 2017-05-27 Michael Eager <eager@eagercon.com>
29582 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29584 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29586 * config/microblaze/microblaze.h
29587 (FIXED_REGISTERS): Update in macro.
29588 (CALL_USED_REGISTERS): Update in macro.
29590 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29592 * doc/install.texi: Add links to macOS binary distributions.
29594 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29598 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29600 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29602 2017-05-26 Martin Liska <mliska@suse.cz>
29604 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29606 2017-05-26 Martin Liska <mliska@suse.cz>
29608 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29609 always leading ';; '.
29610 (dump_bb_info): Likewise.
29611 (brief_dump_cfg): Likewise.
29612 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29613 * dumpfile.c: Remove usage of TDF_VERBOSE.
29614 * dumpfile.h (enum dump_kind): Likewise.
29615 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29616 * print-tree.c (debug_verbose): Remove.
29617 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29618 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29619 * tree-diagnostic.c (default_tree_printer): Replace
29620 TDF_DIAGNOSTIC with TDF_SLIM.
29622 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29624 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29625 in parameter loop, rather than loop_vinfo.
29626 (create_intersect_range_checks): Ditto.
29627 (vect_create_cond_for_alias_checks): Update call to above functions.
29629 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29631 PR tree-optimization/80815
29632 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29633 for merging runtime alias checks. Handle negative DR_STEPs.
29635 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29637 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29639 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29640 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29641 out code pruning runtime alias checks.
29642 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29643 factored out from above.
29644 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29646 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29648 (prune_runtime_alias_test_list): New decalaration.
29650 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29652 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29653 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29654 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29655 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29656 (operator==, comp_dr_with_seg_len_pair): Ditto.
29657 (vect_prune_runtime_alias_test_list): Ditto.
29659 2017-05-26 Martin Liska <mliska@suse.cz>
29662 * params.def: Bound partial-inlining-entry-probability param.
29664 2017-05-26 Marek Polacek <polacek@redhat.com>
29667 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29670 2017-05-26 Richard Biener <rguenther@suse.de>
29672 PR tree-optimization/80842
29673 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29676 2017-05-26 Richard Biener <rguenther@suse.de>
29678 PR tree-optimization/80844
29679 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29681 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29683 * doc/md.texi (Machine Constraints): Update x86 family
29684 machine constraints section to match 'config/i386/constraints.md'.
29686 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29688 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29690 2017-05-25 Nathan Sidwell <nathan@acm.org>
29692 * doc/invoke.texi (--enable-languages): Update documentation.
29694 2017-05-25 Martin Liska <mliska@suse.cz>
29696 * dumpfile.c: Add TDF_FOLDING.
29697 * dumpfile.h (enum dump_kind): Likewise.
29698 * genmatch.c (dt_simplify::gen_1): Use it.
29700 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29702 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29704 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29706 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29707 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29709 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29711 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29712 * match.pd (X == C): Rewrite it here.
29713 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29714 with_certain_nonzero_bits2): New predicates.
29715 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29717 2017-05-24 Nathan Sidwell <nathan@acm.org>
29719 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29722 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29725 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29727 * config/powerpcspe: New port. Files are copied from the rs6000
29728 port, with "rs6000" in filenames replaced by "powerpcspe".
29730 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29732 PR rtl-optimization/80754
29733 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29735 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29737 * config/sparc/sparc.md (length): Return the correct value for -mflat
29738 sibcalls to match output_sibcall.
29740 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29744 * config/rs6000/rs6000.c (struct machine_function): Add new field
29746 (rs6000_get_separate_components): Init that field, use it.
29747 (rs6000_components_for_bb): Use the field.
29749 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29751 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29753 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29755 PR middle-end/80823
29756 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29758 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29761 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29762 address against address_operand predicate.
29763 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29765 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29767 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29768 parameters passed indirectly.
29770 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29772 * config/i386/i386.md (*movdi_internal): Remove SSE4
29773 alternative 18 (?r, *v). Update insn attributes.
29774 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29775 Update insn attributes.
29776 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29777 Update insn attributes.
29778 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29779 alternative 1 (r, v). Remove isa attribute.
29780 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29781 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29782 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29784 2017-05-23 Tom de Vries <tom@codesourcery.com>
29786 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29789 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29791 * cgraphunit.c (symbol_table::process_new_functions): Update.
29792 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29793 (inline_generate_summary): Rename to ...
29794 (ipa_fn_summary_generate): ... this one.
29795 (inline_read_summary): Rename to ...
29796 (ipa_fn_summary_read): ... this one.
29797 (inline_write_summary): Rename to ...
29798 (ipa_fn_summary_write): ... this one.
29799 (inline_free_summary): Rename to ...
29800 (ipa_free_fn_summary): ... this one.
29801 (pass_data_local_fn_summary, pass_local_fn_summary,
29802 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29803 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29804 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29805 make_pass_ipa_fn_summary): New.
29806 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29807 inline_write_summary, inline_free_summary): Remove.
29808 (ipa_free_fn_summary) : New.
29809 * ipa-inline.c (ipa_inline): Update.
29810 (pass_ipa_inline): Do not generate summaries.
29811 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29813 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29814 and add pass_ipa_fn_summary.
29815 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29817 (make_pass_inline_parameters): Remove.
29819 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29821 * omp-low.c (struct omp_context): Remove "default_kind" member.
29824 * omp-offload.c (execute_oacc_device_lower): Remove the
29825 parallelism dimensions function attributes for unparallelized
29826 OpenACC kernels constructs.
29828 2017-05-23 Martin Liska <mliska@suse.cz>
29830 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29832 (cgraph_edge::make_speculative): Likewise.
29833 (cgraph_edge::resolve_speculation): Likewise.
29834 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29835 (cgraph_node::dump): Likewise.
29836 * cgraph.h: Likewise.
29837 * cgraphunit.c (analyze_functions): Likewise.
29838 (symbol_table::compile): Likewise.
29839 * ipa-cp.c (print_all_lattices): Likewise.
29840 (determine_versionability): Likewise.
29841 (initialize_node_lattices): Likewise.
29842 (ipcp_verify_propagated_values): Likewise.
29843 (estimate_local_effects): Likewise.
29844 (update_profiling_info): Likewise.
29845 (create_specialized_node): Likewise.
29846 (perhaps_add_new_callers): Likewise.
29847 (decide_about_value): Likewise.
29848 (decide_whether_version_node): Likewise.
29849 (identify_dead_nodes): Likewise.
29850 (ipcp_store_bits_results): Likewise.
29851 * ipa-devirt.c (dump_targets): Likewise.
29852 (ipa_devirt): Likewise.
29853 * ipa-icf.c (sem_item::dump): Likewise.
29854 (sem_function::equals): Likewise.
29855 (sem_variable::equals): Likewise.
29856 (sem_item_optimizer::read_section): Likewise.
29857 (sem_item_optimizer::execute): Likewise.
29858 (congruence_class::dump): Likewise.
29859 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29860 (dump_inline_summary): Likewise.
29861 (estimate_node_size_and_time): Likewise.
29862 (inline_analyze_function): Likewise.
29863 * ipa-inline-transform.c (inline_call): Likewise.
29864 * ipa-inline.c (report_inline_failed_reason): Likewise.
29865 (want_early_inline_function_p): Likewise.
29866 (edge_badness): Likewise.
29867 (update_edge_key): Likewise.
29868 (inline_small_functions): Likewise.
29869 * ipa-profile.c (ipa_profile): Likewise.
29870 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29871 (ipa_make_edge_direct_to_target): Likewise.
29872 (remove_described_reference): Likewise.
29873 (ipa_impossible_devirt_target): Likewise.
29874 (propagate_controlled_uses): Likewise.
29875 (ipa_print_node_params): Likewise.
29876 (ipcp_transform_function): Likewise.
29877 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29878 (propagate_pure_const): Likewise.
29879 * ipa-reference.c (generate_summary): Likewise.
29880 (read_write_all_from_decl): Likewise.
29881 (propagate): Likewise.
29882 (ipa_reference_read_optimization_summary): Likewise.
29883 * ipa-utils.c (ipa_merge_profiles): Likewise.
29884 * ipa.c (walk_polymorphic_call_targets): Likewise.
29885 (symbol_table::remove_unreachable_nodes): Likewise.
29886 (ipa_single_use): Likewise.
29887 * passes.c (execute_todo): Likewise.
29888 * predict.c (drop_profile): Likewise.
29889 * symtab.c (symtab_node::get_dump_name): New function.
29890 (symtab_node::dump_name): Likewise.
29891 (symtab_node::dump_asm_name): Likewise.
29892 (symtab_node::dump_references): Likewise.
29893 (symtab_node::dump_referring): Likewise.
29894 (symtab_node::dump_base): Likewise.
29895 (symtab_node::debug_symtab): Likewise.
29896 * tree-sra.c (convert_callers_for_node): Likewise.
29897 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29898 * value-prof.c (init_node_map): Likewise.
29900 2017-05-23 Martin Liska <mliska@suse.cz>
29902 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29903 and symtab_node::debug_symtab to symbol_table::debug.
29904 * cgraphunit.c (analyze_functions): Use the renamed function.
29905 (symbol_table::compile): Likewise.
29906 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29907 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29908 * passes.c (execute_todo): Likewise.
29909 * symtab.c (symbol_table::dump): New function.
29910 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29912 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29914 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29915 that nonconst implies exec.
29917 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29919 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29920 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29921 (inline_edge_summary_vec): Turn into ...
29922 (ipa_call_summaries): ... this one.
29923 (redirect_to_unreachable, edge_set_predicate,
29924 evaluate_properties_for_edge, inline_summary_alloc,
29925 reset_ipa_call_summary, reset_inline_summary,
29926 inline_summary_t::duplicate): Update.
29927 (inline_edge_duplication_hook): Turn to ...
29928 (ipa_call_summary_t::duplicate): ... this one.
29929 (inline_edge_removal_hook): Turn to ...
29930 (ipa_call_summary_t::remove): ... this one.
29931 (dump_inline_edge_summary): Turn to ...
29932 (dump_ipa_call_summary): ... this one.
29933 (estimate_function_body_sizes): Update.
29934 (inline_update_callee_summaries): Update.
29935 (remap_edge_change_prob): Update.
29936 (remap_edge_summaries): Update.
29937 (inline_merge_summary): Update.
29938 (do_estimate_edge_time): Update.
29939 (inline_generate_summary): Update.
29940 (inline_read_section): Update.
29941 (inline_read_summary): Update.
29942 (inline_free_summary): Update.
29943 * ipa-inline.c (can_inline_edge_p): Update.
29944 (compute_inlined_call_time): Update.
29945 (want_inline_small_function_p): Update.
29946 (edge_badness): Update.
29947 (early_inliner): Update.
29948 * ipa-inline.h (inline_edge_summary): Turn to ...
29949 (ipa_call_summary): ... this one.
29950 (ipa_call_summary_t): New class.
29951 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29952 (ipa_call_summaries): New.
29953 (inline_edge_summary): Remove.
29954 (estimate_edge_growth): Update.
29955 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29956 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29957 * ipa-split.c (execute_split_functions): Update.
29958 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29960 2017-05-23 Tom de Vries <tom@codesourcery.com>
29962 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29963 attributes): Document rdrand effective target.
29965 2017-05-23 Tom de Vries <tom@codesourcery.com>
29967 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29968 attributes): Sort alphabetically.
29970 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29972 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29974 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29977 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29978 V2DF/V2DI splat into two separate patterns, one that handles
29979 registers, and the other that only handles memory. Drop support
29980 for splatting from a GPR on ISA 2.07 and then splitting the
29981 splat into direct move and splat.
29982 (vsx_splat_<mode>_reg): Likewise.
29983 (vsx_splat_<mode>_mem): Likewise.
29985 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29987 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29989 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29991 PR middle-end/80809
29992 * omp-low.c (finish_taskreg_remap): New function.
29993 (finish_taskreg_scan): If unit size of ctx->record_type
29994 is non-constant, unshare the size expression and replace
29995 decls in it with possible outer var refs.
29997 PR middle-end/80809
29998 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29999 GOVD_SHARED rather than GOVD_PRIVATE with it.
30000 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30001 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30003 PR middle-end/80853
30004 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30005 as last argument to build_outer_var_ref for pointer bases of array
30006 section reductions.
30008 2017-05-19 Martin Sebor <msebor@redhat.com>
30010 * print-tree.c (print_node): Print DECL_READ_P flag.
30012 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30014 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30015 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30016 * cgraph.c: Likewise.
30017 * cgraphunit.c: Likewise.
30018 * gengtype.c: Likewise.
30019 * ipa-cp.c: Likewise.
30020 * ipa-devirt.c: Likewise.
30021 * ipa-icf.c: Likewise.
30022 * ipa-predicate.c: Likewise.
30023 * ipa-profile.c: Likewise.
30024 * ipa-prop.c: Likewise.
30025 * ipa-split.c: Likewise.
30027 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30028 edge_predicate_pool, dump_inline_hints,
30029 inline_summary::account_size_time, redirect_to_unreachable,
30030 edge_set_predicate, set_hint_predicate,
30031 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30032 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30033 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30034 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30035 ipa_call_summary_t::remove, initialize_growth_caches,
30036 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30037 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30038 mark_modified, unmodified_parm_1, unmodified_parm,
30039 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30040 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30041 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30042 will_be_nonconstant_predicate, record_modified_bb_info,
30043 get_minimal_bb, record_modified, param_change_prob,
30044 phi_result_unknown_predicate, predicate_for_phi_result,
30045 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30046 estimate_function_body_sizes, compute_inline_parameters,
30047 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30048 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30049 inline_update_callee_summaries, remap_edge_change_prob,
30050 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30051 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30052 inline_analyze_function, inline_summary_t::insert,
30053 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30054 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30055 inline_free_summary): Move to ipa-fnsummary.h
30056 (predicate_t): Remove.
30057 * ipa-fnsummary.c: New file.
30058 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30059 (enum inline_hints_vals, inline_hints, agg_position_info,
30060 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30061 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30062 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30063 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30064 inline_read_summary, inline_write_summary, inline_free_summary,
30065 inline_analyze_function, initialize_inline_failed,
30066 inline_merge_summary, inline_update_overall_summary,
30067 compute_inline_parameters): Move to ipa-fnsummary.h
30068 * ipa-fnsummary.h: New file.
30069 * ipa-inline-transform.h: Include ipa-inline.h.
30070 * ipa-inline.c: LIkewise.
30072 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30074 * ipa-inline.c (edge_badness): Use inlined_time instead of
30075 inline_summaries->get.
30077 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30079 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30081 2017-05-22 Nathan Sidwell <nathan@acm.org>
30083 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30084 (fdump-lang): Document 'raw' option.
30085 * dumpfile.h (TDI_tu): Delete.
30086 * dumpfile.c (dump_files): Remove translation-unit.
30087 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30089 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30091 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30092 command option from $(AWK) call.
30093 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30094 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30095 [FORMAT]: Remove handling of variable.
30096 * config/avr/t-multilib: Regenerate.
30098 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30100 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30102 (dump_inline_summary): Do not print self_time.
30103 (estimate_function_body_sizes): Do not set self_time.
30104 (compute_inline_parameters): Likewise.
30105 (inline_read_section, inline_write_summary): Do not stream self_time.
30106 * ipa-inline.h (inline_summary): Drop self_time.
30108 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30110 * ipa-inline-analysis.c (account_size_time): Rename to ...
30111 (inline_summary::account_size_time): ... this one.
30112 (reset_ipa_call_summary): Turn to ...
30113 (ipa_call_summary::reset): ... this one.
30114 (reset_inline_summary): Turn to ...
30115 (inline_summary::reset): ... this one.
30116 (inline_summary_t::remove): Update.
30117 (inline_summary_t::duplicate): Update.
30118 (ipa_call_summary_t::remove): Update.
30119 (dump_inline_summary): Update.
30120 (estimate_function_body_sizes): Update.
30121 (compute_inline_parameters): Update.
30122 (estimate_node_size_and_time): Update.
30123 (inline_merge_summary): Update.
30124 (inline_update_overall_summary): Update.
30125 (inline_read_section): Update.
30126 (inline_write_summary): Update.
30127 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30128 add account_size_time and reset member functions.
30129 (ipa_call_summary): Add reset function.
30130 * ipa-predicate.h (predicate::operator &): Constify.
30132 2017-05-22 Richard Biener <rguenther@suse.de>
30134 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30136 2017-05-19 Jason Merrill <jason@redhat.com>
30138 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30140 2017-05-19 Marek Polacek <polacek@redhat.com>
30143 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30144 TYPE_OVERFLOW_WRAPS checks.
30146 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30148 * tree-core.h (enum omp_clause_default_kind): Add
30149 "OMP_CLAUSE_DEFAULT_PRESENT".
30150 * tree-pretty-print.c (dump_omp_clause): Handle it.
30151 * gimplify.c (enum gimplify_omp_var_data): Add
30152 "GOVD_MAP_FORCE_PRESENT".
30153 (gimplify_adjust_omp_clauses_1): Map it to
30154 "GOMP_MAP_FORCE_PRESENT".
30155 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30157 * gimplify.c (oacc_default_clause): Clarify.
30159 2017-05-19 Nathan Sidwell <nathan@acm.org>
30161 LANG_HOOK_REGISTER_DUMPS
30162 * toplev.c (general_init): Call register dump lang hook.
30163 * doc/invoke.texi: Document -fdump-lang option family.
30164 * dumpfile.c (dump_files): Remove class dump here.
30165 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30166 * dumpfile.h (tree_dump_index): Remove TDI_class.
30167 * langhooks-def.h (lhd_register_dumps): Declare.
30168 (LANG_HOOKS_REGISTER_DUMPS): Define.
30169 (LANG_HOOKS_INITIALIZER): Add it.
30170 * langhooks.c (lhd_register_dumps): Define.
30171 * langhooks.h (struct lang_hooks): Add register_dumps.
30173 2017-05-19 Nathan Sidwell <nathan@acm.org>
30175 * context.h (context::set_passes): New.
30176 * context.c (context::context): Do not create pass manager.
30177 * toplev.c (general_init): Create pass manager here.
30179 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30181 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30182 use this splitter if two add or or instructions would also work for
30183 the constant we want to generate.
30185 2017-05-19 Richard Biener <rguenther@suse.de>
30188 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30189 predicate evaluation.
30191 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30193 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30195 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30198 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30200 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30201 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30202 (inline_edge_summary_vec): Turn into ...
30203 (ipa_call_summaries): ... this one.
30204 (redirect_to_unreachable, edge_set_predicate,
30205 evaluate_properties_for_edge, inline_summary_alloc,
30206 reset_ipa_call_summary, reset_inline_summary,
30207 inline_summary_t::duplicate): Update.
30208 (inline_edge_duplication_hook): Turn to ...
30209 (ipa_call_summary_t::duplicate): ... this one.
30210 (inline_edge_removal_hook): Turn to ...
30211 (ipa_call_summary_t::remove): ... this one.
30212 (dump_inline_edge_summary): Turn to ...
30213 (dump_ipa_call_summary): ... this one.
30214 (estimate_function_body_sizes): Update.
30215 (inline_update_callee_summaries): Update.
30216 (remap_edge_change_prob): Update.
30217 (remap_edge_summaries): Update.
30218 (inline_merge_summary): Update.
30219 (do_estimate_edge_time): Update.
30220 (inline_generate_summary): Update.
30221 (inline_read_section): Update.
30222 (inline_read_summary): Update.
30223 (inline_free_summary): Update.
30224 * ipa-inline.c (can_inline_edge_p): Update.
30225 (compute_inlined_call_time): Update.
30226 (want_inline_small_function_p): Update.
30227 (edge_badness): Update.
30228 (early_inliner): Update.
30229 * ipa-inline.h (inline_edge_summary): Turn to ...
30230 (ipa_call_summary): ... this one.
30231 (ipa_call_summary_t): New class.
30232 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30233 (ipa_call_summaries): New.
30234 (inline_edge_summary): Remove.
30235 (estimate_edge_growth): Update.
30236 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30237 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30238 * ipa-split.c (execute_split_functions): Update.
30239 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30241 2017-05-19 Richard Biener <rguenther@suse.de>
30243 PR middle-end/80764
30244 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30246 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30248 * config/rs6000/rs6000.c (struct machine_function): Add field
30249 fpr_is_wrapped_separately.
30250 (rs6000_get_separate_components): Use 64 components. Handle the
30251 new FPR components.
30252 (rs6000_components_for_bb): Handle the FPR components.
30253 (rs6000_emit_prologue_components): Handle the FPR components.
30254 (rs6000_emit_epilogue_components): Handle the FPR components.
30255 (rs6000_set_handled_components): Handle the FPR components.
30256 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30257 that are already separately shrink-wrapped.
30258 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30259 that are already separately shrink-wrapped.
30261 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30264 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30267 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30268 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30269 cases where the register allocator uses a move and an offsettable
30270 memory operation to/from a FPR register on ISA 2.06/2.07.
30271 (define_peephole2 for Altivec d-form store): Likewise.
30273 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30276 * config/i386/mmx.md (*mov<mode>_internal): Enable
30277 alternatives 11, 12, 13 and 14 also for 32bit targets.
30278 Remove alternatives 15, 16, 17 and 18.
30279 * config/i386/sse.md (vec_concatv2di): Change
30280 alternative (!x, *y) to (x, ?!*Yn).
30282 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30284 * dumpfile.h (enum dump_kind): Remove stray comma.
30286 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30288 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30289 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30290 predicate::num_conditions
30291 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30292 (CHANGED): turn into predicate::changed.
30293 (agg_position_info): Move to ipa-predicate.h
30294 (add_condition, predicate::add_clause, predicate::operator &=,
30295 predicate::or_with, predicate::evaluate, predicate::probability,
30296 dump_condition, dump_clause, predicate::dump,
30297 predicate::remap_after_duplication, predicate::remap_after_inlining,
30298 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30299 (evaluate_conditions_for_known_args): Update.
30300 (set_cond_stmt_execution_predicate): Update.
30301 * ipa-inline.h: Include ipa-predicate.h
30302 (condition, inline_param_summary, conditions, agg_position_info,
30303 predicate): Move to ipa-predicate.h
30304 * ipa-predicate.c: New file.
30305 * ipa-predicate.h: New file.
30307 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30309 * final.c (leaf_function_p): Check we are not in a sequence.
30311 2017-05-18 Martin Liska <mliska@suse.cz>
30313 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30314 * dumpfile.c (dump_register): Use new enum dump_kind.
30315 (get_dump_file_name): Likewise.
30316 (dump_enable_all): Likewise.
30317 (dump_switch_p_1): Likewise.
30318 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30319 * dumpfile.h (enum dump_kind): New enum type.
30320 (struct dump_file_info): Create constructor and
30321 format fields and comments.
30322 * passes.c (pass_manager::register_one_dump_file):
30324 * statistics.c (statistics_early_init): Likewise.
30325 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30326 TDF_TREE with TDF_SLIM.
30327 (gather_memory_references_ref): Likewise.
30329 2017-05-18 Martin Liska <mliska@suse.cz>
30331 * vec.h (struct vnull): Use it.
30333 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30335 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30336 (true_predicate, false_predicate, true_predicate_p,
30337 false_predicate_p): Remove.
30338 (single_cond_predicate, not_inlined_predicate): Turn to member function
30340 (add_condition): Update.
30341 (add_clause): Turn to...
30342 (predicate::add_clause): ... this one; update; allow passing NULL
30344 (and_predicates): Turn to ...
30345 (predicate::operator &=): ... this one.
30346 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30347 (or_predicates): Turn to ...
30348 (predicate::or_with): ... this one.
30349 (evaluate_predicate): Turn to ...
30350 (predicate::evaluate): ... this one.
30351 (predicate_probability): Turn to ...
30352 (predicate::probability): ... this one.
30353 (dump_condition): Update.
30354 (dump_predicate): Turn to ...
30355 (predicate::dump): ... this one.
30356 (account_size_time): Update.
30357 (edge_set_predicate): Update.
30358 (set_hint_predicate): UPdate.
30359 (evaluate_conditions_for_known_args): Update.
30360 (evaluate_properties_for_edge): Update.
30361 (remap_predicate_after_duplication): Turn to...
30362 (predicate::remap_after_duplication): ... this one.
30363 (remap_hint_predicate_after_duplication): Update.
30364 (inline_summary_t::duplicate): UPdate.
30365 (dump_inline_edge_summary): Update.
30366 (dump_inline_summary): Update.
30367 (set_cond_stmt_execution_predicate): Update.
30368 (set_switch_stmt_execution_predicate): Update.
30369 (compute_bb_predicates): Update.
30370 (will_be_nonconstant_expr_predicate): Update.
30371 (will_be_nonconstant_predicate): Update.
30372 (phi_result_unknown_predicate): Update.
30373 (predicate_for_phi_result): Update.
30374 (array_index_predicate): Update.
30375 (estimate_function_body_sizes): Update.
30376 (estimate_node_size_and_time): Update.
30377 (estimate_ipcp_clone_size_and_time): Update.
30378 (remap_predicate): Rename to ...
30379 (predicate::remap_after_inlining): ... this one.
30380 (remap_hint_predicate): Update.
30381 (inline_merge_summary): Update.
30382 (inline_update_overall_summary): Update.
30383 (estimate_size_after_inlining): Update.
30384 (read_predicate): Rename to ...
30385 (predicate::stream_in): ... this one.
30386 (read_inline_edge_summary): Update.
30387 (write_predicate): Rename to ...
30388 (predicate::stream_out): ... this one.
30389 (write_inline_edge_summary): Update.
30390 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30391 (clause_t): Turn to uint32_t
30392 (predicate): Turn to class; implement constructor and operators
30394 (size_time_entry): Update.
30395 (inline_summary): Update.
30396 (inline_edge_summary): Update.
30398 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30400 * fold-const.c (fold_binary_loc): Move transformation...
30401 * match.pd (C - X CMP X): ... here.
30403 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30405 * config/sparc/sparc.c (sparc_option_override): Set function
30406 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30407 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30409 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30411 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30413 2017-05-18 Marek Polacek <polacek@redhat.com>
30416 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30417 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30418 gimple_assign_load_p.
30420 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30422 PR middle-end/80692
30423 * real.c (do_compare): Give decimal_do_compare preference over
30424 comparing just the signs.
30426 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30428 * doc/md.texi (Canonicalization of Instructions): Describe the
30429 canonical form of instructions that inherently set a condition
30432 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30434 PR middle-end/80775
30435 * tree-cfg.c: Move deletion of unreachable case statements to after
30436 the merging of consecutive case labels.
30438 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30440 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30441 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30442 restoring of callee-saved registers.
30444 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30446 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30447 * config/visium/visium.c (single_set_and_flags): Likewise.
30448 * config/visium/visium.md (Substitutions): Likewise.
30450 2017-05-17 Martin Liska <mliska@suse.cz>
30452 * cfg.c: Introduce dump_flags_t type and
30453 use it instead of int type.
30455 * cfghooks.c: Likewise.
30456 * cfghooks.h (struct cfg_hooks): Likewise.
30457 * cfgrtl.c: Likewise.
30458 * cfgrtl.h: Likewise.
30459 * cgraph.c (cgraph_node::get_body): Likewise.
30460 * coretypes.h: Likewise.
30461 * domwalk.c: Likewise.
30462 * domwalk.h: Likewise.
30463 * dumpfile.c (struct dump_option_value_info): Likewise.
30464 (dump_enable_all): Likewise.
30465 (dump_switch_p_1): Likewise.
30466 (opt_info_switch_p): Likewise.
30467 * dumpfile.h (enum tree_dump_index): Likewise.
30468 (struct dump_file_info): Likewise.
30469 * genemit.c: Likewise.
30470 * generic-match-head.c: Likewise.
30471 * gengtype.c (open_base_files): Likewise.
30472 * gimple-pretty-print.c: Likewise.
30473 * gimple-pretty-print.h: Likewise.
30474 * graph.c (print_graph_cfg): Likewise.
30475 * graphite-scop-detection.c (dot_all_sese): Likewise.
30476 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30477 * loop-unroll.c (report_unroll): Likewise.
30478 * passes.c (pass_manager::register_one_dump_file): Likewise.
30479 * print-tree.c: Likewise.
30480 * statistics.c: Likewise.
30481 * tree-cfg.c: Likewise.
30482 * tree-cfg.h: Likewise.
30483 * tree-dfa.c: Likewise.
30484 * tree-dfa.h: Likewise.
30485 * tree-dump.c (dump_function): Likewise.
30486 * tree-dump.h (struct dump_info): Likewise.
30487 * tree-pretty-print.c: Likewise.
30488 * tree-pretty-print.h: Likewise.
30489 * tree-ssa-live.c: Likewise.
30490 * tree-ssa-live.h: Likewise.
30491 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30492 * tree-vect-loop.c: Likewise.
30493 * tree-vect-slp.c: Likewise.
30495 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30496 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30498 PR tree-optimization/80457
30499 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30500 of all arguments to a statement as scalar_to_vec operations.
30501 (vectorizable_call): Adjust call to vect_model_simple_cost for
30503 (vectorizable_conversion): Likewise.
30504 (vectorizable_assignment): Likewise.
30505 (vectorizable_shift): Likewise.
30506 (vectorizable_operation): Likewise.
30507 (vectorizable_comparison): Likewise.
30508 (vect_is_simple_cond): Record the def types for operands.
30509 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30510 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30511 for statement argument count.
30513 2017-05-16 Carl Love <cel@us.ibm.com>
30515 * config/rs6000/rs6000-c: Add support for built-in functions
30516 vector unsigned long long vec_bperm (vector unsigned long long,
30517 vector unsigned char)
30518 vector signed long long vec_mule (vector signed int,
30520 vector unsigned long long vec_mule (vector unsigned int,
30521 vector unsigned int)
30522 vector signed long long vec_mulo (vector signed int,
30524 vector unsigned long long vec_mulo (vector unsigned int,
30525 vector unsigned int)
30526 vector signed char vec_sldw (vector signed char,
30527 vector signed char,
30529 vector unsigned char vec_sldw (vector unsigned char,
30530 vector unsigned char,
30532 vector signed short vec_sldw (vector signed short,
30533 vector signed short,
30535 vector unsigned short vec_sldw (vector unsigned short,
30536 vector unsigned short,
30538 vector signed int vec_sldw (vector signed int,
30541 vector unsigned int vec_sldw (vector unsigned int,
30542 vector unsigned int,
30544 vector signed long long vec_sldw (vector signed long long,
30545 vector signed long long,
30547 vector unsigned long long vec_sldw (vector unsigned long long,
30548 vector unsigned long long,
30550 * config/rs6000/rs6000-c: Add support for built-in functions
30551 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30552 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30553 * doc/extend.texi: Update the built-in documentation for the
30554 new built-in functions.
30556 2017-05-16 Marek Polacek <polacek@redhat.com>
30560 * tree.c (save_expr): Don't fold the expression.
30562 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30564 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30565 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30566 and (?*y,m). Update insn attributes.
30568 2017-05-16 Martin Liska <mliska@suse.cz>
30570 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30571 flags argument of print_gimple_stmt, print_gimple_expr,
30572 print_generic_stmt and print_generic_expr.
30573 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30574 * coretypes.h: Likewise.
30575 * except.c (dump_eh_tree): Likewise.
30576 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30577 * gimple-pretty-print.h: Likewise.
30578 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30579 (backprop::push_to_worklist): Likewise.
30580 (backprop::pop_from_worklist): Likewise.
30581 (backprop::process_use): Likewise.
30582 (backprop::intersect_uses): Likewise.
30583 (note_replacement): Likewise.
30584 * gimple-ssa-store-merging.c
30585 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30586 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30587 (pass_store_merging::execute): Likewise.
30588 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30589 (ssa_base_cand_dump_callback): Likewise.
30590 (dump_incr_vec): Likewise.
30591 (replace_refs): Likewise.
30592 (replace_mult_candidate): Likewise.
30593 (create_add_on_incoming_edge): Likewise.
30594 (create_phi_basis): Likewise.
30595 (insert_initializers): Likewise.
30596 (all_phi_incrs_profitable): Likewise.
30597 (introduce_cast_before_cand): Likewise.
30598 (replace_one_candidate): Likewise.
30599 * gimplify.c (gimplify_expr): Likewise.
30600 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30601 (set_rename): Likewise.
30602 (rename_uses): Likewise.
30603 (copy_loop_phi_nodes): Likewise.
30604 (add_close_phis_to_merge_points): Likewise.
30605 (copy_loop_close_phi_args): Likewise.
30606 (copy_cond_phi_args): Likewise.
30607 (graphite_copy_stmts_from_block): Likewise.
30608 (translate_pending_phi_nodes): Likewise.
30609 * graphite-poly.c (print_pdr): Likewise.
30610 (dump_gbb_cases): Likewise.
30611 (dump_gbb_conditions): Likewise.
30612 (print_scop_params): Likewise.
30613 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30614 (build_cross_bb_scalars_use): Likewise.
30615 (gather_bbs::before_dom_children): Likewise.
30616 * hsa-dump.c (dump_hsa_immed): Likewise.
30617 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30618 (get_replacement_map): Likewise.
30619 * ipa-inline-analysis.c (dump_condition): Likewise.
30620 (estimate_function_body_sizes): Likewise.
30621 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30622 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30623 * ipa-prop.c (ipa_dump_param): Likewise.
30624 (ipa_print_node_jump_functions_for_edge): Likewise.
30625 (ipa_modify_call_arguments): Likewise.
30626 (ipa_modify_expr): Likewise.
30627 (ipa_dump_param_adjustments): Likewise.
30628 (ipa_dump_agg_replacement_values): Likewise.
30629 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30630 * ipa-pure-const.c (check_stmt): Likewise.
30631 (pass_nothrow::execute): Likewise.
30632 * ipa-split.c (execute_split_functions): Likewise.
30633 * omp-offload.c (dump_oacc_loop_part): Likewise.
30634 (dump_oacc_loop): Likewise.
30635 * trans-mem.c (tm_log_emit): Likewise.
30636 (tm_memopt_accumulate_memops): Likewise.
30637 (dump_tm_memopt_set): Likewise.
30638 (dump_tm_memopt_transform): Likewise.
30639 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30640 (print_loop): Likewise.
30641 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30642 (chkp_gather_checks_info): Likewise.
30643 (chkp_get_check_result): Likewise.
30644 (chkp_remove_check_if_pass): Likewise.
30645 (chkp_use_outer_bounds_if_possible): Likewise.
30646 (chkp_reduce_bounds_lifetime): Likewise.
30647 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30648 (chkp_mark_completed_bounds): Likewise.
30649 (chkp_register_incomplete_bounds): Likewise.
30650 (chkp_mark_invalid_bounds): Likewise.
30651 (chkp_maybe_copy_and_register_bounds): Likewise.
30652 (chkp_build_returned_bound): Likewise.
30653 (chkp_get_bound_for_parm): Likewise.
30654 (chkp_build_bndldx): Likewise.
30655 (chkp_get_bounds_by_definition): Likewise.
30656 (chkp_generate_extern_var_bounds): Likewise.
30657 (chkp_get_bounds_for_decl_addr): Likewise.
30658 * tree-chrec.c (chrec_apply): Likewise.
30659 * tree-data-ref.c (dump_data_reference): Likewise.
30660 (dump_subscript): Likewise.
30661 (dump_data_dependence_relation): Likewise.
30662 (analyze_overlapping_iterations): Likewise.
30663 * tree-inline.c (expand_call_inline): Likewise.
30664 (tree_function_versioning): Likewise.
30665 * tree-into-ssa.c (dump_defs_stack): Likewise.
30666 (dump_currdefs): Likewise.
30667 (dump_names_replaced_by): Likewise.
30668 (dump_update_ssa): Likewise.
30669 (update_ssa): Likewise.
30670 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30671 * tree-parloops.c (build_new_reduction): Likewise.
30672 (try_create_reduction_list): Likewise.
30673 (ref_conflicts_with_region): Likewise.
30674 (oacc_entry_exit_ok_1): Likewise.
30675 (oacc_entry_exit_single_gang): Likewise.
30676 * tree-pretty-print.h: Likewise.
30677 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30678 (get_scalar_evolution): Likewise.
30679 (add_to_evolution): Likewise.
30680 (get_loop_exit_condition): Likewise.
30681 (analyze_evolution_in_loop): Likewise.
30682 (analyze_initial_condition): Likewise.
30683 (analyze_scalar_evolution): Likewise.
30684 (instantiate_scev): Likewise.
30685 (number_of_latch_executions): Likewise.
30686 (gather_chrec_stats): Likewise.
30687 (final_value_replacement_loop): Likewise.
30688 (scev_const_prop): Likewise.
30689 * tree-sra.c (dump_access): Likewise.
30690 (disqualify_candidate): Likewise.
30691 (create_access): Likewise.
30692 (reject): Likewise.
30693 (maybe_add_sra_candidate): Likewise.
30694 (create_access_replacement): Likewise.
30695 (analyze_access_subtree): Likewise.
30696 (analyze_all_variable_accesses): Likewise.
30697 (sra_modify_assign): Likewise.
30698 (initialize_constant_pool_replacements): Likewise.
30699 (find_param_candidates): Likewise.
30700 (decide_one_param_reduction): Likewise.
30701 (replace_removed_params_ssa_names): Likewise.
30702 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30703 * tree-ssa-copy.c (dump_copy_of): Likewise.
30704 (copy_prop_visit_cond_stmt): Likewise.
30705 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30706 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30707 (record_equivalences_from_stmt): Likewise.
30708 * tree-ssa-dse.c (compute_trims): Likewise.
30709 (delete_dead_call): Likewise.
30710 (delete_dead_assignment): Likewise.
30711 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30712 (forward_propagate_into_cond): Likewise.
30713 (pass_forwprop::execute): Likewise.
30714 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30715 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30717 (move_computations_worker): Likewise.
30718 (execute_sm): Likewise.
30719 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30720 (remove_exits_and_undefined_stmts): Likewise.
30721 (remove_redundant_iv_tests): Likewise.
30722 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30723 (adjust_iv_update_pos): Likewise.
30724 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30725 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30726 (value_replacement): Likewise.
30727 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30728 * tree-ssa-pre.c (print_pre_expr): Likewise.
30729 (get_representative_for): Likewise.
30730 (create_expression_by_pieces): Likewise.
30731 (insert_into_preds_of_block): Likewise.
30732 (eliminate_insert): Likewise.
30733 (eliminate_dom_walker::before_dom_children): Likewise.
30734 (eliminate): Likewise.
30735 (remove_dead_inserted_code): Likewise.
30736 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30737 * tree-ssa-reassoc.c (get_rank): Likewise.
30738 (eliminate_duplicate_pair): Likewise.
30739 (eliminate_plus_minus_pair): Likewise.
30740 (eliminate_not_pairs): Likewise.
30741 (undistribute_ops_list): Likewise.
30742 (eliminate_redundant_comparison): Likewise.
30743 (update_range_test): Likewise.
30744 (optimize_range_tests_var_bound): Likewise.
30745 (optimize_vec_cond_expr): Likewise.
30746 (rewrite_expr_tree): Likewise.
30747 (rewrite_expr_tree_parallel): Likewise.
30748 (linearize_expr): Likewise.
30749 (break_up_subtract): Likewise.
30750 (linearize_expr_tree): Likewise.
30751 (attempt_builtin_powi): Likewise.
30752 (attempt_builtin_copysign): Likewise.
30753 (transform_stmt_to_copy): Likewise.
30754 (transform_stmt_to_multiply): Likewise.
30755 (dump_ops_vector): Likewise.
30756 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30757 (print_scc): Likewise.
30758 (set_ssa_val_to): Likewise.
30759 (visit_reference_op_store): Likewise.
30760 (visit_use): Likewise.
30761 (sccvn_dom_walker::before_dom_children): Likewise.
30762 (run_scc_vn): Likewise.
30763 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30765 (expr_hash_elt::print): Likewise.
30766 (const_and_copies::pop_to_marker): Likewise.
30767 (const_and_copies::record_const_or_copy_raw): Likewise.
30768 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30769 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30770 (dump_predicates): Likewise.
30771 (find_uninit_use): Likewise.
30772 (warn_uninitialized_phi): Likewise.
30773 (pass_late_warn_uninitialized::execute): Likewise.
30774 * tree-ssa.c (verify_vssa): Likewise.
30775 (verify_ssa): Likewise.
30776 (maybe_optimize_var): Likewise.
30777 * tree-vrp.c (dump_value_range): Likewise.
30778 (dump_all_value_ranges): Likewise.
30779 (dump_asserts_for): Likewise.
30780 (register_edge_assert_for_2): Likewise.
30781 (vrp_visit_cond_stmt): Likewise.
30782 (vrp_visit_switch_stmt): Likewise.
30783 (vrp_visit_stmt): Likewise.
30784 (vrp_visit_phi_node): Likewise.
30785 (simplify_cond_using_ranges_1): Likewise.
30786 (fold_predicate_in): Likewise.
30787 (evrp_dom_walker::before_dom_children): Likewise.
30788 (evrp_dom_walker::push_value_range): Likewise.
30789 (evrp_dom_walker::pop_value_range): Likewise.
30790 (execute_early_vrp): Likewise.
30792 2017-05-16 Richard Biener <rguenther@suse.de>
30794 * dwarf2out.c (loc_list_from_tree_1): Do not create
30795 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30797 2017-05-16 Richard Biener <rguenther@suse.de>
30799 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30801 (note_variable_value_in_expr): If we resolved the decl ref
30802 do not push to the stack.
30804 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30806 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30807 operations in fast-math mode.
30808 (vaddq_f16): Likewise.
30809 (vmul_f16): Likewise.
30810 (vmulq_f16): Likewise.
30811 (vsub_f16): Likewise.
30812 (vsubq_f16): Likewise.
30813 * config/arm/neon.md (add<mode>3): New.
30815 (fma:<VH:mode>3): New. Also remove outdated comment.
30818 2017-05-16 Martin Liska <mliska@suse.cz>
30822 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30823 (odr_types_equivalent_p): Likewise.
30825 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30827 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30829 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30832 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30833 non-interunit SSE move alternatives with '?'.
30834 (zero-extendsidi peephole2): New peephole to skip intermediate
30835 general register in SSE zero-extend sequence.
30837 2017-05-15 Jeff Law <law@redhat.com>
30839 * reorg.c (relax_delay_slots): Create a new variable to hold
30840 the temporary target rather than clobbering TARGET_LABEL.
30842 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30843 missing argument to extract_bit_field call.
30844 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30846 2017-05-15 Martin Liska <mliska@suse.cz>
30849 * gcc.c (process_command): Do not allow empty argument of -o option.
30851 2017-05-15 Renlin Li <renlin.li@arm.com>
30853 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30854 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30855 * config/aarch64/constraints.md (Usf): Add long call check.
30856 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30857 (call_value): Likewise.
30858 (sibcall): Likewise.
30859 (sibcall_value): Likewise.
30861 (call_value_insn): New.
30862 (sibcall_insn): Update rtx pattern.
30863 (sibcall_value_insn): Likewise.
30864 (call_internal): Remove.
30865 (call_value_internal): Likewise.
30866 (sibcall_internal): Likewise.
30867 (sibcall_value_internal): Likewise.
30868 (call_reg): Likewise.
30869 (call_symbol): Likewise.
30870 (call_value_reg): Likewise.
30871 (call_value_symbol): Likewise.
30873 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30876 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30878 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30880 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30881 compatible with CCGOCmode and with CCZmode.
30883 2017-05-14 Martin Sebor <msebor@redhat.com>
30885 PR middle-end/77671
30886 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30887 (gimple_fold_builtin_snprintf): Same.
30888 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30889 (gimple_fold_builtin_snprintf): Same.
30890 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30891 of character types.
30892 (is_call_safe): New function.
30893 (try_substitute_return_value): Call it.
30894 (try_simplify_call): New function.
30895 (pass_sprintf_length::handle_gimple_call): Call it.
30897 2017-05-14 Martin Sebor <msebor@redhat.com>
30899 PR middle-end/80669
30900 * builtins.c (expand_builtin_stpncpy): Simplify.
30902 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30904 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30905 * config/i386/i386.h
30906 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30907 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30908 (struct machine_function): Add new members call_ms2sysv,
30909 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30910 (struct machine_frame_state): New fields sp_realigned and
30911 sp_realigned_offset.
30912 * config/i386/i386.c
30913 (enum xlogue_stub): New enum.
30914 (enum xlogue_stub_sets): New enum.
30915 (class xlogue_layout): New class.
30916 (struct ix86_frame): New fields stack_realign_allocate_offset,
30917 stack_realign_offset and outlined_save_offset. Modify comments to
30918 detail stack layout when using out-of-line stubs.
30919 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30920 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30921 -mcall-ms2sysv-xlogues.
30922 (stub_managed_regs): New static variable.
30923 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30924 registers managed by out-of-line stub.
30925 (disable_call_ms2sysv_xlogues): New function.
30926 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30927 m->call_ms2sysv when appropriate and compute frame layout for
30929 (sp_valid_at, fp_valid_at): New inline functions.
30930 (choose_basereg): New function.
30931 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30933 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30934 Use align parameter of choose_baseaddr to generated aligned SSE movs
30936 (pro_epilogue_adjust_stack): Modify to track
30937 machine_frame_state::sp_realigned.
30938 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30939 (ix86_nsaved_sseregs): Likewise.
30940 (ix86_emit_save_regs): Likewise.
30941 (ix86_emit_save_regs_using_mov): Likewise.
30942 (ix86_emit_save_sse_regs_using_mov): Likewise.
30943 (get_scratch_register_on_entry): Likewise.
30944 (gen_frame_set): New function.
30945 (gen_frame_load): Likewise.
30946 (gen_frame_store): Likewise.
30947 (emit_outlined_ms2sysv_save): Likewise.
30948 (emit_outlined_ms2sysv_restore): Likewise.
30949 (ix86_expand_prologue): Modify stack re-alignment code and call
30950 emit_outlined_ms2sysv_save when appropriate.
30951 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30952 parameter rtx_insn *insn, which allows the function to be used to only
30953 generate the notes.
30954 (ix86_expand_epilogue): Modify validity checks of frame and stack
30955 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30956 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30957 * config/i386/predicates.md
30958 (save_multiple): New predicate.
30959 (restore_multiple): Likewise.
30960 * config/i386/sse.md
30961 (save_multiple<mode>): New pattern.
30962 (save_multiple_realign<mode>): Likewise.
30963 (restore_multiple<mode>): Likewise.
30964 (restore_multiple_and_return<mode>): Likewise.
30965 (restore_multiple_leave_return<mode>): Likewise.
30966 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30968 2017-05-14 Julia Koval <julia.koval@intel.com>
30970 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30971 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30972 (__builtin_ia32_xsetbv): New builtins.
30973 * config/i386/i386.c (ix86_expand_special_args_builtin):
30975 (ix86_expand_builtin): Special expand for new intrinsics.
30976 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30977 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30978 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30980 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30982 * cfganal.c (inverted_post_order_compute): Change argument type
30984 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30985 * df-core.c (rest_of_handle_df_initialize): Adjust.
30986 (rest_of_handle_df_finish): Likewise.
30987 (df_analyze_1): Likewise.
30988 (df_analyze): Likewise.
30989 (loop_inverted_post_order_compute): Change argument to be a vec *.
30990 (df_analyze_loop): Adjust.
30991 (df_get_n_blocks): Likewise.
30992 (df_get_postorder): Likewise.
30993 * df.h (struct df_d): Change field to be a vec.
30994 * lcm.c (compute_laterin): Adjust.
30995 (compute_available): Likewise.
30996 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30997 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30998 * tree-ssa-pre.c (compute_antic): Likewise.
31000 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31002 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31003 (depth_first_search::depth_first_search): Change structure init
31004 function to this constructor.
31005 (depth_first_search::add_bb): Rename function to this member.
31006 (depth_first_search::execute): Likewise.
31007 (flow_dfs_compute_reverse_finish): Adjust.
31009 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31011 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31012 (longest_simple_path): Likewise.
31013 * shrink-wrap.c (spread_components): Likewise.
31014 (disqualify_problematic_components): Likewise.
31015 (emit_common_heads_for_components): Likewise.
31016 (emit_common_tails_for_components): Likewise.
31017 (insert_prologue_epilogue_for_components): Likewise.
31019 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31021 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31024 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31026 * df-core.c (df_set_blocks): Start using auto_bitmap.
31027 (df_compact_blocks): Likewise.
31028 * df-problems.c (df_rd_confluence_n): Likewise.
31029 * df-scan.c (df_insn_rescan_all): Likewise.
31030 (df_process_deferred_rescans): Likewise.
31031 (df_update_entry_block_defs): Likewise.
31032 (df_update_exit_block_uses): Likewise.
31033 (df_entry_block_bitmap_verify): Likewise.
31034 (df_exit_block_bitmap_verify): Likewise.
31035 (df_scan_verify): Likewise.
31036 * lra-constraints.c (lra_constraints): Likewise.
31037 (undo_optional_reloads): Likewise.
31038 (lra_undo_inheritance): Likewise.
31039 * lra-remat.c (calculate_gen_cands): Likewise.
31040 (do_remat): Likewise.
31041 * lra-spills.c (assign_spill_hard_regs): Likewise.
31042 (spill_pseudos): Likewise.
31043 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31044 (bitmap_set_subtract_values): Likewise.
31046 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31048 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31049 management with auto_bitmap.
31050 (fix_inter_tick): Likewise.
31051 (fix_recovery_deps): Likewise.
31052 * ira.c (add_store_equivs): Likewise.
31053 (find_moveable_pseudos): Likewise.
31054 (split_live_ranges_for_shrink_wrap): Likewise.
31055 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31056 (rtx_reuse_manager::seen_def_p): Likewise.
31057 (rtx_reuse_manager::set_seen_def): Likewise.
31058 * print-rtl.h (class rtx_reuse_manager): Likewise.
31060 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31062 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31064 (migrate_btr_def): Likewise.
31065 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31066 * df-core.c (loop_post_order_compute): Likewise.
31067 (loop_inverted_post_order_compute): Likewise.
31068 * hsa-common.h: Likewise.
31069 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31070 * init-regs.c (initialize_uninitialized_regs): Likewise.
31071 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31072 (inline_small_functions): Likewise.
31073 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31074 * ira.c (combine_and_move_insns): Likewise.
31075 (build_insn_chain): Likewise.
31076 * loop-invariant.c (find_invariants): Likewise.
31077 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31078 * predict.c (tree_predict_by_opcode): Likewise.
31079 (predict_paths_leading_to): Likewise.
31080 (predict_paths_leading_to_edge): Likewise.
31081 (estimate_loops_at_level): Likewise.
31082 (estimate_loops): Likewise.
31083 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31084 (spread_components): Likewise.
31085 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31086 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31087 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31088 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31089 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31090 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31091 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31092 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31093 (mark_threaded_blocks): Likewise.
31094 (thread_through_all_blocks): Likewise.
31095 * tree-ssa.c (verify_ssa): Likewise.
31096 (execute_update_addresses_taken): Likewise.
31097 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31099 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31101 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31103 (post_order_compute): Likewise.
31104 (inverted_post_order_compute): Likewise.
31105 (pre_and_rev_post_order_compute_fn): Likewise.
31107 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31109 * genrecog.c (int_set::int_set): Explicitly construct our
31110 auto_vec base class.
31111 * vec.h (auto_vec::auto_vec): New constructor.
31113 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31115 * bitmap.h (class auto_bitmap): New constructor taking
31116 bitmap_obstack * argument.
31118 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31120 * bitmap.h (class auto_bitmap): Change type of m_bits to
31121 bitmap_head, and adjust ctor / dtor and member operators.
31123 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31125 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31126 when returned register mode doesn't match original mode.
31128 2017-05-12 Jeff Law <law@redhat.com>
31129 Jakub Jelinek <jakub@redhat.com>
31131 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31132 we look for cc setter after the compare-elim changes.
31133 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31134 within the vector to match what compare-elim now expects.
31135 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31136 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31138 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31139 after the compare-elim changes.
31140 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31141 the vector to match what compare-elim now expects.
31142 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31143 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31144 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31145 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31146 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31148 * config/visium/visium.c (single_set_and_flags): Fix where
31149 we look for cc setter after the compare-elim changes.
31150 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31151 with the vector to match what compare-elim now expects.
31152 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31153 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31154 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31155 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31156 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31157 (neg<mode>2_insn_set_overflow): Likewise.
31159 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31161 PR middle-end/79794
31162 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31163 maybe_expand_insn call, set ops[0].target. If still set after call,
31164 set alt_rtl. Add extra arg to recursive calls.
31165 (extract_bit_field): Add alt_rtl argument. Pass to
31167 * expmed.h (extract_bit_field): Fix prototype.
31168 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31169 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31170 to extract_bit_field_calls.
31171 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31172 Pass alt_rtl to extract_bit_field calls.
31173 * calls.c (store_unaligned_arguments_into_psuedos)
31174 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31175 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31177 * optabs.h (struct expand_operand): Add target bitfield.
31179 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31181 * compare-elim.c (try_eliminate_compare): Canonicalize
31182 operation with embedded compare to
31183 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31184 (set (reg) (operation)].
31186 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31188 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31191 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31192 cost of adding a carry flag for ADC instruction.
31193 [case MINUS]: Ignore the cost of subtracting a carry flag
31194 for SBB instruction.
31196 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31198 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31200 * config/rs6000/bmiintrin.h: New file.
31201 * config/rs6000/bmi2intrin.h: New file.
31202 * config/rs6000/x86intrin.h: New file.
31204 2017-05-12 Jeff Law <law@redhat.com>
31206 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31209 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31211 PR middle-end/80707
31212 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31214 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31216 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31217 early expansion of vector divide builtins.
31218 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31219 builtins identified as having unsigned arguments.
31221 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31223 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31224 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31225 expansion of vector logical operations (and, andc, or, xor,
31228 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31230 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31231 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31233 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31235 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31236 early GIMPLE expansion of vector multiplies.
31238 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31240 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31241 TARGET_HAVE_MOVT conditional.
31242 (movt splitter): Likewise.
31244 2017-05-12 Richard Biener <rguenther@suse.de>
31246 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31247 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31248 Fold all stmts not inplace.
31250 2017-05-12 Richard Biener <rguenther@suse.de>
31252 PR tree-optimization/80713
31253 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31254 inserted_exprs bit for not removed stmts.
31256 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31258 PR middle-end/69921
31259 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31260 parallelized" attribute for parallelized OpenACC kernels.
31261 * omp-offload.c (execute_oacc_device_lower): Use it.
31263 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31264 Set "oacc kernels" attribute.
31265 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31266 parameter. Adjust all users.
31267 (oacc_fn_attrib_kernels_p): Remove function.
31268 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31269 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31270 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31271 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31272 assert "oacc kernels" attribute is set.
31274 2017-05-11 Carl Love <cel@us.ibm.com>
31276 * config/rs6000/rs6000-c: Add support for built-in functions
31277 vector unsigned char vec_popcnt (vector signed char)
31278 vector unsigned char vec_popcnt (vector unsigned char)
31279 vector unsigned short vec_popcnt (vector signed short)
31280 vector unsigned short vec_popcnt (vector unsigned short)
31281 vector unsigned int vec_popcnt (vector signed int)
31282 vector unsigned int vec_popcnt (vector unsigned int)
31283 vector unsigned long long vec_popcnt (vector signed long long)
31284 vector unsigned long long vec_popcnt (vector unsigned long long)
31285 vector signed long long vec_slo (vector signed long long,
31286 vector signed char)
31287 vector signed long long vec_slo (vector signed long long,
31288 vector unsigned char)
31289 vector unsigned long long vec_slo (vector unsigned long long,
31290 vector signed char)
31291 vector unsigned long long vec_slo (vector unsigned long long,
31292 vector unsigned char)
31293 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31294 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31295 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31296 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31297 * doc/extend.texi: Update the built-in documentation file for the
31298 new built-in functions.
31300 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31302 * attribs.h (sorted_attr_string): Move machine independent
31303 functions for target clone support from the i386 port to common
31304 code. Rename ix86_function_versions to common_function_versions.
31305 Rename make_name to make_unique_name.
31306 (common_function_versions): Likewise.
31307 (make_unique_name): Likewise.
31308 (make_dispatcher_decl): Likewise.
31309 (is_function_default_version): Likewise.
31310 * attribs.c (attr_strcmp): Likewise.
31311 (sorted_attr_string): Likewise.
31312 (common_function_versions): Likewise.
31313 (make_unique_name): Likewise.
31314 (make_dispatcher_decl): Likewise.
31315 (is_function_default_version): Likewise.
31316 * config/i386/i386.c (attr_strcmp): Likewise.
31317 (sorted_attr_string): Likewise.
31318 (ix86_function_versions): Likewise.
31319 (make_name): Likewise.
31320 (make_dispatcher_decl): Likewise.
31321 (is_function_default_version): Likewise.
31322 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31324 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31327 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31328 Account for direct move costs for vec_construct of integer
31331 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31334 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31335 (UNSPEC_STX_ATOMIC): Ditto.
31336 (loaddi_via_sse): New insn.
31337 (storedi_via_sse): Ditto.
31338 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31339 Update corresponding peephole2 patterns.
31340 (atomic_storedi_fpu): Ditto.
31342 2017-05-11 Julia Koval <julia.koval@intel.com>
31344 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31345 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31347 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31348 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31349 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31351 2017-05-11 Nathan Sidwell <nathan@acm.org>
31353 * graphite-poly.c: Include dumpfile.h.
31355 * dumpfle.h (dump_function): Declare here ...
31356 * tree-dump.h (dump_function): ... not here.
31357 * dumpfile.c: #include tree-cfg.h.
31358 (dump_function): Move here from ...
31359 * tree-dump.c (dump_function): ... here.
31360 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31361 * graphite-poly.c: Don't include tree-dump.h.
31362 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31363 * print-tree.c: Likewise.
31364 * stor-layout.c: Likewise.
31365 * tree-nested.c: Likewise.
31367 * dumpfile.c (dump_start): Use TDF_FLAGS.
31368 (dump_enable_all): Fix TDF_KIND check thinko.
31370 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31372 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31373 array entries to represent two legal parameterizations of the
31374 overloaded __builtin_cmpb function, as represented by the
31375 P6_OV_BUILTIN_CMPB constant.
31376 (altivec_resolve_overloaded_builtin): Add special case handling
31377 for the __builtin_cmpb function, as represented by the
31378 P6_OV_BUILTIN_CMPB constant.
31379 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31380 (BU_P6_64BIT_2): New macro.
31381 (BU_P6_OVERLOAD_2): New macro
31382 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31383 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31384 (CMPB): Add overload support to represent both 32-bit and 64-bit
31385 compare-bytes function.
31386 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31387 support for TARGET_CMPB.
31388 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31389 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31390 documentation of the __builtin_cmpb overloaded built-in function.
31392 2017-05-11 Richard Biener <rguenther@suse.de>
31394 PR tree-optimization/80705
31395 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31396 bases are not vectorizable.
31398 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31400 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31401 when counting register pressure.
31403 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31405 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31406 (iv_ca_more_deps): Renamed to ...
31407 (iv_ca_compare_deps): ... this.
31408 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31410 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31412 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31414 (determine_group_iv_costs): ... here.
31415 (find_inv_vars_cb): Record inv var if it's not recorded before.
31417 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31419 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31420 (get_shiftadd_cost): Ditto.
31422 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31424 * tree-ssa-address.c: Include header file.
31425 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31427 (add_to_parts): Refactor.
31428 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31429 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31432 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31434 PR tree-optimization/53090
31435 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31437 (extract_cond_operands): Detect condition with IV on both sides
31438 and return COMP_IV_EXPR_2.
31439 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31440 (rewrite_use_compare): Simplify by removing call to function
31441 extract_cond_operands.
31443 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31445 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31446 (extract_cond_operands): Detect condition comparing against non-
31447 invariant bound and return appropriate enum value.
31448 (find_interesting_uses_cond): Update use of extract_cond_operands.
31449 Handle its return value accordingly.
31450 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31452 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31454 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31455 nonlinear iv_use computation in loop invariant sensitive way.
31457 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31459 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31460 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31462 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31464 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31465 (dump_cand): Support iv_cand.inv_exprs.
31466 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31468 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31471 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31473 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31475 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31476 as local function. Include necessary header files.
31477 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31479 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31481 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31483 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31485 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31486 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31487 RSHIFT_EXPR and BIT_NOT_EXPR.
31489 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31491 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31492 (adjust_setup_cost): New parameter supporting round up adjustment.
31493 (struct address_cost_data): Delete.
31494 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31495 (split_address_cost, ptr_difference_cost): Delete.
31496 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31497 (struct ainc_cost_data): New struct.
31498 (get_address_cost_ainc): New function.
31499 (get_address_cost, get_computation_cost): Reimplement.
31500 (determine_group_iv_cost_address): Record inv_expr for all uses of
31502 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31503 (iv_ca_has_deps): Reimplemented to ...
31504 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31506 (iv_ca_extend): Call iv_ca_more_deps.
31508 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31510 * tree-ssa-address.c (struct mem_address): Move to header file.
31511 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31512 * tree-ssa-address.h (struct mem_address): Move from C file.
31513 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31515 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31517 * tree-affine.h (aff_combination_type): New interface.
31518 (aff_combination_zero_p): Remove static.
31519 (aff_combination_const_p): New interface.
31520 (aff_combination_singleton_var_p): New interfaces.
31522 2017-05-11 Richard Biener <rguenther@suse.de>
31524 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31525 Skip unreachable blocks and destinations.
31526 (eliminate): Move stmt removal and fixup ...
31527 (fini_eliminate): ... here. Skip inserted exprs.
31528 (pass_pre::execute): Move fini_pre after fini_eliminate.
31529 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31530 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31531 PRE to get rid of dead code that has invalid SSA form and
31532 split critical edges again.
31534 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31536 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31538 2017-05-11 Richard Biener <rguenther@suse.de>
31540 * passes.c (execute_function_todo): Verify loops if they are
31541 said to be up-to-date.
31542 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31543 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31545 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31548 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31549 handle calling assemble_external ourself.
31552 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31553 modes with zero size. Enhance comment.
31555 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31557 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31558 built-ins for vec_xl and vec_xst with short and char pointer
31561 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31563 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31564 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31565 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31566 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31567 (_mm_maskz_min_round_ss): New intrinsics.
31568 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31569 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31570 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31571 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31572 (__builtin_ia32_minss_mask_round): New builtins.
31573 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31574 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31575 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31577 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31578 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31580 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31583 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31585 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31586 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31587 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31588 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31589 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31590 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31591 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31592 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31593 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31594 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31595 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31596 (__builtin_ia32_mulss_mask_round): New builtins.
31597 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31598 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31599 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31601 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31602 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31604 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31607 2017-05-10 Julia Koval <julia.koval@intel.com>
31609 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31610 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31611 (_mm256_setr_m128i): New intrinsics.
31613 2017-05-10 Julia Koval <julia.koval@intel.com>
31615 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31616 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31617 (_mm_maskz_rcp14_ss): New intrinsics.
31618 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31619 (__builtin_ia32_rcp14ss_mask): New builtins.
31620 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31622 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31624 PR tree-optimization/51513
31625 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31626 (assert_unreachable_fallthru_edge_p): Use it.
31627 (group_case_labels_stmt): Likewise.
31628 * tree-cfg.h: Prototype it.
31629 * stmt.c: Include cfghooks.h and tree-cfg.h.
31630 (emit_case_dispatch_table) <gap_label>: New local variable.
31631 Use it to fill dispatch table gaps.
31632 Test for default_label before updating probabilities.
31633 (expand_case) <default_label>: Remove unneeded initialization.
31634 Test for unreachable default case statement and remove its edge.
31635 Set default_label accordingly.
31636 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31638 2017-05-10 Carl Love <cel@us.ibm.com>
31640 * config/rs6000/rs6000-c: Add support for built-in functions
31641 vector signed char vec_neg (vector signed char)
31642 vector signed short int vec_neg (vector short int)
31643 vector signed int vec_neg (vector signed int)
31644 vector signed long long vec_neg (vector signed long long)
31645 vector float vec_neg (vector float)
31646 vector double vec_neg (vector double)
31647 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31649 * config/rs6000/altivec.h: Add define for vec_neg
31650 * doc/extend.texi: Update the built-in documentation for the
31651 new built-in functions.
31653 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31655 PR tree-optimization/77644
31656 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31658 2017-05-10 Nathan Sidwell <nathan@acm.org>
31660 * dumpfile.h (TDI_lang_all): New.
31661 (TDF_KIND): New. Renumber others
31662 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31664 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31666 (get_dump_file_name): Adjust suffix generation.
31667 (dump_enable_all): Use TDF_KIND.
31668 * doc/invoke.texi (-fdump-lang-all): Document.
31670 * dumpfile.h: Tabify.
31672 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31675 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31676 Move member access before delete.
31678 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31680 * tree-inline.c (expand_call_inline): Split block at stmt
31683 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31686 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31687 are now unused after splitting mov{sf,sd}_hardfloat.
31688 (f32_lr2): Likewise.
31689 (f32_lm): Likewise.
31690 (f32_lm2): Likewise.
31691 (f32_li): Likewise.
31692 (f32_li2): Likewise.
31693 (f32_lv): Likewise.
31694 (f32_sr): Likewise.
31695 (f32_sr2): Likewise.
31696 (f32_sm): Likewise.
31697 (f32_sm2): Likewise.
31698 (f32_si): Likewise.
31699 (f32_si2): Likewise.
31700 (f32_sv): Likewise.
31701 (f32_dm): Likewise.
31702 (f32_vsx): Likewise.
31703 (f32_av): Likewise.
31704 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31705 For movsf, order stores so the VSX stores occur before the GPR
31706 store which encourages the register allocator to use a traditional
31707 FPR instead of a GPR. For movsd, order the stores so that the GPR
31708 store comes before the VSX stores to allow the power6 to work.
31709 This is due to the power6 not having a 32-bit integer store
31710 instruction from a FPR.
31711 (movsf_hardfloat): Likewise.
31712 (movsd_hardfloat): Likewise.
31714 2017-05-09 Martin Sebor <msebor@redhat.com>
31716 PR translation/80280
31717 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31720 PR translation/80280
31721 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31722 data member added in r247778.
31723 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31725 2017-05-09 Nathan Sidwell <nathan@acm.org>
31727 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31729 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31732 2017-05-09 Marek Polacek <polacek@redhat.com>
31734 * doc/invoke.texi: Fix typo.
31736 2017-05-09 Richard Biener <rguenther@suse.de>
31738 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31739 (vrp_val_is_min): Likewise.
31740 (set_value_range_to_value): Likewise.
31741 (set_value_range_to_nonnegative): Likewise.
31742 (gimple_assign_nonzero_p): Likewise.
31743 (gimple_stmt_nonzero_p): Likewise.
31744 (vrp_int_const_binop): Likewise. Remove unreachable case.
31745 (adjust_range_with_scev): Adjust comments.
31746 (compare_range_with_value): Likewise.
31747 (extract_range_from_phi_node): Likewise.
31748 (test_for_singularity): Likewise.
31750 2017-05-09 Richard Biener <rguenther@suse.de>
31752 * tree-vrp.c (get_single_symbol): Add assert that we don't
31753 get overflowed constants as invariant part.
31754 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31755 checks. Use wi::cmp instead of recursing for integer constants.
31756 (compare_values): Just ignore whether we assumed undefined
31757 overflow instead of failing the compare.
31758 (extract_range_for_var_from_comparison_expr): Add comment before the
31759 TREE_NO_WARNING sets.
31760 (test_for_singularity): Likewise.
31761 (extract_range_from_comparison): Do not disable optimization
31762 when we assumed undefined overflow.
31763 (extract_range_basic): Remove init of unused var.
31765 2017-05-09 Richard Biener <rguenther@suse.de>
31767 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31768 (extract_range_from_multiplicative_op_1): Adjust.
31769 (extract_range_from_binary_expr_1): Use int_const_binop.
31771 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31774 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31775 rs6000_store_data_bypass_p in seven define_bypass directives and
31776 in several comments.
31777 * config/rs6000/rs6000-protos.h: Add prototype for
31778 rs6000_store_data_bypass_p function.
31779 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31780 function implements slightly different (rs6000-specific) semantics
31781 than store_data_bypass_p, returning false rather than aborting
31782 with assertion error when arguments do not satisfy the
31783 requirements of store data bypass.
31784 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31785 rs6000_store_data_bypass_p.
31787 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31789 * config/xtensa/xtensa-protos.h
31790 (xtensa_initial_elimination_offset): New declaration.
31791 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31792 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31793 macro definition, add case for FRAME_POINTER_REGNUM when
31794 FRAME_GROWS_DOWNWARD.
31795 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31796 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31797 xtensa_initial_elimination_offset.
31799 2017-05-08 Nathan Sidwell <nathan@acm.org>
31801 * doc/invoke.texi: Alphabetize -fdump options.
31803 2017-05-08 Martin Sebor <msebor@redhat.com>
31805 PR translation/80280
31806 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31808 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31810 * target.def (compute_frame_layout): New optional target hook.
31811 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31812 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31813 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31815 * reload1.c (verify_initial_elim_offsets): Likewise.
31816 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31817 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31818 (arm_get_frame_offsets): Split up into this ...
31819 (arm_compute_frame_layout): ... and this function.
31821 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31823 * config/aarch64/constraints.md (Usa): New constraint.
31824 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31826 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31828 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31829 with_multilib_list after it has been checked.
31831 2017-05-08 Richard Biener <rguenther@suse.de>
31833 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31834 (bitmap_set_subtract_values): Likewise.
31836 2017-05-08 Richard Biener <rguenther@suse.de>
31838 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31839 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31841 (gimple_stmt_nonzero_warnv_p): Rename to ...
31842 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31844 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31845 (extract_range_basic): Adjust, do not disable propagation on
31846 strict overflow sensitive simplification.
31847 (vrp_visit_cond_stmt): Likewise.
31849 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31851 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31852 body size unconditionally.
31854 2017-05-07 Jeff Law <law@redhat.com>
31857 2017-05-06 Jeff Law <law@redhat.com>
31858 PR tree-optimization/78496
31859 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31862 PR tree-optimization/78496
31863 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31864 (simplify_stmt_using_ranges): Call it.
31865 (vrp_dom_walker::before_dom_children): Extract equivalences
31866 from an ASSERT_EXPR with an equality comparison against a
31869 2017-05-06 Jeff Law <law@redhat.com>
31871 PR tree-optimization/78496
31872 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31875 PR tree-optimization/78496
31876 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31877 (simplify_stmt_using_ranges): Call it.
31878 (vrp_dom_walker::before_dom_children): Extract equivalences
31879 from an ASSERT_EXPR with an equality comparison against a
31882 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31884 * lra-constraints.c (lra_copy_reg_equiv): New function.
31885 (split_reg): Use it to copy equivalence information from the
31886 original register to the spill register.
31888 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31890 PR rtl-optimization/75964
31891 * simplify-rtx.c (simplify_const_relational_operation): Remove
31892 invalid handling of comparisons of integer ABS.
31894 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31896 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31897 initialize to zero.
31898 (init_regs): Remove declaration.
31899 (function_arg_advance_32): Initialize error_p as boolean variable.
31901 2017-05-05 Nathan Sidwell <nathan@acm.org>
31903 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31904 lines. Use for (;;).
31906 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31908 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31909 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31910 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31911 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31912 VF=2 that require versioning.
31914 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31916 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31919 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31921 * diagnostic.h (diagnostic_override_option_index): Convert from
31922 macro to inline function.
31924 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31926 * diagnostic.c (last_module_changed_p): New function.
31927 (set_last_module): New function.
31928 (diagnostic_report_current_module): Convert macro usage to
31929 the above functions.
31930 * diagnostic.h (diagnostic_context::last_module): Strengthen
31931 from const line_map * to const line_map_ordinary *.
31932 (diagnostic_last_module_changed): Delete macro.
31933 (diagnostic_set_last_module): Delete macro.
31935 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31937 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31938 with diagnostic_report_diagnostic.
31939 (diagnostic_n_impl_richloc): Likewise.
31940 * diagnostic.h (report_diagnostic): Delete macro.
31941 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31942 with diagnostic_report_diagnostic.
31943 * substring-locations.c (format_warning_va): Likewise.
31945 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31947 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31948 save/restor of format_spec. Move option-printing code to...
31949 (print_option_information): ...this new function, and
31950 reimplement by simply printing to the pretty_printer,
31951 rather than appending to the format string.
31953 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31955 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31956 handling logic into...
31957 (update_effective_level_from_pragmas): ...this new function.
31959 2017-05-04 Andrew Waterman <andrew@sifive.com>
31961 * config/riscv/riscv.opt (mstrict-align): New option.
31962 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31963 (SLOW_UNALIGNED_ACCESS): Define.
31964 (riscv_slow_unaligned_access): Declare.
31965 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31967 (riscv_slow_unaligned_access): New variable.
31968 (rocket_tune_info): Set slow_unaligned_access to true.
31969 (optimize_size_tune_info): Set slow_unaligned_access to false.
31970 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31971 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31972 (riscv_option_override): Set riscv_slow_unaligned_access.
31973 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31975 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31977 * config/riscv/riscv.md: Unify indentation.
31979 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31984 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31986 (extendsi<mode>2): Add support for doing sign extension via
31987 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31988 don't have ISA 3.0 instructions.
31989 (extendsi<mode>2 splitter): Likewise.
31990 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31991 generate the normal insns since SImode can now go in vector
31992 registers. Disallow the special UNSPECs needed for previous
31993 machines to hide SImode being used. Add new insns
31994 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31995 (fix_trunc<mode>si2_stfiwx): Likewise.
31996 (fix_trunc<mode>si2_internal): Likewise.
31997 (fixuns_trunc<mode>si2): Likewise.
31998 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31999 (fctiw<u>z_<mode>_smallint): Likewise.
32000 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32001 of floating point to 32-bit integer from doing a direct move to
32002 the GPR registers to do a store.
32003 (fctiwz_<mode>): Break long line.
32005 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32007 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32008 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32009 (addr_list, addr_offset_valid_p): New.
32010 (split_address_groups): Check offset validity with above function.
32011 (gt-tree-ssa-loop-ivopts.h): Include header file.
32013 2017-05-05 Nathan Sidwell <nathan@acm.org>
32015 * config.gcc (arm*-*-*): Add missing 'fi'.
32017 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32019 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32020 included in -fopt-info does not matter.
32021 * doc/optinfo.texi (-fopt-info): Fix description of default
32022 behavour. Explicitly say order of options included in -fopt-info
32025 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32027 * config.gcc: Allow combinations of aprofile and rmprofile values for
32028 --with-multilib-list.
32029 * config/arm/t-multilib: New file.
32030 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32031 variables. Remove setting of ISA and floating-point ABI in
32032 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32033 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32034 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32035 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32037 * config/arm/t-rmprofile: Likewise except for the matches changes.
32038 * doc/install.texi (--with-multilib-list): Document the combination of
32039 aprofile and rmprofile values and warn about pitfalls in doing that.
32041 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32043 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32044 (movdi_aarch64): Likewise.
32046 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32048 PR tree-optimization/80632
32049 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32051 (build_arrays): Initialize it for virtual phis.
32052 (fix_phi_nodes): Use it for virtual phis.
32054 PR tree-optimization/80558
32055 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32056 [x, y] op z into [x op, y op z] for op & or | if conditions
32059 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32060 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32063 * config/arm/arm.md (use_literal_pool): Remove.
32064 (64-bit immediate split): No longer takes cost into consideration
32065 if arm_disable_literal_pool is enabled.
32066 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32067 used when arm_disable_literal_pool is enabled.
32068 (arm_max_const_double_inline_cost): Remove use of
32069 arm_disable_literal_pool.
32070 (push_minipool_fix): Add assert.
32071 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32072 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32073 (no_literal_pool_sf_immediate): New.
32075 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32077 PR tree-optimization/80613
32078 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32079 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32081 2017-05-05 Richard Biener <rguenther@suse.de>
32083 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32085 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32087 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32088 of this flag from insn conditions due to removal from r247495.
32090 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32092 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32094 (arm_early_store_addr_dep_ptr): Likewise.
32095 * config/arm/aarch-common-protos.h
32096 (arm_early_load_addr_dep_ptr): Add prototype.
32097 (arm_early_store_addr_dep_ptr): Likewise.
32098 * config/arm/cortex-a53.md: Add new bypasses.
32100 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32102 * tree.c (next_type_uid): Change type to unsigned.
32103 (type_hash_canon): Decrement back next_type_uid if
32104 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32105 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32108 2017-05-04 Martin Sebor <msebor@redhat.com>
32110 * builtins.c: Fix a trivial typo in a comment.
32112 PR middle-end/79234
32113 * builtins.c (check_sizes): Adjust to handle reading past the end.
32114 Avoid printing excessive upper bound of ranges. Use %E to print
32115 tree nodes instead of converting them to %wu.
32116 (expand_builtin_memchr): New function.
32117 (compute_dest_size): Rename...
32118 (compute_objsize): ...to this.
32119 (expand_builtin_memcpy): Adjust.
32120 (expand_builtin_mempcpy): Adjust.
32121 (expand_builtin_strcat): Adjust.
32122 (expand_builtin_strcpy): Adjust.
32123 (check_strncat_sizes): Adjust.
32124 (expand_builtin_strncat): Adjust.
32125 (expand_builtin_strncpy): Adjust and simplify.
32126 (expand_builtin_memset): Adjust.
32127 (expand_builtin_bzero): Adjust.
32128 (expand_builtin_memcmp): Adjust.
32129 (expand_builtin): Handle memcmp.
32130 (maybe_emit_chk_warning): Check strncat just once.
32132 2017-05-04 Martin Sebor <msebor@redhat.com>
32134 PR preprocessor/79214
32135 PR middle-end/79222
32136 PR middle-end/79223
32137 * builtins.c (check_sizes): Add inlining context and issue
32138 warnings even when -Wno-system-headers is set.
32139 (check_strncat_sizes): Same.
32140 (expand_builtin_strncat): Same.
32141 (expand_builtin_memmove): New function.
32142 (expand_builtin_stpncpy): Same.
32143 (expand_builtin): Handle memmove and stpncpy.
32145 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32147 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32148 which is not used any more.
32150 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32152 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32154 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32156 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32157 (cortexa53_tunings): Likewise.
32158 (cortexa57_tunings): Likewise.
32159 (cortexa72_tunings): Likewise.
32160 (cortexa73_tunings): Likewise.
32162 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32164 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32165 Set loop alignment to 8.
32167 2017-05-04 Martin Sebor <msebor@redhat.com>
32169 PR translation/80280
32170 * builtins.c (expand_builtin_object_size): Add missing quoting to
32171 %D and like directives.
32172 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32173 (hsa_type_for_tree_type): Same.
32174 (verify_function_arguments): Same.
32175 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32176 * varasm.c (get_section): Same.
32179 2017-05-04 Martin Sebor <msebor@redhat.com>
32181 PR translation/80280
32182 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32184 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32186 * config/aarch64/aarch64.c (generic_addrcost_table):
32187 Change HI/TI mode setting.
32189 2017-05-04 Martin Jambor <mjambor@suse.cz>
32191 PR tree-optimization/80622
32192 * tree-sra.c (comes_initialized_p): New function.
32193 (build_accesses_from_assign): Only set write lazily when
32194 comes_initialized_p is false.
32195 (analyze_access_subtree): Use comes_initialized_p.
32196 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32197 instead of testing for PARM_DECL.
32199 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32201 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32202 constraint on operand 0 to allow more general addressing modes.
32203 Adjust output template.
32204 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32206 * config/aarch64/aarch64-protos.h
32207 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32208 * config/aarch64/constraints.md (Dp): New address constraint.
32209 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32212 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32214 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32215 update use of estimate_ipcp_clone_size_and_time.
32216 (estimate_local_effects): Update use of
32217 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32218 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32219 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32220 Return nonspecialized time.
32222 2017-05-04 Richard Biener <rguenther@suse.de>
32224 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32225 for the last VUSE which def dominates the PHI. Directly call
32227 (get_continuation_for_phi_1): Remove.
32229 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32231 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32232 to explain the use of truncating division. Cap the number of
32233 iterations to the maximum given by nb_iterations_upper_bound,
32236 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32238 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32239 * configure: Regenerate.
32240 * config.in: Regenerate.
32241 * config/i386/driver-mingw32.c: new file.
32242 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32243 * config.host: Link driver-mingw32.o on MinGW host.
32244 * doc/install.texi: Document new --enable-mingw-wildcard configure
32247 2017-05-04 Marek Polacek <polacek@redhat.com>
32249 PR tree-optimization/80612
32250 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32252 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32253 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32255 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32256 (movt splitter): Likewise.
32257 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32258 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32259 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32260 block for Thumb-1 with MOVT.
32261 (thumb2_legitimate_address_p): Move code block ...
32262 (can_avoid_literal_pool_for_label_p): ... into this new function.
32263 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32265 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32266 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32267 "M-profile targets with the MOVT instruction".
32269 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32271 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32272 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32273 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32275 2017-05-04 Martin Liska <mliska@suse.cz>
32277 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32278 variable cond_code.
32280 2017-05-04 Richard Biener <rguenther@suse.de>
32282 * tree.c (array_at_struct_end_p): Handle arrays at struct
32283 end with flexarrays more conservatively. Refactor and treat
32284 arrays of arrays or aggregates more strict. Fix
32285 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32286 * tree.c (array_at_struct_end_p): Adjust prototype.
32287 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32288 * gimple-fold.c (get_range_strlen): Likewise.
32289 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32291 2017-05-04 Richard Biener <rguenther@suse.de>
32293 PR tree-optimization/31130
32294 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32296 (supports_overflow_infinity): Likewise.
32297 (is_negative_overflow_infinity): Likewise.
32298 (is_positive_overflow_infinity): Likewise.
32299 (is_overflow_infinity): Likewise.
32300 (stmt_overflow_infinity): Likewise.
32301 (overflow_infinity_range_p): Likewise.
32302 (usable_range_p): Remove as always returning true.
32303 (make_overflow_infinity): Remove.
32304 (negative_overflow_infinity): Likewise.
32305 (positive_overflow_infinity): Likewise.
32306 (avoid_overflow_infinity): Likewise.
32307 (set_value_range): Adjust accordingly.
32308 (set_value_range_to_nonnegative): Likewise, remove now unused
32309 overflow_infinity arg.
32310 (vrp_operand_equal_p): Adjust.
32311 (update_value_range): Likewise.
32312 (range_int_cst_singleton_p): Likewise.
32313 (operand_less_p): Likewise.
32314 (compare_values_warnv): Likewise.
32315 (extract_range_for_var_from_comparison_expr): Likewise.
32316 (vrp_int_const_binop): Likewise.
32317 (zero_nonzero_bits_from_vr): Likewise.
32318 (extract_range_from_multiplicative_op_1): Likewise.
32319 (extract_range_from_binary_expr_1): Likewise.
32320 (extract_range_from_unary_expr): Likewise.
32321 (extract_range_from_comparison): Likewise.
32322 (extract_range_basic): Likewise.
32323 (adjust_range_with_scev): Likewise.
32324 (compare_ranges): Likewise.
32325 (compare_range_with_value): Likewise.
32326 (dump_value_range): Likewise.
32327 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32329 (simplify_cond_using_ranges): Adjust.
32331 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32333 * brig-builtins.def: Added a builtin for class_f64.
32334 * builtin-types.def: Added a builtin type needed by class_f64.
32336 2017-05-03 Jason Merrill <jason@redhat.com>
32338 * timevar.def: Add TV_CONSTEXPR.
32340 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32342 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32344 2017-05-03 Martin Jambor <mjambor@suse.cz>
32346 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32347 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32348 * ipa-cp.c (ipcp_propagate_stage): Do not call
32349 ipa_update_after_lto_read.
32350 * ipa-inline.c (ipa_inline): Likewise.
32352 2017-05-03 Martin Jambor <mjambor@suse.cz>
32354 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32355 tag. Added a default constructor and a destructor.
32356 (ipa_edge_args_sum_t): New class;
32357 (ipa_edge_args_sum): Declare.
32358 (ipa_edge_args_vector): Remove declaration.
32359 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32360 (ipa_free_edge_args_substructures): Remove declaration.
32361 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32362 (ipa_edge_args_info_available_for_edge_p): Likewise.
32363 * ipa-prop.c (ipa_edge_args_vector): Removed.
32364 (edge_removal_hook_holder): Likewise.
32365 (edge_duplication_hook_holder): Likewise.
32366 (ipa_edge_args_sum): New variable.
32367 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32368 ipa_edge_args_vector.
32369 (ipa_free_edge_args_substructures): Likewise.
32370 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32371 ipa_edge_args_vector.
32372 (ipa_edge_removal_hook): Turned into method
32373 ipa_edge_args_sum_t::remove.
32374 (ipa_edge_duplication_hook): Turned into method
32375 ipa_edge_args_sum_t::duplicate.
32376 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32377 registering edge hooks.
32378 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32379 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32380 ipa_edge_args_sum instead of ipa_edge_args_vector.
32381 * ipa-profile.c (ipa_profile): Likewise.
32383 2017-05-03 Martin Jambor <mjambor@suse.cz>
32385 * symbol-summary.h (function_summary): New method exists.
32386 (function_summary::symtab_removal): Deallocate through release.
32387 (call_summary): New class.
32388 (gt_ggc_mx): New overload.
32389 (gt_pch_nx): Likewise.
32390 (gt_pch_nx): Likewise.
32392 2017-05-03 Jeff Law <law@redhat.com>
32394 PR tree-optimization/78496
32395 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32396 from simplify_cond_using_ranges. Split off code to walk
32397 backwards through casts into ...
32398 (simplify_cond_using_ranges_2): New function.
32399 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32400 (execute_vrp): After identifying jump threads, call
32401 simplify_cond_using_ranges_2.
32403 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32406 * ipa-inline.h (inline_summary): Add ctor.
32407 (create_ggc): Do not use ggc_cleared_alloc.
32409 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32410 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32412 * gcc.c (handle_braces): Support escaping in switch matching
32414 * doc/invoke.texi (Spec Files): Document it.
32415 Remove superfluous @code markup in items.
32417 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32419 * diagnostic-show-locus.c (struct column_range): New struct.
32420 (get_affected_columns): New function.
32421 (get_printed_columns): New function.
32422 (struct correction): New struct.
32423 (correction::ensure_capacity): New function.
32424 (correction::ensure_terminated): New function.
32425 (struct line_corrections): New struct.
32426 (line_corrections::~line_corrections): New dtor.
32427 (line_corrections::add_hint): New function.
32428 (layout::print_trailing_fixits): Reimplement in terms of the new
32430 (selftest::test_overlapped_fixit_printing): New function.
32431 (selftest::diagnostic_show_locus_c_tests): Call it.
32433 2017-05-03 Nathan Sidwell <nathan@acm.org>
32435 Canonicalize canonical type hashing
32436 * tree.h (type_hash_canon_hash): Declare.
32437 * tree.c (type_hash_list, attribute_hash_list): Move into
32438 type_hash_canon_hash.
32439 (build_type_attribute_qual_variant): Break out hash code calc into
32440 type_hash_canon_hash.
32441 (type_hash_canon_hash): New. Generic type hash computation.
32442 (build_range_type_1, build_array_type_1, build_function_type,
32443 build_method_type_directly, build_offset_type, build_complex_type,
32444 make_vector_type): Call it.
32446 2017-05-03 Richard Biener <rguenther@suse.de>
32448 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32449 When all DRs have unknown misaligned do not always peel
32450 when there is a store but apply the same costing model as if
32451 there were only loads.
32453 2017-05-03 Richard Biener <rguenther@suse.de>
32456 PR tree-optimization/80492
32457 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32458 compare_base_decls returning dont-know properly.
32460 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32462 * config/arm/iterators.md (CCSI): New mode iterator.
32463 (arch): New mode attribute.
32464 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32465 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32466 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32467 code iterator for success result mode.
32468 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32469 the corresponding new insn generators.
32471 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32474 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32475 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32477 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32479 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32480 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32481 (DDR_A): Wrap DDR argument in brackets.
32482 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32483 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32484 (DDR_REVERSED_P): Likewise.
32486 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32488 PR tree-optimization/79472
32489 * tree-switch-conversion.c (struct switch_conv_info): Add
32490 contiguous_range and default_case_nonstandard fields.
32491 (collect_switch_conv_info): Compute contiguous_range and
32492 default_case_nonstandard fields, don't clear final_bb if
32493 contiguous_range and only the default case doesn't have the required
32495 (check_all_empty_except_final): Set default_case_nonstandard instead
32496 of failing if contiguous_range and the default case doesn't have empty
32498 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32499 and only the default case doesn't have the required constants. Skip
32501 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32502 if default_case_nonstandard.
32503 (build_constructors): Build constant 1 just once. Assert that default
32504 values aren't inserted in between cases if contiguous_range. Skip
32506 (build_arrays): Skip virtual phis.
32507 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32508 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32509 Handle virtual phis.
32510 (gen_inbound_check): Handle default_case_nonstandard case.
32511 (process_switch): Adjust check_final_bb caller. Call
32512 gather_default_values with the first non-default case instead of
32513 default case if default_case_nonstandard.
32515 2017-05-02 Nathan Sidwell <nathan@acm.org>
32517 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32518 check. Fix formatting.
32520 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32522 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32523 errors when comparing specialized and unspecialized times.
32525 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32527 * diagnostic-show-locus.c
32528 (layout::should_print_annotation_line_p): Make private.
32529 (layout::print_annotation_line): Make private.
32530 (layout::annotation_line_showed_range_p): Make private.
32531 (layout::show_ruler): Make private.
32532 (layout::print_source_line): Make private. Pass in line and
32533 line_width, rather than calling location_get_source_line. Drop
32535 (layout::print_leading_fixits): New method.
32536 (layout::print_any_fixits): Rename to...
32537 (layout::print_trailing_fixits): ...this, and make private.
32538 Don't print newline fixits.
32539 (diagnostic_show_locus): Move logic for printing one row into...
32540 (layout::print_line): ...this new function. Move the
32541 location_get_source_line call and error-handling from
32542 print_source_line to here. Call print_leading_fixits, and rename
32543 print_any_fixits to print_trailing_fixits.
32544 (selftest::test_fixit_insert_containing_newline): Update now that
32545 newlines are partially supported.
32546 (selftest::test_fixit_insert_containing_newline_2): New test.
32547 (selftest::test_fixit_replace_containing_newline): Update comments.
32548 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32549 * edit-context.c (class added_line): New class.
32550 (class edited_line): Describe newline handling in comment.
32551 (edited_line::actually_edited_p): New method.
32552 (edited_line::print_content): Delete redundant decl.
32553 (edited_line::m_predecessors): New field.
32554 (edited_file::print_content): Call edited_line::print_content.
32555 (edited_file::print_diff): Update to support newlines.
32556 (edited_file::print_diff_hunk): Likewise.
32557 (edited_file::print_run_of_changed_lines): New function.
32558 (edited_file::print_diff_line): Convert to...
32559 (print_diff_line): ...this.
32560 (edited_file::get_effective_line_count): New function.
32561 (edited_line::edited_line): Initialize new field m_predecessors.
32562 (edited_line::~edited_line): Clean up m_predecessors.
32563 (edited_line::apply_fixit): Handle newlines.
32564 (edited_line::get_effective_line_count): New function.
32565 (edited_line::print_content): New function.
32566 (edited_line::print_diff_lines): New function.
32567 (selftest::test_applying_fixits_insert_containing_newline): New
32569 (selftest::test_applying_fixits_replace_containing_newline): New
32571 (selftest::insert_line): New function.
32572 (selftest::test_applying_fixits_multiple_lines): Add example of
32574 (selftest::edit_context_c_tests): Call the new tests.
32576 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32578 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32579 parameter cand. Update dump information.
32580 (get_computation_cost): Update uses.
32582 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32584 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32585 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32586 (get_computation_at, rewrite_use_address): Update use of
32587 get_computation_aff.
32589 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32591 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32592 (get_computation): Delete.
32593 (get_computation_cost): Implement like get_computation_cost_at.
32594 Use get_computation_at.
32595 (get_computation_cost_at): Delete.
32596 (rewrite_use_nonlinear_expr): Use get_computation_at.
32597 (rewrite_use_compare, remove_unused_ivs): Ditto.
32599 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32601 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32603 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32605 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32606 (ivopts_global_cost_for_size): Rename parameter and update uses.
32607 (iv_ca_recount_cost): Update uses.
32608 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32609 candidates seperately in n_invs and n_cands.
32610 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32612 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32614 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32615 (find_inv_vars_cb): New.
32616 (find_depends): Renamed to ...
32617 (find_inv_vars): ... this.
32618 (add_candidate_1, force_var_cost): Call find_inv_vars.
32619 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32621 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32623 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32624 inv_vars. Add inv_exprs.
32625 (struct iv_cand): Rename depends_on to inv_vars.
32626 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32627 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32628 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32629 (dump_cand): Dump inv_vars.
32630 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32631 (record_invariant, find_depends, add_candidate_1): Ditto.
32632 (set_group_iv_cost, force_var_cost): Ditto.
32633 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32634 (get_computation_cost_at, get_computation_cost): Ditto.
32635 (determine_group_iv_cost_generic): Ditto.
32636 (determine_group_iv_cost_address): Ditto.
32637 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32638 (determine_group_iv_costs): Ditto.
32639 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32640 (iv_ca_set_remove_invariants): Renamed to ...
32641 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32642 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32643 (iv_ca_set_add_invariants): Renamed to ...
32644 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32645 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32646 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32647 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32648 (create_new_ivs): Remove useless dump.
32650 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32652 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32654 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32655 (iv_ca_set_no_cp, create_new_iv): Ditto.
32657 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32659 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32661 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32663 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32664 function tree_check2.
32666 2017-05-02 Martin Liska <mliska@suse.cz>
32668 * doc/gcov.texi: Add missing preposition.
32669 * gcov.c (function_info::function_info): Properly fill up
32670 all member variables.
32672 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32674 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32676 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32678 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32680 2017-05-02 Martin Liska <mliska@suse.cz>
32683 * lto-streamer-in.c (lto_read_tree_1): Remove
32684 LTO_STREAMER_DEBUG.
32685 * lto-streamer.c (struct tree_hash_entry): Likewise.
32686 (struct tree_entry_hasher): Likewise.
32687 (tree_entry_hasher::hash): Likewise.
32688 (tree_entry_hasher::equal): Likewise.
32689 (lto_streamer_init): Likewise.
32690 (lto_orig_address_map): Likewise.
32691 (lto_orig_address_get): Likewise.
32692 (lto_orig_address_remove): Likewise.
32693 * lto-streamer.h: Likewise.
32694 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32695 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32697 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32699 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32700 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32701 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32702 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32703 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32704 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32705 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32706 (mm_maskz_sub_ss): New intrinsics.
32707 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32708 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32709 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32710 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32711 (__builtin_ia32_subss_mask_round): New builtins.
32712 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32713 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32714 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32716 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32717 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32719 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32722 2017-05-02 Martin Jambor <mjambor@suse.cz>
32724 PR tree-optimization/78687
32725 * tree-sra.c (access): New field parent.
32726 (process_subtree_disqualification): New function.
32727 (disqualify_candidate): Call it.
32728 (build_accesses_from_assign): Reset write flag if creating an
32730 (build_access_subtree): Fill in parent field and also prpagate
32731 down grp_write flag.
32732 (create_artificial_child_access): New parameter set_grp_write, set
32733 grp_write to its value.
32734 (propagate_subaccesses_across_link): Also propagate grp_write flag
32736 (propagate_all_subaccesses): Push the closest parent back to work
32737 queue if add_access_to_work_queue returned true.
32739 2017-05-02 Richard Biener <rguenther@suse.de>
32741 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32742 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32743 -fstrict-overflow documentation.
32744 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32745 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32746 flag_strict_overflow.
32747 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32748 * lto-opts.c (lto_write_options): Do not stream it.
32749 * lto-wrapper.c (merge_and_complain): Do not handle it.
32750 * opts.c (default_options_table): Do not set -fstrict-overflow.
32751 (finish_options): Likewise do not clear it when sanitizing.
32752 * simplify-rtx.c (simplify_const_relational_operation): Do not
32753 test flag_strict_overflow.
32755 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32757 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32758 using enabled attribute.
32759 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32760 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32761 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32762 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32763 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32764 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32765 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32766 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32767 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32768 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32770 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32772 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32774 2017-05-02 Richard Biener <rguenther@suse.de>
32776 PR tree-optimization/80591
32778 2017-04-10 Richard Biener <rguenther@suse.de>
32780 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32783 2017-05-02 Richard Biener <rguenther@suse.de>
32785 PR tree-optimization/80549
32786 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32787 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32788 headers if they do not have a preheader.
32790 2017-05-02 Martin Liska <mliska@suse.cz>
32793 * common.opt: Fix typo.
32794 * doc/invoke.texi: Likewise.
32796 2017-05-01 Jan Beulich <jbeulich@suse.com>
32798 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32799 swapping, add (x,x,m,x,n) alternative.
32801 2017-05-01 Nathan Sidwell <nathan@acm.org>
32803 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32804 unnecessary unadjusted_alignment check.
32806 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32809 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32811 * gimplify.c (gimplify_cilk_detach): New function.
32812 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32813 * tree-core.h: Document EXPR_CILK_SPAWN.
32814 * tree.h (EXPR_CILK_SPAWN): Define.
32816 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32818 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32819 to use new fixit_hint representation, using the "replace" logic.
32820 (get_line_span_for_fixit_hint): Likewise.
32821 (layout::print_any_fixits): Likewise.
32822 (selftest::test_one_liner_many_fixits): Rename to...
32823 (selftest::test_one_liner_many_fixits_1): ...this, and update
32824 comment and expected output to reflect that the multiple fix-it
32825 hints are now consolidated into one insertion.
32826 (selftest::test_one_liner_many_fixits_2): New test.
32827 (selftest::test_diagnostic_show_locus_one_liner): Update for
32829 (selftest::test_fixit_consolidation): Update for fix-it API
32831 * diagnostic.c (print_parseable_fixits): Likewise.
32832 * edit-context.c (edited_line::m_line_events): Convert from
32833 auto_vec <line_event *> to auto_vec <line_event>.
32834 (class line_event): Convert from abstract base class to a concrete
32835 class, taking over the role of replace_event.
32836 (class insert_event): Delete.
32837 (class replace_event): Rename to class line_event. Convert to
32839 (edit_context::add_fixits): Reimplement.
32840 (edit_context::apply_insert): Delete.
32841 (edit_context::apply_replace): Rename to...
32842 (edit_context::apply_fixit): ...this. Convert to half-open range.
32843 (edited_file::apply_insert): Delete.
32844 (edited_file::apply_replace): Rename to...
32845 (edited_file::apply_fixit): ...this.
32846 (edited_line::~edited_line): Drop deletion of events.
32847 (edited_line::apply_insert): Delete.
32848 (edited_line::apply_replace): Rename to...
32849 (edited_line::apply_fixit): ...this. Convert to half-open range.
32850 Update for change to type of m_line_events.
32851 * edit-context.h (edit_context::apply_insert): Delete.
32852 (edit_context::apply_replace): Rename to...
32853 (edit_context::apply_fixit): ...this.
32855 2017-05-01 Martin Sebor <msebor@redhat.com>
32857 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32860 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32863 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32864 __get_cpuid_max returns 0.
32865 (__get_cpuid_count): Ditto.
32867 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32869 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32870 replacement expression is another instance of one of its arguments.
32872 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32875 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32876 check for stack push/pop autoinc.
32877 * config/i386/i386.c (ix86_agi_dependent): Return false
32878 if the only reason why modified_in_p returned true is that
32879 addr is SP based and set_insn is a push or pop.
32881 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32883 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32886 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32889 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32890 (account_size_time): Use two predicates - exec_pred and
32892 (evaluate_conditions_for_known_args): Compute both clause and
32894 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32895 (inline_summary_t::duplicate): Update.
32896 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32898 (compute_inline_parameters): Likewise.
32899 (estimate_edge_size_and_time): Update caluclation of time.
32900 (estimate_node_size_and_time): Compute both time and nonspecialized
32902 (estimate_ipcp_clone_size_and_time): Update.
32903 (inline_merge_summary): Update.
32904 (do_estimate_edge_time): Update.
32905 (do_estimate_edge_size): Update.
32906 (do_estimate_edge_hints): Update.
32907 (inline_read_section, inline_write_summary): Stream both new predicates.
32908 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32910 (compute_inlined_call_time): Cleanup.
32911 (big_speedup_p): Update.
32912 (edge_badness): Update.
32913 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32914 (size_time_entry): Replace predicate by exec_predicate and
32915 nonconst_predicate.
32916 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32917 (estimate_edge_time): Return also nonspec_time.
32918 (reset_edge_growth_cache): Update.
32920 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32922 PR rtl-optimization/80491
32923 * ifcvt.c (noce_process_if_block): When looking for x setter
32924 with missing else_bb, don't check only the insn right before
32925 cond_earliest, but look for the last insn that x is modified in
32926 within the same bb.
32928 PR rtl-optimization/80491
32929 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32931 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32933 PR tree-optimization/80487
32934 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32936 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32938 PR tree-optimization/79697
32939 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32940 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32941 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32943 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32944 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32946 2017-04-28 Martin Sebor <msebor@redhat.com>
32948 PR tree-optimization/80523
32949 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32950 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32952 (maybe_warn, format_directive, parse_directive): Use new functions.
32953 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32955 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32957 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32959 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32961 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32962 target_header_dir): Set correctly.
32963 * configure: Regenerated.
32964 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32965 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32966 instead of SYSTEM_HEADER_DIR.
32968 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32970 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32971 (estimate_local_effects): Likewise.
32972 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32973 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32974 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32975 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32976 do_estimate_edge_time, estimate_edge_time): Likewise.
32977 * ipa-inline-analysis.c (estimate_node_size_and_time,
32978 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32979 (estimate_time_after_inlining): Remove.
32981 2017-04-28 Martin Liska <mliska@suse.cz>
32983 * doc/gcov.texi: Enhance documentation of gcov.
32985 2017-04-28 Martin Liska <mliska@suse.cz>
32987 * doc/gcov.texi: Sort options in alphabetic order.
32988 * doc/gcov-dump.texi: Likewise.
32989 * doc/gcov-tool.texi: Likewise.
32990 * gcov.c (print_usage): Likewise.
32991 * gcov-dump.c (print_usage): Likewise.
32992 * gcov-tool.c (print_merge_usage_message): Likewise.
32993 (print_rewrite_usage_message): Likewise.
32994 (print_overlap_usage_message): Likewise.
32996 2017-04-28 Martin Liska <mliska@suse.cz>
32998 PR gcov-profile/53915
32999 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33001 2017-04-28 Martin Liska <mliska@suse.cz>
33003 PR gcov-profile/79891
33004 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33005 is marked by compiler as living on a line.
33006 (get_cycles_count): Remove usage of the union.
33007 (output_intermediate_file): Likewise.
33008 (find_source): Fix GNU coding style.
33009 (accumulate_line_counts): Remove old non-all block mode.
33010 (output_lines): Remove usage of the union.
33011 * profile.c (output_location): Include all BBs, even if
33012 belonging to a same line (and file) as a previous BB.
33014 2017-04-28 Martin Liska <mliska@suse.cz>
33016 * gcov.c (process_args): Handle new argument 'w'.
33017 (read_graph_file): Assign ID to BBs.
33018 (output_branch_count): Display BB # if verbose flag is set.
33019 (output_lines): Likewise for arcs.
33020 (print_usage): Add '--verbose' option help.
33021 * doc/gcov.texi: Document --verbose (-w) option.
33023 2017-04-28 Martin Liska <mliska@suse.cz>
33025 * gcov.c (struct block_location_info): New struct.
33026 (process_file): Fill up the new structure.
33027 (read_graph_file): Replace usage of encoding by the newly added
33029 (add_line_counts): Likewise.
33030 (accumulate_line_counts): Remove usage of the union.
33031 (function_info::function_info): New function.
33032 (function_info::~function_info): Likewise.
33033 (process_file): Call delete instead of release_function.
33034 (release_function): Release the function.
33035 (release_structures): Call delete instead of release_function.
33036 (solve_flow_graph): Replace usage of num_blocks.
33037 (find_exception_blocks): Likewise.
33038 (output_lines): Fix GNU coding style.
33040 2017-04-28 Martin Liska <mliska@suse.cz>
33043 * coverage.c (coverage_remove_note_file): New function.
33044 * coverage.h: Declare the function.
33045 * toplev.c (finalize): Clean if an error has been seen.
33047 2017-04-28 Martin Liska <mliska@suse.cz>
33049 PR gcov-profile/80031
33050 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33051 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33052 * gcov.c (read_graph_file): Read just number of blocks.
33053 * profile.c (branch_prob): Do not stream 0 flags per a basic
33056 2017-04-28 Martin Liska <mliska@suse.cz>
33058 * gcov-dump.c (tag_*): Add new argument to declarations.
33059 (dump_gcov_file): Likewise.
33060 (tag_blocks): Add and use new argument depth.
33061 (tag_arcs): Likewise.
33062 (tag_lines): Likewise.
33063 (tag_counters): Likewise.
33064 (tag_summary): Likewise.
33065 (dump_working_sets): Use depth to do a proper indentation.
33067 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33070 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33071 * symtab.c (symtab_node::debug_symtab): Move definition here.
33073 2017-04-28 Richard Biener <rguenther@suse.de>
33075 * lto-streamer.h (LTO_major_version): Bump to 7.
33077 2017-04-28 Richard Biener <rguenther@suse.de>
33079 * tree-vrp.c (assert_info): New struct.
33080 (add_assert_info): New helper.
33081 (register_edge_assert_for_2): Refactor to add asserts to a vector
33083 (register_edge_assert_for_1): Likewise.
33084 (register_edge_assert_for): Likewise.
33085 (finish_register_edge_assert_for): New helper actually registering
33086 asserts where live on edge.
33087 (find_conditional_asserts): Adjust.
33088 (find_switch_asserts): Likewise.
33089 (evrp_dom_walker::try_find_new_range): Generalize.
33090 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33092 2017-04-27 Marek Polacek <polacek@redhat.com>
33095 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33096 arg10 and arg11 to itype.
33098 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33100 * doc/extend.texi (Object Size Checking): Improve grammar.
33102 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33105 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33106 that the logic for permitting reciprocal estimates matches that
33109 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33112 * tree.c (type_cache_hasher::equal): Only compare
33113 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33114 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33115 non-aggregate element types.
33116 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33117 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33119 2017-04-27 Richard Biener <rguenther@suse.de>
33121 PR middle-end/80533
33122 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33123 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33124 keeping a reference to a trailing array.
33126 2017-04-27 Richard Biener <rguenther@suse.de>
33128 PR middle-end/80539
33129 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33130 being in loop-closed SSA form conservatively.
33131 (chrec_fold_multiply_poly_poly): Likewise.
33133 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33135 PR middle-end/79665
33136 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33137 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33139 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33142 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33143 (aarch64_function_arg_alignment): Return unsigned int again, but still
33144 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33145 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33146 Don't emit -Wpsabi note.
33147 (aarch64_function_arg_boundary): Likewise.
33148 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33151 2017-04-26 Nathan Sidwell <nathan@acm.org>
33153 * tree.h (crc32_unsigned_n): Declare.
33154 (crc32_unsigned, crc32_unsigned): Make inline.
33155 * tree.c (crc32_unsigned_bits): Replace with ...
33156 (crc32_unsigned_n): ... this.
33157 (crc32_unsigned, crc32_byte): Remove.
33158 (crc32_string): Remove unnecessary braces.
33160 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33162 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33163 * ipa-inline-analysis.c (MAX_TIME): Remove.
33164 (account_size_time): Use sreal for time.
33165 (dump_inline_summary): Update.
33166 (estimate_function_body_sizes): Update.
33167 (estimate_edge_size_and_time): Update.
33168 (estimate_calls_size_and_time): Update.
33169 (estimate_node_size_and_time): Update.
33170 (inline_merge_summary): Update.
33171 (inline_update_overall_summary): Update.
33172 (estimate_time_after_inlining): Update.
33173 (inline_read_section): Update.
33174 (inline_write_summary): Update.
33175 * ipa-inline.c (compute_uninlined_call_time): Update.
33176 (compute_inlined_call_time): Update.
33177 (recursive_inlining): Update.
33178 (inline_small_functions): Update.
33179 (dump_overall_stats): Update.
33180 * ipa-inline.h: Include sreal.h.
33181 (size_time_entry): Turn time to sreal.
33182 (inline_summary): Turn self_time nad time to sreal.
33184 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33186 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33188 (sreal::stream_out, sreal::stream_in): New.
33189 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33191 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33193 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33196 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33199 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33201 (dimode_scalar_chain::compute_convert_gain): Ditto.
33202 (dimode_scalar_chain::make_vector_copies): Ditto.
33203 (dimode_scalar_chain::convert_reg): Ditto.
33204 (dimode_scalar_chain::convert_insn): Ditto.
33205 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33206 (VI248_AVX512BW_1): New mode iterator.
33207 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33208 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33211 2017-04-25 Martin Sebor <msebor@redhat.com>
33213 PR tree-optimization/80497
33214 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33215 constants are representable in HOST_WIDE_INT.
33216 (parse_directive): Ditto.
33218 2017-04-25 Martin Sebor <msebor@redhat.com>
33221 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33222 (new_zero_array): Adjust signature.
33223 (dom_info::dom_init): Used unsigned rather that size_t.
33224 (dom_info::dom_info): Same.
33226 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33227 Jakub Jelinek <jakub@redhat.com>
33230 * config/arm/arm.c: Include gimple.h.
33231 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33232 returns negative, increment ncrn only if it returned positive.
33233 (arm_needs_doubleword_align): Return int instead of bool,
33234 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33235 members, but if there is any such non-FIELD_DECL
33236 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33237 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33238 returns negative, increment nregs only if it returned positive.
33239 (arm_setup_incoming_varargs): Likewise.
33240 (arm_function_arg_boundary): Emit -Wpsabi note if
33241 arm_needs_doubleword_align returns negative, return
33242 DOUBLEWORD_ALIGNMENT only if it returned positive.
33244 2017-04-25 Marek Polacek <polacek@redhat.com>
33247 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33248 first argument to type.
33250 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33253 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33254 type checks to test for compatibility instead of equality.
33256 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33257 Jakub Jelinek <jakub@redhat.com>
33260 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33262 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33263 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33264 the alignment computation, but return their maximum in warn_alignment.
33265 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33266 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33268 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33269 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33272 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33274 * config/arc/simdext.md (dmpyh): Fix typo.
33276 2017-04-25 Richard Biener <rguenther@suse.de>
33278 PR tree-optimization/80492
33279 * alias.c (compare_base_decls): Handle registers with asm
33280 specification conservatively.
33281 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33282 compare_base_decls returning dont-know properly.
33284 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33286 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33287 (legitimate_offset_address_p): New function.
33288 (arc_legitimate_address_p): Use above function.
33290 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33292 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33294 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33296 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33297 ACCH registers whenever they are available.
33299 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33301 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33302 double regs fix when not used.
33304 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33306 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33308 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33309 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33311 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33313 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33314 when emitting short ADD instructions.
33316 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33318 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33320 (cmpsi_cc_c_insn): Likewise.
33321 (cbranchsi4_scratch): Compute proper instruction length using
33322 compact_hreg_operand.
33323 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33325 2017-04-25 Richard Biener <rguenther@suse.de>
33327 PR middle-end/80509
33328 * passes.c (pass_manager::pass_manager): Initialize
33329 m_name_to_pass_map.
33331 2017-04-25 Richard Biener <rguenther@suse.de>
33333 PR tree-optimization/79201
33334 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33336 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33339 * config/s390/vector.md: Split MEM->GPR vector moves for
33340 non-s_operand addresses.
33342 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33345 * config/s390/predicates.md (reload_const_wide_int_operand): New
33347 * config/s390/s390.md ("movti"): Remove d/P alternative.
33348 ("movti_bigconst"): New pattern definition.
33350 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33353 * s390-protos.h (s390_expand_cs_hqi): Removed.
33354 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33355 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33356 modes as well as CCZ1mode and CCZmode.
33357 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33358 signature of s390_emit_compare_and_swap.
33359 (s390_expand_cs_hqi): Likewise, make static.
33360 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33361 compare-and-swap, in some cases.
33362 (s390_expand_cs): Wrapper function.
33363 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33365 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33366 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33367 patterns for small and large integers. Forbid symref memory operands.
33368 Move expander to s390.c. Require cc register.
33369 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33370 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33371 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33372 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33373 symref memory operands. Remove CC mode and call s390_match_ccmode
33375 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33377 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33379 * config/s390/s390.md (define_peephole2): New peephole to help
33380 combining the load-and-test pattern with volatile memory.
33382 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33384 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33385 with CCZmode for TARGET_Z196.
33387 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33389 PR rtl-optimization/80501
33390 * combine.c (make_compound_operation_int): Set subreg_code to SET
33391 even for AND with mask of the sign bit of mode.
33393 PR rtl-optimization/80500
33394 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33395 sum's initial value.
33397 2017-04-25 Julian Brown <julian@codesourcery.com>
33398 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33400 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33402 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33404 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33406 2017-04-25 Julian Brown <julian@codesourcery.com>
33407 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33409 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33410 (thunderx2t99_sha): New Reservation.
33412 2017-04-25 Julian Brown <julian@codesourcery.com>
33413 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33415 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33416 type for 1-element load.
33418 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33420 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33422 2017-04-24 Martin Jambor <mjambor@suse.cz>
33424 PR tree-optimization/80293
33425 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33426 char arrays not totally scalarizable if it is false.
33427 (analyze_all_variable_accesses): Pass correct value in the new
33428 parameter. Add a statistics counter.
33430 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33432 PR middle-end/79931
33433 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33435 2017-04-24 Richard Biener <rguenther@suse.de>
33437 PR tree-optimization/80494
33438 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33439 out for complex types.
33441 2017-04-24 Richard Biener <rguenther@suse.de>
33443 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33444 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33445 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33446 (DFS): Adjust and never fail.
33447 (sccvn_dom_walker::fail): Remove.
33448 (sccvn_dom_walker::before_dom_children): Adjust.
33449 (run_scc_vn): Likewise and never fail.
33450 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33451 (pass_fre::execute): Likewise.
33453 2017-04-24 Richard Biener <rguenther@suse.de>
33455 PR tree-optimization/79725
33456 * tree-ssa-sink.c (statement_sink_location): Return whether
33457 failure reason was zero uses. Move that check later.
33458 (sink_code_in_bb): Deal with zero uses by removing the stmt
33461 2017-04-24 Richard Biener <rguenther@suse.de>
33464 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33465 pointer-based references.
33467 2017-04-24 Richard Biener <rguenther@suse.de>
33470 * pass_manager.h (pass_manager::operator new): Remove.
33471 (pass_manager::operator delete): Likewise.
33472 * passes.c (pass_manager::operator new): Remove.
33473 (pass_manager::operator delete): Likewise.
33474 (pass_manager::pass_manager): Zero individual pass members.
33476 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33479 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33480 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33481 Check "XEXP (src, 1)" operand here.
33482 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33483 Check "XEXP (src, 1)" operand here.
33484 (dimode_scalar_chain::make_vector_copies): Detect count register
33485 of a shift instruction. Zero extend count register from QImode
33486 to DImode to satisfy vector shift pattern count operand predicate.
33487 Substitute vector shift count operand with a DImode copy.
33488 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33491 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33493 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33494 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33495 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33496 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33497 (UNSPEC_NOREX_MEM): Remove definition.
33499 2017-04-21 Richard Biener <rguenther@suse.de>
33501 PR tree-optimization/79547
33502 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33503 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33504 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33505 without any constraints.
33507 2017-04-21 Richard Biener <rguenther@suse.de>
33509 PR tree-optimization/78847
33510 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33512 2017-04-21 Richard Biener <rguenther@suse.de>
33514 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33515 (build_distinct_type_copy): Likewise.
33516 (build_variant_type_copy): Likewise.
33517 * tree.c (build_qualified_type): Pass down mem-stat info.
33518 (build_distinct_type_copy): Likewise.
33519 (build_variant_type_copy): Likewise.
33521 2017-04-21 Richard Biener <rguenther@suse.de>
33523 PR tree-optimization/80237
33524 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33526 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33527 for a simplified result.
33529 2016-04-21 Richard Biener <rguenther@suse.de>
33531 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33532 sth as strict as a simple_iv but a chrec without symbols and an
33533 operand defined in the loop we are peeling (and not some subloop).
33534 (propagate_constants_for_unrolling): Propagate all constants.
33536 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33539 * config/i386/i386.c (print_reg): Remove assert for disalowed
33540 regno values, call output_operand_lossage instead.
33542 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33545 * config/i386/constraints.md (Yc): New register constraint.
33546 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33547 Use Yc constraint for alternative 2 of operand 0. Remove
33548 preferred_for_speed attribute.
33550 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33552 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33553 lastprivate clauses in SIMT case.
33555 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33557 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33559 2017-04-20 Richard Biener <rguenther@suse.de>
33561 PR tree-optimization/57796
33562 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33563 as N scalar stores.
33564 (vect_model_load_cost): Cost gathers as N scalar loads.
33566 2017-04-20 Richard Biener <rguenther@suse.de>
33568 * ggc-page.c (ggc_allocated_p): Rename to ...
33569 (safe_lookup_page_table_entry): ... this and return the lookup
33571 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33573 2017-04-20 Richard Biener <rguenther@suse.de>
33575 PR tree-optimization/80453
33576 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33577 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33578 from the conditions.
33579 (vn_phi_eq): Pass them down.
33580 (vn_phi_lookup): Record them.
33581 (vn_phi_insert): Likewise.
33583 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33585 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33586 uninitialized variable warning to avoid buffer overrun.
33588 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33591 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33592 is suppressed for '{ 0 }' in C.
33594 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33596 * BASE-VER: Set to 8.0.0.
33598 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33600 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33601 priority .init_array and .fini_array section with SECTION_NOTYPE
33604 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33606 PR middle-end/80423
33607 * tree.h (build_array_type): Add typeless_storage default argument.
33608 * tree.c (type_cache_hasher::equal): Also compare
33609 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33610 (build_array_type): Add typeless_storage argument, set
33611 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33613 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33614 (build_array_type): Likewise. Add typeless_storage argument.
33616 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33617 Jakub Jelinek <jakub@redhat.com>
33619 PR tree-optimization/80426
33620 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33621 operation on symbolic operands, also compute the overflow for the
33622 invariant part when the operation degenerates into a negation.
33624 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33627 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33628 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33631 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33633 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33636 * config/avr/avr.c (tree.h): Include it.
33637 (cgraph.h): Include it.
33638 (avr_encode_section_info): Don't warn for uninitialized progmem
33639 variable if it's just an alias.
33641 2017-04-19 Richard Biener <rguenther@suse.de>
33644 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33645 when needed by AutoPGO.
33647 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33650 * doc/lto.texi: Remove an extra 'that'.
33652 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33654 PR rtl-optimization/80429
33655 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33656 are only used in debug insns.
33658 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33659 Vladimir Makarov <vmakarov@redhat.com>
33661 * config/sparc/predicates.md (input_operand): Add comment. Return
33662 true for any memory operand when LRA is in progress.
33663 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33665 2017-04-18 Jeff Law <law@redhat.com>
33668 * mips.md ({return,simple_return}_internal): Do not overwrite
33671 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33673 PR tree-optimization/80443
33674 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33675 instead of adding 1, subtract -1 and similarly instead of subtracting
33678 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33680 PR rtl-optimization/80357
33681 * haifa-sched.c (tmp_bitmap): New variable.
33682 (model_recompute): Handle duplicate use records.
33683 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33684 (free_global_sched_pressure_data): Free it.
33686 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33689 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33690 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33691 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33692 instead of SYSTEM_HEADER_DIR.
33694 2017-04-18 Jeff Law <law@redhat.com>
33696 PR middle-end/80422
33697 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33698 predecessors after walking up the insn chain.
33700 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33703 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33704 sizetype type into debug info.
33706 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33709 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33710 unneeded test for TARGET_UPPER_REGS_SF.
33711 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33713 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33716 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33717 instead of gsi_after_labels.
33719 2017-04-18 Jeff Law <law@redhat.com>
33721 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33725 2017-04-13 Jeff Law <law@redhat.com>
33726 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33727 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33729 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33732 * config/avr/avr.c (intl.h): Include it.
33733 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33735 2017-04-18 Martin Liska <mliska@suse.cz>
33737 PR gcov-profile/78783
33738 * gcov-tool.c (gcov_output_files): Validate that destination
33739 file is either removed by the tool or by a user.
33741 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33742 Guy Benyei <guybe@mellanox.com>
33744 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33745 block, and do not negate it, the stored id is already negative.
33747 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33749 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33751 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33754 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33755 masks of options that should be turned off if the VSX vector
33756 options are turned off.
33757 (OTHER_P8_VECTOR_MASKS): Likewise.
33758 (OTHER_VSX_VECTOR_MASKS): Likewise.
33759 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33760 rs6000_disable_incompatible_switches to validate no type switches
33762 (rs6000_incompatible_switch): New function to disallow turning on
33763 other vector options if -mno-vsx, -mno-power8-vector, or
33764 -mno-power9-vector are specified.
33766 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33768 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33770 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33772 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33773 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33774 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33775 (ARG_POINTER_CFA_OFFSET): Likewise.
33777 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33779 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33780 conditions to take advantage of various optimizations.
33782 2017-04-13 Jeff Law <law@redhat.com>
33784 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33785 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33786 (zero_extendsidi2_dext): Likewise.
33788 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33791 * fold-const.c (fold_ternary_loc): Revert
33792 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33795 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33797 PR rtl-optimization/80343
33798 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33799 new scratch pseudo.
33801 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33804 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33805 to ubsan_encode_value.
33807 2017-04-13 Jeff Law <law@redhat.com>
33809 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33810 appearing in DEBUG_INSNs.
33812 2017-04-13 Martin Liska <mliska@suse.cz>
33814 PR gcov-profile/80413
33815 * gcov-io.c (gcov_write_string): Copy to buffer just when
33816 allocated size is greater than zero.
33818 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33821 * dwarf2out.c (decls_for_scope): Ignore declarations of
33822 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33824 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33827 * ipa-visibility.c (non_local_p): Fix typos.
33828 (localize_node): When localizing symbol in same comdat group,
33829 dissolve the group only when we know external symbols are going
33831 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33833 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33835 PR tree-optimization/79390
33836 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33837 order does not result in usable sequence, retry with reversed operand
33843 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33844 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33845 op0 instead of fold_convert_loc (loc, type, arg0).
33847 2017-04-12 Jeff Law <law@redhat.com>
33849 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33850 has a delay slot in the generated code.
33852 * config/cris/cris.md (cris_preferred_reload_class): Return
33853 GENNONACR_REGS rather than GENERAL_REGS.
33855 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33858 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33859 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33860 signedness of the result type.
33862 2017-04-12 Richard Biener <rguenther@suse.de>
33863 Jeff Law <law@redhat.com>
33865 PR tree-optimization/80359
33866 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33867 trim stores to TARGET_MEM_REFs.
33869 2017-04-12 Richard Biener <rguenther@suse.de>
33871 PR tree-optimization/79390
33872 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33873 threading case even more.
33875 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33878 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33879 for quad_address_p for TImode, instead of just not indexed_address.
33881 2017-04-12 Richard Biener <rguenther@suse.de>
33882 Bernd Edlinger <bernd.edlinger@hotmail.de>
33884 PR middle-end/79671
33885 * alias.c (component_uses_parent_alias_set_from): Handle
33886 TYPE_TYPELESS_STORAGE.
33887 (get_alias_set): Likewise.
33888 * tree-core.h (tree_type_common): Add typeless_storage flag.
33889 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33890 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33891 for types containing members with TYPE_TYPELESS_STORAGE.
33892 (place_field): Likewise.
33893 (layout_type): Likewise for ARRAY_TYPE.
33894 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33895 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33896 TYPE_TYPELESS_STORAGE.
33897 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33899 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33902 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33903 first argument to type.
33905 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33909 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33910 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33911 (rs6000_expand_binop_builtin): Likewise.
33912 (rs6000_expand_ternop_builtin): Likewise; also add missing
33913 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33914 vshasigma built-ins.
33915 * doc/extend.texi: Document that vec_xxpermdi's third argument
33916 must be a constant.
33918 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33920 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33921 Use shift_const cost parameter when calculating gain of STV shifts.
33923 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33925 PR rtl-optimization/70478
33926 * lra-constraints.c (process_alt_operands): Check memory for
33927 disfavoring memory insn operand.
33929 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33931 PR middle-end/80100
33932 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33933 left shift in unsigned HOST_WIDE_INT type.
33935 PR rtl-optimization/80385
33936 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33937 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33940 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33941 if they have any depend clauses.
33943 2017-04-11 Martin Liska <mliska@suse.cz>
33946 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33947 * ipa-split.c (split_function): Create a local comdat symbol
33948 if caller is in a comdat group.
33950 2017-04-11 Martin Liska <mliska@suse.cz>
33953 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33956 2017-04-11 Martin Sebor <msebor@redhat.com>
33958 PR middle-end/80364
33959 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33960 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33962 (directive::set_width, directive::set_precision, format_character):
33964 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33967 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33970 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33971 conflict, set target->arch_name instead of target->cpu_name.
33973 2017-04-11 Richard Biener <rguenther@suse.de>
33975 PR tree-optimization/80374
33976 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33977 build_zero_cst, remove fold_convertible_p check again.
33979 2017-04-11 Martin Liska <mliska@suse.cz>
33982 * ubsan.c (instrument_object_size): Do not instrument register
33985 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33988 * config/i386/i386-builtin-types.def
33989 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33990 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33991 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33992 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33993 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33994 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33995 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33996 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33997 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33998 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33999 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34000 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34001 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34002 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34003 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34004 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34005 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34006 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34007 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34008 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34009 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34010 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34011 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34012 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34013 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34014 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34015 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34016 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34017 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34018 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34019 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34020 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34021 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34022 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34023 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34024 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34025 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34026 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34027 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34028 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34029 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34030 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34031 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34032 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34033 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34034 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34036 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34037 flag to second_arg_count, handle 4 argument function type _COUNT
34038 aliases, handle second_arg_count on second argument rather than last.
34040 2017-04-10 Jeff Law <law@redhat.com>
34042 PR tree-optimization/80374
34043 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34044 record anything if we can not convert integer_zero_node to the
34047 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34050 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34051 Enhance special handling given to the TARGET_P9_MINMAX option in
34052 relation to certain other options.
34054 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34056 PR tree-optimization/80153
34057 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34058 remove POINTER_PLUS_EXPR's base part directly, rather than through
34061 2017-04-10 Richard Biener <rguenther@suse.de>
34062 Bin Cheng <bin.cheng@arm.com>
34064 PR tree-optimization/80153
34065 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34066 the first element of pointer type aff_tree. Build result expr in
34068 (add_elt_to_tree): Convert to type unconditionally. Remove other
34069 fold_convert calls.
34070 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34071 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34073 2017-04-10 Richard Biener <rguenther@suse.de>
34075 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34078 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34080 PR rtl-optimization/70478
34081 * lra-constraints.c (curr_small_class_check): New.
34082 (update_and_check_small_class_inputs): New.
34083 (process_alt_operands): Update curr_small_class_check. Disfavor
34084 alternative insn memory operands. Check available regs for small
34087 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34090 * config/mips/mips.opt (-mvirt): Update description.
34091 * doc/invoke.texi (-mvirt): Likewise.
34093 2017-04-10 Richard Biener <rguenther@suse.de>
34095 PR middle-end/80362
34096 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34097 looking for NEGATE_EXPR in -A / -B to A / B folding.
34099 2017-04-10 Martin Liska <mliska@suse.cz>
34101 PR gcov-profile/80224
34102 * gcov.c (print_usage): Fix usage string.
34103 (get_gcov_intermediate_filename): Remove.
34104 (output_gcov_file): Use both for normal and intermediate format.
34105 (generate_results): Do not initialize special file for
34106 intermediate format.
34108 2017-04-10 Richard Biener <rguenther@suse.de>
34110 PR tree-optimization/80304
34111 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34114 2017-04-10 Nathan Sidwell <nathan@acm.org>
34117 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34118 (rs6000_init_builtins): Use it.
34120 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34122 * config/arm/arm.md (<mrc>): Add mode to SET source.
34123 (<mrrc>): Likewise.
34125 2017-04-10 Richard Biener <rguenther@suse.de>
34127 PR middle-end/80344
34128 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34130 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34133 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34134 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34135 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34136 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34137 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34138 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34139 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34140 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34141 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34142 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34143 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34144 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34145 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34146 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34147 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34148 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34149 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34150 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34151 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34152 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34153 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34154 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34155 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34157 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34159 PR rtl-optimization/70478
34160 * lra-constraints.c: Reverse the last patch.
34162 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34164 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34165 Add comment for WCHAR_T.
34167 2017-04-08 Martin Liska <mliska@suse.cz>
34170 2017-04-07 Martin Liska <mliska@suse.cz>
34173 * ipa-split.c (split_function): Add function part to a same comdat
34176 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34179 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34181 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34183 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34184 (vec_pairing): ...to this.
34185 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34186 (rs6000_sched_init): Adjust for name change.
34187 (struct rs6000_sched_context): Likewise.
34188 (rs6000_init_sched_context): Likewise.
34189 (rs6000_set_sched_context): Likewise.
34191 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34197 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34199 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34200 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34201 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34203 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34205 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34207 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34209 PR rtl-optimization/70703
34210 * ira-color.c (update_conflict_hard_regno_costs): Use
34211 int64_t instead of HOST_WIDE_INT.
34213 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34215 PR rtl-optimization/70478
34216 * lra-constraints.c (process_alt_operands): Disfavor alternative
34217 insn memory operands.
34219 2017-04-07 Jeff Law <law@redhat.com>
34221 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34222 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34224 2017-04-07 Martin Liska <mliska@suse.cz>
34227 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34228 Show error message instead of an ICE.
34230 2017-04-07 Martin Liska <mliska@suse.cz>
34233 * ipa-split.c (split_function): Add function part to a same comdat
34236 2017-04-07 Richard Biener <rguenther@suse.de>
34238 PR middle-end/80341
34239 * tree.c (get_unwidened): Also handle ! for_type case for
34241 * convert.c (do_narrow): Split out from ...
34242 (convert_to_integer_1): ... here. Do not pass final truncation
34243 type to get_unwidened for TRUNC_DIV_EXPR.
34245 2017-04-07 Richard Biener <rguenther@suse.de>
34247 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34249 (aff_combination_const): Adjust.
34250 (aff_combination_scale): Likewise.
34251 (aff_combination_add_elt): Likewise.
34252 (aff_combination_add_cst): Likewise.
34253 (aff_combination_convert): Likewise.
34254 (add_elt_to_tree): Likewise. Remove unused argument.
34255 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34257 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34259 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34261 * config/arm/arm.c (arm_default_short_enums): Use
34262 ARM_DEFAULT_SHORT_ENUMS.
34263 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34265 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34268 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34269 members with redundant out-of-class redeclaration.
34271 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34274 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34275 * config/i386/i386.md (*zero_extendsidi2):
34276 Add (?*x,*x) and (?*v,*v) alternatives.
34278 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34281 * config/i386/i386.c (ix86_expand_builtin)
34282 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34283 mode from insn data. Convert operands to insn operand mode.
34284 Copy operands that don't satisfy insn predicate to a register.
34286 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34288 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34291 2017-04-06 Richard Biener <rguenther@suse.de>
34293 PR tree-optimization/80334
34294 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34295 preserve alignment of accesses.
34297 2017-04-06 Richard Biener <rguenther@suse.de>
34299 PR tree-optimization/80262
34300 * tree-sra.c (build_ref_for_offset): Preserve address-space
34302 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34303 Drop useless address-space information on MEM_REF offsets.
34305 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34307 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34309 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34311 PR rtl-optimization/70703
34312 * ira-color.c (update_conflict_hard_regno_costs): Use
34313 HOST_WIDE_INT instead of long.
34315 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34318 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34319 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34321 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34322 for x86_64 target. Handle -m3dnowa option.
34324 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34326 PR rtl-optimization/70703
34327 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34328 (update_conflict_hard_regno_costs): Use long instead of unsigned
34329 arithmetic for cost calculation.
34331 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34332 Bernd Edlinger <bernd.edlinger@hotmail.de>
34335 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34338 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34341 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34342 ptr_mode with Pmode throughout.
34343 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34344 into probe_stack_range and use DImode.
34346 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34349 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34350 call_eh_return is true.
34352 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34354 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34355 Initialize last_match_fntype_index.
34357 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34360 * tree-nvr.c: Include internal-fn.h.
34361 (pass_return_slot::execute): Ignore internal calls without
34364 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34365 Richard Biener <rguenther@suse.de>
34368 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34369 captures used multiple times, except for the last use.
34370 * generic-match-head.c: Include gimplify.h.
34372 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34374 PR tree-optimization/79390
34375 * target.h (struct noce_if_info): Declare.
34376 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34377 * target.def (noce_conversion_profitable_p): New target hook.
34378 * ifcvt.h (struct noce_if_info): New type, moved from ...
34379 * ifcvt.c (struct noce_if_info): ... here.
34380 (noce_conversion_profitable_p): Renamed to ...
34381 (default_noce_conversion_profitable_p): ... this. No longer
34383 (noce_try_store_flag_constants, noce_try_addcc,
34384 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34385 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34386 instead of noce_conversion_profitable_p.
34387 * config/i386/i386.c: Include ifcvt.h.
34388 (ix86_option_override_internal): Don't override
34389 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34390 (ix86_noce_conversion_profitable_p): New function.
34391 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34392 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34393 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34394 * doc/tm.texi: Regenerated.
34396 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34398 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34401 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34404 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34405 instructions for small multiply cores.
34407 2017-04-04 Jeff Law <law@redhat.com>
34409 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34411 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34412 that are not set by the loop over the elements.
34414 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34417 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34418 int mode, convert_modes it to mode as unsigned, otherwise use
34419 lowpart_subreg to mode rather than SImode.
34420 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34421 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34422 Use DImode instead of SImode for the shift count operand.
34423 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34426 2017-04-04 Richard Biener <rguenther@suse.de>
34428 PR middle-end/80281
34429 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34430 arithmetic done for the negate or the plus. Simplify.
34431 (A - (-B) -> A + B): Likewise.
34432 * fold-const.c (split_tree): Make sure to not negate pointers.
34434 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34436 PR rtl-optimization/60818
34437 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34438 a compare of comparisons with the thing compared if this results
34439 in a different machine mode.
34441 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34443 * alias.c (base_alias_check): Fix typo in comment.
34444 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34445 * cgraphunit.c (symbol_table::compile): Likewise.
34446 * collect2.c (maybe_run_lto_and_relink): Likewise.
34447 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34448 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34449 * config/avr/avr.c (avr_map_op_t): Likewise.
34450 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34451 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34452 * config/epiphany/epiphany.md (movcc): Likewise.
34453 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34454 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34456 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34457 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34458 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34459 * config/sh/sh.c (sh_rtx_costs): Likewise.
34460 * fold-const.c (fold_truth_andor): Likewise.
34461 * genautomata.c (collapse_flag): Likewise.
34462 * gengtype.h (struct type::u::s): Likewise.
34463 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34464 * input.c (FORMAT_AMOUNT): Likewise.
34465 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34466 (known_aggs_to_agg_replacement_list): Likewise.
34467 * ipa-inline-analysis.c: Likewise.
34468 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34469 * ipa-polymorphic-call.c
34470 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34471 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34472 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34474 * modulo-sched.c (apply_reg_moves): Likewise.
34475 * omp-expand.c (build_omp_regions_1): Likewise.
34476 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34477 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34478 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34479 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34480 * value-prof.c: Likewise.
34481 * var-tracking.c (val_reset): Likewise.
34483 2017-04-03 Richard Biener <rguenther@suse.de>
34485 PR tree-optimization/80275
34486 * fold-const.c (split_address_to_core_and_offset): Handle
34489 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34491 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34492 descriptors is at least equal to that of functions.
34494 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34496 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34498 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34501 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34502 (mov<IMOD4:mode>): New expander.
34503 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34505 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34507 PR rtl-optimization/79405
34508 * fwprop.c (propagations_left): New variable.
34509 (forward_propagate_into): Decrement it.
34510 (fwprop_init): Initialize it.
34511 (fw_prop): If the variable has reached zero, stop propagating.
34512 (fwprop_addr): Ditto.
34514 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34517 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34518 a FUNCTION_DECL, pass it as decl instead of origin to
34521 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34523 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34526 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34529 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34530 TARGET_VSX_SMALL_INTEGER.
34532 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34534 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34535 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34537 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34539 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34540 extraction from odd-numbered MSA register.
34542 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34544 PR middle-end/80173
34545 * expmed.c (store_bit_field_1): Don't attempt to create
34546 a word subreg out of hard registers wider than word if they
34547 have HARD_REGNO_NREGS of 1 for their mode.
34549 PR middle-end/80163
34550 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34551 conversions to integer types wider than word and pointer.
34554 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34555 (rtx_equal_for_cselib_p): Pass 0 to it.
34556 * cselib.c (cselib_hasher::equal): Likewise.
34557 (rtx_equal_for_cselib_1): Add depth argument. If depth
34558 is 128, don't look up VALUE locs and punt. Increment
34559 depth in recursive calls when walking VALUE locs.
34561 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34563 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34564 (make_gcov_file_name): Use the canonical path name for generating
34566 (read_line): Fix handling of files with ascii null bytes.
34568 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34570 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34571 to initialise a vector register instead
34572 of using a const_int.
34574 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34576 PR translation/80189
34577 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34578 diagnostic messages.
34580 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34583 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34584 (dfp_diex_<mode>): Update mode of operand 1.
34585 * doc/extend.texi (dxex, dxexq): Document change to return type.
34586 (diex, diexq): Document change to argument type.
34588 2017-03-30 Martin Jambor <mjambor@suse.cz>
34591 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34592 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34593 it reflects the signature changes performed at the callee side.
34594 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34595 to cgraph_build_function_type_skip_args.
34596 (build_function_decl_skip_args): Adjust call to the above function.
34598 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34601 * config/i386/sse.md
34602 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34603 register as dest whenever it is a MEM not rtx_equal_p to the
34604 corresponding dup operand, and when forcing into reg move the
34605 reg into the memory afterwards.
34606 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34607 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34608 for the force_reg mode.
34609 (avx512vl_vextractf128<mode>): Use register as dest either
34610 always when a MEM, or when it is a MEM not rtx_equal_p to the
34611 corresponding dup operand, or even not when it is a CONST_VECTOR
34612 depending on the mode and lo vs. hi.
34613 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34615 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34616 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34617 Likewise. Require that operands[2] is even.
34618 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34619 Remove extraneous parens. Require that operands[2] is a multiple
34621 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34622 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34623 disallow memory then.
34625 2017-03-30 Richard Biener <rguenther@suse.de>
34627 PR tree-optimization/77498
34628 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34629 to non-constants over backedges.
34631 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34633 PR rtl-optimization/80233
34634 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34635 as last_combined_insn. Do not test for BARRIER_P separately.
34637 2017-03-29 Andreas Schwab <schwab@suse.de>
34640 * calls.c (prepare_call_address): Convert funexp to Pmode before
34641 copying to temp reg.
34643 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34645 PR tree-optimization/80158
34646 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34647 Handle possible future case of more than one alternate
34649 (replace_rhs_if_not_dup): Likewise.
34650 (replace_one_candidate): Likewise.
34652 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34654 PR rtl-optimization/80193
34655 * ira.c (ira): Do not check allocation for LRA.
34657 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34659 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34660 (nvptx_output_simt_exit): Declare.
34661 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34662 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34663 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34664 (nvptx_declare_function_name): ...here. Emit declaration of local
34665 memory space buffer for omp_simt_enter insn.
34666 (nvptx_output_unisimt_switch): New.
34667 (nvptx_output_softstack_switch): New.
34668 (nvptx_output_simt_enter): New.
34669 (nvptx_output_simt_exit): New.
34670 * config/nvptx/nvptx.h (struct machine_function): New fields
34671 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34672 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34673 (UNSPECV_SIMT_EXIT): Ditto.
34674 (omp_simt_enter_insn): New insn.
34675 (omp_simt_enter): New expansion.
34676 (omp_simt_exit): New insn.
34677 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34679 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34680 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34681 (expand_GOMP_SIMT_EXIT): New.
34682 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34683 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34684 (GOMP_SIMT_EXIT): Ditto.
34685 * target-insns.def (omp_simt_enter): New insn.
34686 (omp_simt_exit): Ditto.
34687 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34689 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34690 (lower_rec_input_clauses): Likewise.
34691 (lower_lastprivate_clauses): Handle SIMT privatization.
34693 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34694 (ompdevlow_adjust_simt_enter): New.
34695 (find_simtpriv_var_op): New.
34696 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34697 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34699 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34700 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34701 (copy_decl_for_dup_finish): Ditto.
34703 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34705 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34708 * config/i386/i386.c (ix86_option_override_internal): Always
34709 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34711 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34713 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34715 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34717 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34718 mark new edge's irreducible flag accordign to it.
34719 (vect_do_peeling): Check loop preheader edge's irreducible flag
34720 and pass it to function slpeel_add_loop_guard.
34722 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34724 PR tree-optimization/80218
34725 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34726 Update block frequencies and counts.
34728 2017-03-28 Richard Biener <rguenther@suse.de>
34730 PR tree-optimization/78644
34731 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34732 of a simplification result we may not use it at all.
34734 2017-03-28 Richard Biener <rguenther@suse.de>
34737 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34738 without arguments, generate default definition of a SSA name.
34740 2017-03-28 Richard Biener <rguenther@suse.de>
34742 PR middle-end/80222
34743 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34744 TYPE_REF_CAN_ALIAS_ALL references.
34745 * fold-const.c (fold_indirect_ref_1): Likewise.
34747 2017-03-28 Martin Liska <mliska@suse.cz>
34750 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34751 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34753 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34754 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34756 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34757 (EXTRA_SPECS): Define.
34758 (SUBTARGET_EXTRA_SPECS): Likewise.
34759 (SUBTARGET_CPP_SPEC): Likewise.
34760 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34761 SUBTARGET_EXTRA_SPECS.
34762 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34764 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34766 * config/arc/simdext.md (vst64_insn): Update pattern.
34767 (vld32wh_insn): Likewise.
34768 (vld32wl_insn): Likewise.
34769 (vld64_insn): Likewise.
34770 (vld32_insn): Likewise.
34772 2017-03-28 Marek Polacek <polacek@redhat.com>
34775 * fold-const.c (fold_comparison): Use protected_set_expr_location
34776 instead of SET_EXPR_LOCATION.
34778 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34780 * tree.c (add_expr): Avoid name lookup warning.
34782 2017-03-27 Jeff Law <law@redhat.com>
34784 PR tree-optimization/80216
34785 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34786 function name. Limit recursion depth.
34787 (record_temporary_equivalences): Corresponding changes.
34789 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34791 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34794 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34797 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34799 * cfgcleanup.c (reg_note_cfa_p): New array.
34800 (insns_have_identical_cfa_notes): New function.
34801 (old_insns_match_p): Don't cross-jump in between /f
34802 and non-/f instructions. If both i1 and i2 are frame related,
34803 verify all CFA notes, their order and content.
34805 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34808 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34809 HImode and SImode with zero extend to DImode to one insn.
34810 (bswap<mode>2_extenddi): Likewise.
34811 (bswapsi2_extenddi): Likewise.
34812 (bswaphi2_extendsi): Likewise.
34813 (bswaphi2): Combine bswap HImode and SImode into one insn.
34814 Separate memory insns from swapping register.
34815 (bswapsi2): Likewise.
34816 (bswap<mode>2): Likewise.
34817 (bswaphi2_internal): Delete, no longer used.
34818 (bswapsi2_internal): Likewise.
34819 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34820 store, and gpr<-gpr swap insns.
34821 (bswap<mode>2_store): Likewise.
34822 (bswaphi2_reg): Register only splitter, combine with the splitter.
34823 (bswaphi2 splitter): Likewise.
34824 (bswapsi2_reg): Likewise.
34825 (bswapsi2 splitter): Likewise.
34826 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34827 the insns into load, store, and register/register insns.
34828 (bswapdi2_ldbrx): Likewise.
34829 (bswapdi2_load): Likewise.
34830 (bswapdi2_store): Likewise.
34831 (bswapdi2_reg): Likewise.
34833 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34835 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34836 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34838 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34841 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34843 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34844 special handling for target option conflicts between dform
34845 options (-mpower9-dform, -mpower9-dform-vector,
34846 -mpower9-dform-scalar) and -mno-direct-move.
34848 2017-03-27 Richard Biener <rguenther@suse.de>
34850 PR tree-optimization/80181
34851 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34853 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34855 * config/arc/predicates.md (move_double_src_operand): Replace the
34856 call to move_double_src_operand with a call to address_operand.
34858 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34860 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34861 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34862 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34864 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34866 * config/arc/predicates.md (long_immediate_loadstore_operand):
34867 Consider scaled addresses cases.
34869 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34871 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34872 restored when in interrupt.
34873 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34874 doesn't have delay slot.
34876 2017-03-27 Richard Biener <rguenther@suse.de>
34879 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34880 inlined thunk clones.
34882 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34885 * asan.c (instrument_derefs): Copy over last operand from
34886 original COMPONENT_REF to the new COMPONENT_REF with
34887 DECL_BIT_FIELD_REPRESENTATIVE.
34888 * ubsan.c (instrument_object_size): Likewise.
34890 2017-03-27 Richard Biener <rguenther@suse.de>
34892 PR tree-optimization/80170
34893 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34894 sure DR/SCEV didnt fold in constants we do not see when looking
34895 at the reference base alignment.
34897 2017-03-27 Richard Biener <rguenther@suse.de>
34899 PR middle-end/80171
34900 * gimple-fold.c (fold_ctor_reference): Properly guard against
34901 NULL return value from canonicalize_constructor_val.
34903 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34906 * config/i386/i386.c (ix86_expand_builtin)
34907 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34908 flags reg setting and flags reg using instructions.
34909 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34910 clobbering instructions to zero extend op2.
34912 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34914 * doc/install.texi (Configuration) <--with-aix-soname>:
34915 Update link to AIX ld.
34917 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34919 PR rtl-optimization/80160
34920 PR rtl-optimization/80159
34921 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34922 reg_alternate_class into account.
34924 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34927 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34928 to consider in curr_insn_transform.
34930 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34932 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34933 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34934 and emit_mode_inner.
34936 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34938 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34939 argument to the overloaded builtin variants. Use the new flag to
34940 deprecate certain builtin variants.
34941 * config/s390/s390-builtin-types.def: Add new builtin types.
34942 * config/s390/s390-builtins.h: Support new flags field for
34943 overloaded builtins.
34944 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34945 (s390_macro_to_expand): Enable vector float data type.
34946 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34947 builtins by incrementing the __VEC__ version number.
34948 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34950 (s390_resolve_overloaded_builtin): Emit error messages depending
34951 on the builtin flags.
34952 * config/s390/s390.c (s390_expand_builtin): Support additional
34953 flags argument. Change error message to match the messages
34954 emitted in s390-c.c.
34955 * config/s390/s390.md: New UNSPEC_* constants.
34956 (op_type): Add new instruction types.
34957 * config/s390/vecintrin.h: Add new builtins and test data class
34959 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34960 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34961 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34962 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34963 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34964 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34966 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34967 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34968 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34969 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34971 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34972 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34973 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34974 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34976 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34977 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34978 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34979 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34980 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34981 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34982 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34984 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34985 ("vec_scatter_element<V_HW_4:mode>_DI")
34986 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34987 ("vec_fpint<mode>", "vflls")
34988 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34989 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34990 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34991 ("*vec_cmphe<mode>_cc"): ... these.
34993 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34994 mode constant instead of magic value.
34996 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34998 * config/s390/s390.c (s390_expand_vec_compare): Support other
34999 vector floating point modes than just V2DF.
35000 (s390_expand_vcond): Likewise.
35001 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35002 (s390_cannot_change_mode_class): Prevent mode changes between TF
35003 and V1TF in vector registers.
35004 * config/s390/s390.md (DF, SF): New mode attributes.
35005 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35006 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35007 SFmode support for VRs.
35008 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35010 (VFT, VF_HW): New mode iterators.
35011 (vw, sdx): New mode attributes.
35012 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35013 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35014 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35015 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35016 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35017 also the new vector floating point modes. Renaming to ...
35019 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35020 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35021 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35022 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35023 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35024 ("vec_unordered<mode>"): ... these.
35026 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35027 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35028 ("*vec_extendv2df"): New insn definitions.
35030 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35032 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35033 ("mulditi3_2", "*muldi3_sign"): New patterns.
35034 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35035 rename the pattern definition.
35037 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35039 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35041 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35043 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35045 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35046 instruction if possible.
35047 * config/s390/vector.md (vec_halfnumelts): New mode
35049 ("*vec_vllezlf<mode>"): New pattern.
35051 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35053 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35054 ("popcountv4si2", "popcountv2di2"): Rename to ...
35055 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35056 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35058 ("popcount<mode>2_vxe"): New pattern.
35060 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35062 * common/config/s390/s390-common.c (processor_flags_table): Add
35064 * config.gcc: Add arch12.
35065 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35066 Default to arch12 for unknown CPU model numbers.
35067 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35068 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35069 PROCESSOR_max sanity check.
35070 * config/s390/s390-opts.h (enum processor_type): Add
35072 * config/s390/s390.c (processor_table): Add arch12.
35073 (s390_expand_builtin): Add check for B_VXE flag.
35074 (s390_issue_rate): Add PROCESSOR_ARCH12.
35075 (s390_get_sched_attrmask): Likewise.
35076 (s390_get_unit_mask): Likewise.
35077 (s390_sched_score): Enable z13 scheduling for arch12.
35078 (s390_sched_reorder): Likewise.
35079 (s390_sched_variable_issue): Likewise.
35080 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35082 (s390_tune_attr): Use z13 scheduling also for arch12.
35083 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35084 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35085 (TARGET_VXE_P): New macros.
35086 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35087 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35088 * config/s390/s390.opt: Add arch12 as processor_type.
35090 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35092 * config/s390/s390.md
35093 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35094 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35095 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35097 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35098 Rename expanders to ...
35100 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35101 ("fixuns_truncdddi2_emu"): ... these.
35103 ("fixuns_trunc<mode>si2_emu"): New expander.
35105 ("*fixuns_truncdfdi2_z13"): Rename to ...
35106 ("*fixuns_truncdfdi2_vx"): ... this.
35108 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35110 * config/s390/2964.md: Remove the single element vector compare
35111 instructions which are no longer used.
35112 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35114 (s390_canonicalize_comparison): Remove handling of DFmode
35116 (s390_expand_vec_compare_scalar): Remove function.
35117 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35118 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35120 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35122 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35124 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35126 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35127 will anyway by matched by mov<mode>_64dfp.
35129 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35131 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35132 vlef/vstef. Add missing operand to vleif.
35134 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35136 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35137 pair for all vector types with 64 bit elements.
35138 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35139 * config/s390/vector.md (V_HW_64): ... here.
35140 (V_128_NOSINGLE): New mode iterator.
35141 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35142 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35143 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35144 ("*vec_load_pairv2di"): Change to ...
35145 ("*vec_load_pair<mode>"): ... this one.
35147 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35149 * config/s390/constraints.md: Add comments.
35150 (jKK): Reject element sizes > 8 bytes.
35151 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35153 * config/s390/s390.md: Add the s_operand checks formerly in
35154 s390_split_ok_p to various splitters where they are still
35156 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35157 for 128 bit vectors. Plus two splitters.
35159 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35161 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35164 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35167 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35168 error if the boundary argument is not constant.
35170 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35172 PR rtl-optimization/80112
35173 * loop-doloop.c (doloop_condition_get): Don't check condition
35174 if cmp isn't SET with IF_THEN_ELSE src.
35176 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35178 PR tree-optimization/80158
35179 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35180 replacing a candidate statement, also replace it for the
35181 candidate's alternate interpretation.
35182 (replace_rhs_if_not_dup): Likewise.
35183 (replace_one_candidate): Likewise.
35185 2017-03-24 Richard Biener <rguenther@suse.de>
35187 PR tree-optimization/80167
35188 * graphite-isl-ast-to-gimple.c
35189 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35191 (translate_isl_ast_to_gimple::get_rename): Likewise.
35193 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35195 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35196 handling of certain combinations of target options, including the
35197 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35198 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35200 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35203 * config/arm/arm.md (*load_multiple): Add reload_completed to
35204 matching condition.
35206 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35207 Richard Biener <rguenth@suse.de>
35209 PR tree-optimization/79908
35210 PR tree-optimization/80136
35211 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35212 been cast away, gimplify_and_add suffices.
35214 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35216 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35218 2017-03-23 Richard Biener <rguenther@suse.de>
35220 PR tree-optimization/80032
35221 * gimplify.c (gimple_push_cleanup): Forced unconditional
35222 cleanups still have to go to the conditional_cleanups
35225 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35227 PR tree-optimization/80072
35228 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35230 (next_operand_entry_id): Change type to unsigned int.
35231 (sort_by_operand_rank): Make sure to return the right return value
35232 even if unsigned fields are bigger than INT_MAX.
35233 (struct oecount): Change cnt and id type to unsigned int.
35234 (oecount_hasher::equal): Formatting fix.
35235 (oecount_cmp): Make sure to return the right return value
35236 even if unsigned fields are bigger than INT_MAX.
35237 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35240 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35241 TREE_READONLY on result if writing it more than once.
35244 * doc/invoke.texi (-fsanitize=thread): Document that with
35245 -fnon-call-exceptions atomics are not able to throw
35249 * tsan.c: Include tree-eh.h.
35250 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35251 maybe_clean_or_replace_eh_stmt where needed.
35252 (instrument_memory_accesses): Add cfg_changed argument.
35253 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35254 if it returned true.
35255 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35257 PR rtl-optimization/63191
35258 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35259 wrapper function, moved the whole old content into ...
35260 (ix86_delegitimize_address_1): ... this. New inline function.
35261 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35262 true as last argument instead of ix86_delegitimize_address.
35264 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35266 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35267 cortexa57_branch_cost.
35269 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35271 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35273 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35276 * doc/md.texi (Constraints): Document wA constraint.
35277 * config/rs6000/constraints.md (wA): New.
35278 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35279 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35280 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35281 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35283 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35286 * gimplify.c (is_oacc_declared): New function.
35287 (oacc_default_clause): Use it to set default flags for acc declared
35288 variables inside parallel regions.
35289 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35290 declared variables.
35291 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35292 declare attribute to any decl as necessary.
35294 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35297 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35298 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35299 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35300 (arm_arch_lpae): This.
35301 * config/arm/arm.c (arm_arch7ve): Rename into ...
35302 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35303 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35306 2017-03-22 Martin Liska <mliska@suse.cz>
35309 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35310 error message instead of an ICE.
35312 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35314 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35316 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35318 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35320 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35323 2017-03-21 Martin Sebor <msebor@redhat.com>
35325 * doc/extend.texi: Use "cannot" instead of "can't."
35326 * doc/hostconfig.texi: Same.
35327 * doc/install.texi: Same.
35328 * doc/invoke.texi: Same.
35329 * doc/loop.texi: Same.
35330 * doc/md.texi: Same.
35331 * doc/objc.texi: Same.
35332 * doc/rtl.texi: Same.
35333 * doc/tm.texi: Same.
35334 * doc/tm.texi.in: Same.
35335 * doc/trouble.texi: Same.
35337 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35340 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35341 (collect_checksum_attributes): Set it.
35342 (die_checksum_ordered): Use it.
35344 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35346 PR tree-optimization/79908
35347 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35348 change: For a VA_ARG whose LHS has been cast away, use
35349 force_gimple_operand to construct the side effects.
35351 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35353 PR translation/80001
35354 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35355 more amenable to translation.
35356 (oacc_loop_auto_partitions): Likewise.
35358 2017-03-21 Marek Polacek <polacek@redhat.com>
35359 Martin Sebor <msebor@redhat.com>
35361 PR tree-optimization/80109
35362 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35363 on INTEGRAL_TYPE_P.
35365 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35366 Segher Boessenkool <segher@kernel.crashing.org>
35369 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35370 check reg_used_between_p between insn and one of succ or succ2
35371 depending on if succ is artificial insn not inserted into insn
35374 2017-03-21 Martin Liska <mliska@suse.cz>
35376 PR gcov-profile/80081
35377 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35378 * doc/gcc.texi: Include gcov-dump stuff.
35379 * doc/gcov-dump.texi: New file.
35381 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35383 PR rtl-optimization/79150
35384 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35385 conditional jump, if the jump is the last insn of the loop.
35387 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35388 Richard Biener <rguenth@suse.de>
35390 PR tree-optimization/79908
35391 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35392 been cast away, use force_gimple_operand to construct the side
35395 2017-03-21 Martin Liska <mliska@suse.cz>
35397 PR libfortran/79956
35398 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35401 2017-03-21 Brad Spengler <spender@grsecurity.net>
35404 * plugin.c (htab_hash_plugin): New function.
35405 (add_new_plugin): Use it and adjust.
35406 (parse_plugin_arg_opt): Adjust.
35407 (init_one_plugin): Likewise.
35409 2017-03-21 Richard Biener <rguenther@suse.de>
35411 PR tree-optimization/80032
35412 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35413 if set force the cleanup to happen unconditionally.
35414 (gimplify_target_expr): Push inserted clobbers with force_uncond
35415 to avoid them being removed by control-dependent DCE.
35417 2017-03-21 Richard Biener <rguenther@suse.de>
35419 PR tree-optimization/80122
35420 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35421 va_arg_pack_len when the inlined call stmt requires pack
35423 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35425 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35428 * tsan.c (instrument_builtin_call): If the memory model argument
35429 is not a constant, assume it is valid.
35432 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35435 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35437 PR rtl-optimization/79910
35438 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35439 if its dest is used by an insn before I2 (other than the combined
35440 insns themselves, which are properly handled already).
35442 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35445 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35447 * combine.c (record_used_regs): New static function.
35448 (try_combine): Handle situations where there is an additional
35449 instruction between I2 and I3 which needs to have a LOG_LINK
35453 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35455 * combine.c (try_combine): Delete redundant i1 test. Call
35456 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35458 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35461 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35462 alternatives 13/14.
35464 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35466 PR tree-optimization/80054
35467 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35468 the optimization if a PHI or any of its arguments is not dominated
35469 by the candidate's basis. Use gphi* rather than gimple* as
35471 (replace_profitable_candidates): Clean up a gimple* variable that
35472 should be a gphi* variable.
35474 2017-03-20 Martin Sebor <msebor@redhat.com>
35477 * doc/extend.texi (attribute constructor): Document present limitation.
35479 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35482 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35483 __POWER9_VECTOR__ #ifdef control, change template definition to
35484 use Power9-specific built-in function.
35485 (vec_any_eq): Likewise.
35486 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35487 to control outcomes from this test.
35488 (vector_ae_<mode>p): For VEC_F modes, likewise.
35490 2017-03-20 Ian Lance Taylor <iant@google.com>
35492 * config/i386/i386.c (ix86_function_regparm): Save an extra
35493 register for -fsplit-stack with DECL_STATIC_CHAIN.
35495 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35498 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35499 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35501 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35503 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35505 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35508 2017-03-20 Marek Polacek <polacek@redhat.com>
35511 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35513 2017-03-20 Richard Biener <rguenther@suse.de>
35515 PR tree-optimization/80113
35516 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35517 allocate extra SSA name for PHI def.
35518 (add_close_phis_to_outer_loops): Likewise.
35519 (add_close_phis_to_merge_points): Likewise.
35520 (copy_loop_close_phi_args): Likewise.
35521 (copy_cond_phi_nodes): Likewise.
35523 2017-03-20 Martin Liska <mliska@suse.cz>
35525 PR middle-end/79753
35526 * tree-chkp.c (chkp_build_returned_bound): Do not build
35527 returned bounds for a LHS that's not a BOUNDED_P type.
35529 2017-03-20 Martin Liska <mliska@suse.cz>
35533 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35534 COMPLEX_CST and VECTOR_CST.
35536 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35539 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35540 target operand. A new splitter adds the clobber statement in case
35541 the target operand is dead anyway.
35543 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35545 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35546 to age-old versions of binutils and glibc.
35548 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35550 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35552 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35554 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35556 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35558 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35559 requirement for binutils 2.13.
35561 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35563 * combine.c (try_combine): Delete redundant i1 test. Call
35564 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35566 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35568 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35569 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35571 <riscv64-*-elf>: Re-arrange section
35572 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35573 <riscv32-*-linux>: Likewise.
35574 <riscv64-*-elf>: Likewise
35575 <riscv64-*-linux>: Likewise.
35577 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35580 * aarch64.opt(verbose-cost-dump): Fix typo.
35582 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35585 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35586 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35588 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35590 * reload.c (find_reloads): When reloading a nonoffsettable address,
35591 use RELOAD_OTHER for it and its address reloads.
35593 PR rtl-optimization/79910
35594 * combine.c (record_used_regs): New static function.
35595 (try_combine): Handle situations where there is an additional
35596 instruction between I2 and I3 which needs to have a LOG_LINK
35599 2017-03-17 Jeff Law <law@redhat.com>
35601 PR tree-optimization/71437
35602 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35603 conditional in the hash table first.
35604 (vrp_dom_walker::before_dom_children): Extract condition from
35605 ASSERT_EXPR. Record condition, its inverion and any implied
35606 conditions as well.
35608 2017-03-17 Marek Polacek <polacek@redhat.com>
35609 Markus Trippelsdorf <markus@trippelsdorf.de>
35611 PR tree-optimization/80079
35612 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35615 2017-03-17 Richard Biener <rguenther@suse.de>
35617 PR middle-end/80075
35618 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35619 Properly verify the LHS before the RHS possibly claims to be
35621 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35624 2017-03-17 Martin Jambor <mjambor@suse.cz>
35626 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35627 (List of -O2 options): Likewise.
35628 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35631 2017-03-17 Tom de Vries <tom@codesourcery.com>
35633 * gcov-dump.c (print_usage): Print bug_report_url.
35635 2017-03-17 Richard Biener <rguenther@suse.de>
35637 PR middle-end/80050
35638 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35639 (parser::peek): Likewise.
35641 2017-03-17 Richard Biener <rguenther@suse.de>
35643 PR tree-optimization/80048
35644 * sese.c (free_sese_info): Properly release rename_map and
35645 copied_bb_map elements.
35647 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35649 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35650 Add linked-list forward and backlinks. Insert on
35651 construction, remove on destruction.
35652 (class pass_store_merging): Add m_stores_head field.
35653 (pass_store_merging::terminate_and_process_all_chains):
35654 Iterate over m_stores_head list.
35655 (pass_store_merging::terminate_all_aliasing_chains):
35657 (pass_store_merging::execute): Check for debug stmts first.
35658 Push new chains onto the m_stores_head stack.
35660 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35663 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35664 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35665 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35667 2017-03-16 Jeff Law <law@redhat.com>
35669 PR tree-optimization/71437
35670 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35671 member function. Implementation moved into after_dom_children
35672 member function and into the threader's thread_outgoing_edges
35674 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35675 some code into new thread_outgoing_edges.
35676 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35677 definition. Simplify marker handling (do it here). Assume we always
35678 have the available expression and the const/copies tables.
35679 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35681 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35682 * tree-vrp.c (equiv_stack): No longer file scoped.
35683 (vrp_dom_walker): New class.
35684 (vrp_dom_walker::before_dom_children): New member function.
35685 (vrp_dom_walker::after_dom_children): Likewise.
35686 (identify_jump_threads): Setup domwalker. Use it rather than
35687 walking edges in a random order by hand. Simplify setup/finalization.
35688 (finalize_jump_threads): Remove.
35689 (vrp_finalize): Do not call identify_jump_threads here.
35690 (execute_vrp): Do it here instead and call thread_through_all_blocks
35693 PR tree-optimization/71437
35694 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35696 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35698 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35699 (dom_opt_dom_walker::thread_across_edge): Remove
35700 handle_dominating_asserts argument. All callers changed.
35701 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35702 changes. Remove calls to lhs_of_dominating_assert. Other
35703 uses of handle_dominating_asserts turn into unconditional code
35704 (simplify_control_stmt_condition_1): Likewise.
35705 (simplify_control_stmt_condition): Likewise.
35706 (thread_through_normal_block, thread_across_edge): Likewise.
35707 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35708 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35709 object if it is not an SSA_NAME.
35710 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35711 before calling into the VRP specific simplifiers.
35712 (identify_jump_threads): Remove handle_dominating_asserts
35715 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35718 * tree-diagnostic.c (default_tree_printer): No longer static.
35719 * tree-diagnostic.h (default_tree_printer): New prototype.
35721 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35723 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35724 Change ins into fmov.
35726 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35728 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35729 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35730 Use h_con constraint for operand 1.
35731 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35732 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35734 2017-03-15 Jeff Law <law@redhat.com>
35736 PR tree-optimization/71437
35737 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35738 (record_temporary_equivalences): Use it.
35740 PR tree-optimization/71437
35741 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35742 tree-ssa-scopedtables.
35743 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35744 (record_conditions, record_cond, vuse_eq): Likewise.
35745 (record_edge_info): Adjust to API tweak of record_conditions.
35746 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35747 (record_temporary_equivalences, optimize_stmt): Likewise.
35748 (eliminate_redundant_computations): Likewise.
35749 (record_equivalences_from_stmt): Likewise.
35750 * tree-ssa-scopedtables.c: Include options.h and params.h.
35751 (vuse_eq): New function, moved from tree-ssa-dom.c
35752 (build_and_record_new_cond): Likewise.
35753 (record_conditions): Likewise. Accept vector of conditions rather
35754 than edge_equivalence structure for first argument.
35755 for the first argument.
35756 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35757 from tree-ssa-dom.c.
35758 (avail_exprs_stack::record_cond): Likewise.
35759 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35760 from tree-ssa-dom.c.
35761 (avail_exprs_stack): Add new member functions lookup_avail_expr
35763 (record_conditions): Declare.
35765 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35768 * lra-constraints.c (process_alt_operands): Increase reject for
35769 reloading an input/output operand.
35771 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35774 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35775 insns to convert from signed/unsigned char/short to IEEE 128-bit
35777 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35779 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35782 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35783 subreg of inner mode for values already in registers.
35785 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35787 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35788 iteration reg is used after the loop.
35790 2017-03-14 Martin Sebor <msebor@redhat.com>
35792 PR tree-optimization/79800
35793 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35794 precision in negative-positive range.
35795 (format_floating): Call non-const overload with adjusted precision.
35797 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35800 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35803 2017-03-14 Martin Sebor <msebor@redhat.com>
35805 PR middle-end/80020
35806 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35807 * builtins.def (aligned_alloc): Use it.
35810 * Makefile.in (GTFILES): Add calls.c.
35811 * calls.c: Include "gt-calls.h".
35813 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35815 PR rtl-optimization/79728
35816 * regs.h (struct target_regs): New field
35817 x_contains_allocatable_regs_of_mode.
35818 (contains_allocatable_regs_of_mode): New macro.
35819 * reginfo.c (init_reg_sets_1): Initialize it, and change
35820 contains_reg_of_mode so it includes global regs as well.
35821 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35822 rather than contains_regs_of_mode.
35824 2017-03-14 Martin Liska <mliska@suse.cz>
35826 * doc/invoke.texi: Document options that can't be combined with
35827 -fcheck-pointer-bounds.
35829 2017-03-14 Martin Liska <mliska@suse.cz>
35831 PR middle-end/79831
35832 * doc/invoke.texi (-Wchkp): Document the option.
35834 2017-03-14 Martin Liska <mliska@suse.cz>
35836 * Makefile.in: Install gcov-dump.
35838 2017-03-14 Martin Liska <mliska@suse.cz>
35840 * multiple_target.c (expand_target_clones): Bail out for
35841 an invalid attribute.
35843 2017-03-14 Richard Biener <rguenther@suse.de>
35845 * alias.c (struct alias_set_entry): Pack properly.
35846 * cfgloop.h (struct loop): Likewise.
35847 * cse.c (struct set): Likewise.
35848 * ipa-utils.c (struct searchc_env): Likewise.
35849 * loop-invariant.c (struct invariant): Likewise.
35850 * lra-remat.c (struct cand): Likewise.
35851 * recog.c (struct change_t): Likewise.
35852 * rtl.h (struct address_info): Likewise.
35853 * symbol-summary.h (function_summary): Likewise.
35854 * tree-loop-distribution.c (struct partition): Likewise.
35855 * tree-object-size.c (struct object_size_info): Likewise.
35856 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35857 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35858 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35859 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35860 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35861 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35862 (struct _stmt_vec_info): Likewise.
35864 2017-03-14 Martin Liska <mliska@suse.cz>
35867 * multiple_target.c (create_dispatcher_calls): Check that
35868 a target can create a function dispatcher.
35870 2017-03-14 Martin Liska <mliska@suse.cz>
35873 * multiple_target.c (expand_target_clones): Drop local.local
35874 flag for default implementation.
35876 2017-03-14 Richard Biener <rguenther@suse.de>
35878 PR tree-optimization/80030
35879 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35881 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35883 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35884 gcc_fallthrough() instead of __attribute__((fallthrough));
35886 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35888 * doc/gcc.texi: Remove "up" link to (DIR).
35889 * doc/gccint.texi: Ditto.
35891 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35893 * doc/install.texi (Specific) <avr>: Remove reference to
35896 2017-03-13 Jeff Law <law@redhat.com>
35898 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35899 attribute rather than comments.
35901 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35902 match_scratch operand is highest.
35904 2017-03-13 Martin Liska <mliska@suse.cz>
35906 PR middle-end/78339
35907 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35908 is a CHKP clone, use original declaration.
35910 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35912 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35913 (arc_conditional_register_usage): Use a different allocation order
35914 when optimizing for size.
35915 * common/config/arc/arc-common.c (arc_option_optimization_table):
35916 Section anchors default on when optimizing for size.
35918 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35920 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35922 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35924 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35925 * config/arc/arc.md (cpu_facility): Add cd variant.
35926 (*movqi_insn): Add code density variant.
35927 (*movhi_insn): Likewise.
35928 (*movqi_insn): Likewise.
35929 (*addsi3_mixed): Likewise.
35930 (subsi3_insn): Likewise.
35932 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35934 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35936 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35938 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35939 expressions with MINUS and UNARY ops.
35941 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35944 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35946 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35947 between vec_select and vector argument.
35948 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35949 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35950 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35951 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35952 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35953 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35955 2017-03-13 Richard Biener <rguenther@suse.de>
35958 * params.def (vect-max-peeling-for-alignment): Fix typo.
35960 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35962 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35963 issue that only occurred with binutils below 2.18.
35965 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35967 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35968 refer to binutils 2.11/2.12 minimum.
35970 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35972 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35973 ftp.kernel.org and simplify binutils requirement.
35975 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35977 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35979 (Optimize Options): Ditto. Also remove redundancy.
35981 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35983 PR translation/79848
35984 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35986 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35987 to G_ to avoid double translation.
35989 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35991 PR translation/79923
35992 * auto-profile.c (get_combined_location): Convert leading
35993 character of diagnostics to lower case and remove trailing period.
35994 (read_profile): Likewise for various diagnostics.
35995 * config/arm/arm.c (arm_option_override): Remove trailing period
35996 from various diagnostics.
35997 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35998 (msp430_expand_delay_cycles): Likewise.
36000 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36003 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36004 full command-line argument, rather than just "str".
36005 (aarch64_validate_march): Likewise.
36006 (aarch64_validate_mtune): Likewise.
36008 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36010 PR rtl-optimization/78911
36011 * lra-assigns.c (must_not_spill_p): New function.
36012 (spill_for): Use it.
36014 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36016 PR tree-optimization/79981
36017 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36018 ATOMIC_COMPARE_EXCHANGE ifn result.
36019 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36020 IFN_ATOMIC_COMPARE_EXCHANGE.
36022 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36025 * opts.c (parse_sanitizer_options): Add missing question mark to
36026 "did you mean" message.
36028 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36030 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36032 (VMULEUH_UNS): Likewise.
36033 (VMULOUB_UNS): Likewise.
36034 (VMULOUH_UNS): Likewise.
36035 * config/rs6000/rs6000.c (builtin_function_type): Remove
36036 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36038 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36041 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36042 x with result of extra_parsing_for_operand_code_0.
36043 (function_reader::extra_parsing_for_operand_code_0): Convert
36044 return type from void to rtx, returning x. When reading
36045 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36046 larger size containing struct block_symbol.
36048 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36050 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36051 -mfloat128-hardware without -m64.
36053 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36056 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36057 entries to the case statement that marks unsigned arguments to
36058 overloaded functions.
36060 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36062 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36063 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36065 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36068 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36069 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36071 2017-03-10 Martin Liska <mliska@suse.cz>
36075 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36076 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36079 2017-03-10 Olivier Hainque <hainque@adacore.com>
36081 * tree-switch-conversion (array_value_type): Start by resetting
36082 candidate type to it's main variant.
36084 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36086 PR rtl-optimization/79909
36087 * combine.c (try_combine): Use simplify_replace_rtx on individual
36088 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36089 of the whole CALL_INSN_FUNCTION_USAGE.
36091 PR tree-optimization/79972
36092 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36093 get_range_info on SSA_NAMEs. Formatting fixes.
36095 2017-03-10 Richard Biener <rguenther@suse.de>
36096 Jakub Jelinek <jakub@redhat.com>
36098 PR tree-optimization/77975
36099 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36100 edge to be constant.
36101 (get_val_for): For constant x return it. Formatting fix.
36102 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36103 would use the same bases as the current one.
36105 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36107 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36108 instead of vec_select for V1TImode.
36109 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36111 (VSX_LE_128): Add V1TI to this mode iterator.
36112 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36113 (*vsx_le_perm_store_<mode>): Likewise.
36114 (pre-reload splitter for VSX stores): Likewise.
36115 (post-reload splitter for VSX stores): Likewise.
36116 (*vsx_xxpermdi2_le_<mode>): Likewise.
36117 (*vsx_lxvd2x2_le_<mode>): Likewise.
36118 (*vsx_stxvd2x2_le_<mode>): Likewise.
36120 2017-03-09 Michael Eager <eager@eagercon.com>
36122 Correct failures with --enable-checking=yes,rtl.
36124 * config/microblaze/microblaze.c (microblaze_expand_shift):
36125 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36126 test for const0_rtx.
36127 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36128 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36130 2017-03-09 Richard Biener <rguenther@suse.de>
36132 PR tree-optimization/79977
36133 * graphite-scop-detection.c (scop_detection::merge_sese):
36134 Handle the case of extra exits to blocks dominating the entry.
36136 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36138 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36141 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36143 PR rtl-optimization/79949
36144 * lra-constraints.c (process_alt_operands): Check memory when
36145 trying to predict a cycle. Print about the overall increase.
36147 2017-03-09 Richard Biener <rguenther@suse.de>
36149 PR middle-end/79971
36150 * gimple-expr.c (useless_type_conversion_p): Preserve
36151 TYPE_SATURATING for fixed-point types.
36153 2017-03-09 Richard Biener <rguenther@suse.de>
36156 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36157 alignment of BLKmode params.
36159 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36162 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36163 (VALL_NO_V2Q): Likewise.
36165 * config/aarch64/aarch64-simd.md
36166 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36168 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36169 VALL_NO_V2Q mode iterator.
36170 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36172 2017-03-09 Martin Liska <mliska@suse.cz>
36174 PR tree-optimization/79631
36175 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36176 tree_int_cst_sign_bit just for INTEGER constants.
36178 2017-03-09 Martin Liska <mliska@suse.cz>
36182 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36185 2017-03-09 Marek Polacek <polacek@redhat.com>
36188 * tree.c (inchash::add_expr): Handle TREE_VEC.
36190 2017-03-09 Martin Liska <mliska@suse.cz>
36193 (chkp_narrow_size_and_offset): New function.
36194 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36195 (void chkp_parse_bit_field_ref): New function.
36196 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36197 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36199 2017-03-09 Martin Liska <mliska@suse.cz>
36202 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36203 (chkp_find_bounds_1): Remove gcc_unreachable.
36205 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36208 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36209 BUILT_IN_SYNC*, determine the access type from the size suffix and
36210 always build a MEM_REF with that type. Handle forgotten
36211 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36214 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36215 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36216 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36217 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36218 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36219 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36220 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36221 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36222 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36223 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36224 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36225 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36226 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36227 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36228 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36229 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36230 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36231 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36232 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36233 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36234 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36235 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36236 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36237 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36238 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36239 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36240 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36241 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36242 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36243 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36244 definitions outside of __OPTIMIZE__ guarded section.
36247 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36248 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36249 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36250 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36253 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36255 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36256 ("vfenez<mode>"): Add missing constraints.
36258 2017-03-08 Martin Sebor <msebor@redhat.com>
36261 * config/nds32/nds32.c (nds32_option_override):
36262 Fix misspelled diagnostic.
36264 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36267 * gimplify.c (gimplify_omp_for): Replace index var in outer
36268 taskloop statement with an artificial variable and add
36269 OMP_CLAUSE_PRIVATE clause for it.
36271 2017-03-08 Richard Biener <rguenther@suse.de>
36273 PR tree-optimization/79955
36274 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36275 for accesses that are completely outside of the variable.
36277 2017-03-08 Andrew Haley <aph@redhat.com>
36279 PR tree-optimization/79943
36280 * tree-ssa-loop-split.c (compute_new_first_bound): When
36281 calculating the new upper bound, (END-BEG) should be added, not
36284 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36286 * config/avr/avr.md (setmemhi): Make sure match_dup
36287 operand number comes before match_scratch.
36289 2017-03-08 Richard Biener <rguenther@suse.de>
36291 PR tree-optimization/79920
36292 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36293 with ncopies == 1 to ...
36294 (vect_transform_slp_perm_load): ... here. Properly compute
36295 all element loads by iterating VF times over the group. Do
36296 not handle ncopies (computed in a broken way) in
36297 vect_create_mask_and_perm.
36299 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36302 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36303 is a uniform vector, use uniform_vector_p return value instead of
36304 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36306 2017-03-07 Marek Polacek <polacek@redhat.com>
36308 PR middle-end/79809
36309 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36310 (alloca_call_type): Likewise.
36312 2017-03-07 Martin Liska <mliska@suse.cz>
36314 * gcov.c (process_args): Put comment to correct location.
36316 2017-03-07 Martin Liska <mliska@suse.cz>
36318 PR middle-end/68270
36319 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36320 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36321 (chkp_narrow_bounds_for_field): Likewise.
36322 (chkp_parse_array_and_component_ref): Pass one more argument to
36325 2017-03-07 Richard Biener <rguenther@suse.de>
36327 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36330 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36332 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36333 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36335 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36338 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36339 to end of description.
36340 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36342 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36344 PR rtl-optimization/79901
36345 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36347 (*avx512f_<code><mode>3<mask_name>): ... this.
36348 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36349 iterator instead of VI8_AVX2_AVX512BW.
36351 PR rtl-optimization/79901
36352 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36353 min/max expander, expand it using expand_vec_cond_expr.
36356 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36359 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36362 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36363 to void * for PCH reasons.
36364 * dwarf2out.c (output_loc_operands, output_die): Cast
36365 v.val_vec.array to unsigned char *.
36367 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36370 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36373 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36375 PR rtl-optimization/79571
36376 * lra-constraints.c (process_alt_operands): Calculate static
36377 reject and subtract it from overall when only addresses will be
36380 2017-03-06 Julia Koval <julia.koval@intel.com>
36383 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36384 incoming stack boundary to 128 for 64-bit targets.
36386 2017-03-06 Richard Biener <rguenther@suse.de>
36388 PR tree-optimization/79894
36389 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36390 to NULL after folding it.
36392 2017-03-06 Richard Biener <rguenther@suse.de>
36394 PR tree-optimization/79824
36395 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36396 check disabling peeling for gaps.
36398 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36400 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36401 attributes): Document gettimeofday.
36403 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36405 * config/s390/s390.c (s390_option_override_internal): Set
36406 PARAM_MIN_VECT_LOOP_BOUND
36408 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36410 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36411 * config/s390/s390.md: Likewise.
36413 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36416 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36417 (<avx2_avx512>_perm<mode>): Rename to ...
36418 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36420 (<avx512>_perm<mode>_mask): Rename to ...
36421 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36423 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36424 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36425 instead of VI8F_256_512.
36426 (avx512f_perm<mode>): New define_expand.
36427 (avx512f_perm<mode>_mask): Likewise.
36428 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36429 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36431 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36433 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36434 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36436 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36438 2017-03-06 Martin Liska <mliska@suse.cz>
36441 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36442 when having a SSA NAME w/o VAR_DECL assigned to it.
36444 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36446 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36447 msa_dpsub_<su>_d): Fix MODE for vec_select.
36449 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36451 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36453 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36455 2017-03-06 Richard Biener <rguenther@suse.de>
36457 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36458 * plugin.c (register_plugin_info): Likewise.
36459 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36461 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36463 * config/i386/sse.md (sse_storehps, sse_storelps,
36464 avx_<castmode><avxsizesuffix>_<castmode>,
36465 avx512f_<castmode><avxsizesuffix>_<castmode>,
36466 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36467 in condition that at least one operand is not a MEM.
36469 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36471 PR middle-end/79805
36472 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36473 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36475 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36476 gimple_call_nothrow_p flag based on whether original builtin can throw.
36477 If it can, emit following stmts on the fallthrough edge.
36478 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36479 don't create new bb if inserting just debug stmts on the edge, try to
36480 insert them on the fallthru bb or just reset debug stmts.
36482 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36485 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36486 restore recog_data (including the operand rtxes inside it) around
36487 the call to get_insn_template.
36489 2017-03-03 Martin Sebor <msebor@redhat.com>
36491 PR tree-optimization/79699
36492 * context.c (context::~context): Free MPFR caches to avoid
36493 a memory leak on program exit.
36495 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36497 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36498 Use wide_int::ulow () instead of .elt (0).
36500 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36502 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36503 (*pushxf): Limit oF constraint to 32bit targets and add oC
36504 constraint for 64bit targets.
36505 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36506 (*pushdf): Change rmF constraint to rmC.
36508 2017-03-03 Martin Liska <mliska@suse.cz>
36510 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36511 Remove unused variable.
36513 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36516 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36517 is a memory operand, increase num_memory.
36518 (ix86_expand_args_builtin): Likewise.
36520 2017-03-03 Jan Hubicka <jh@suse.cz>
36523 * ipa-devirt.c (maybe_record_node): Properly handle
36524 __cxa_pure_virtual visibility.
36526 2017-03-03 Martin Liska <mliska@suse.cz>
36528 PR tree-optimization/79803
36529 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36531 (pass_loop_prefetch::execute): Disabled optimization if an
36532 assumption about L1 cache size is not met.
36534 2017-03-03 Martin Liska <mliska@suse.cz>
36536 PR rtl-optimization/79574
36537 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36538 (hash_scan_set): Likewise.
36539 (dump_hash_table): Likewise.
36540 (hoist_code): Likewise.
36542 2017-03-03 Richard Biener <rguenther@suse.de>
36544 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36545 in place of uhigh ().
36546 (fixed_convert_from_real): Likewise.
36548 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36551 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36553 2017-03-03 Richard Biener <rguenther@suse.de>
36555 PR middle-end/79818
36556 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36557 TYPE_OVERFLOW_UNDEFINED check.
36559 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36561 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36563 (vector_ae_<mode>_p): Likewise.
36564 (vector_nez_<mode>_p): Likewise.
36565 (vector_ne_v2di_p): Likewise.
36566 (vector_ae_v2di_p): Likewise.
36567 (vector_ne_<mode>_p): Likewise.
36568 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36570 (vsx_tsqrt<mode>2_fe): Likewise.
36572 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36575 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36577 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36579 PR rtl-optimization/79780
36580 * cprop.c (one_cprop_pass): When second and further conditional trap
36581 in a single basic block is turned into an unconditional trap, turn it
36582 into a deleted note to avoid RTL verification failures.
36584 2017-03-02 Richard Biener <rguenther@suse.de>
36586 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36588 2017-03-02 Richard Biener <rguenther@suse.de>
36590 PR tree-optimization/79345
36592 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36593 param and abort the walk, returning -1 if it is hit.
36594 (walk_aliased_vdefs): Take a limit param and pass it on.
36595 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36596 defaulting to 0 and return a signed int.
36597 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36598 (check_defs): New helper.
36599 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36600 about uninitialized memory.
36601 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36602 bogus uninitialized warning.
36603 (fixed_convert_from_real): Likewise.
36605 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36607 PR tree-optimization/66768
36608 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36609 iv_use if base object can't be determined.
36611 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36613 PR tree-optimization/79345
36614 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36615 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36616 (get_pattern_stats): Initialize it.
36617 * genemit.c (gen_expand): Verify match_scratch numbers come after
36618 match_operand/match_dup numbers.
36619 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36620 match_scratch numbers.
36621 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36623 * config/s390/s390.md (trunctdsd2): Likewise.
36625 2017-03-02 Richard Biener <rguenther@suse.de>
36627 * wide-int.h (wide_int_storage::operator=): Implement in terms
36630 2017-03-02 Richard Biener <rguenther@suse.de>
36632 PR tree-optimization/79777
36633 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36634 the to insert expression to sth existing.
36636 2017-03-01 Martin Sebor <msebor@redhat.com>
36638 PR middle-end/79692
36639 * gimple-ssa-sprintf.c
36640 (directive::known_width_and_precision): New function.
36641 (format_integer): Use it.
36642 (get_mpfr_format_length): Consider the full range of precision
36643 when computing %g output with the # flag. Set the likely byte
36644 count to 3 rather than 1 when precision is indeterminate.
36645 (format_floating): Correct the lower bound of precision.
36647 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36649 * doc/invoke.texi: Document default code model for 64-bit Linux.
36651 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36654 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36655 udiv rather than div since input pattern is unsigned.
36657 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36659 * config/i386/i386.c (print_reg): Warn for values of
36660 unsupported size in integer register.
36662 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36665 * config/rs6000/predicates.md (current_file_function_operand): Do
36666 not allow self calls to be local if the function is replaceable.
36668 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36671 * config/rs6000/altivec.h (vec_ctz and others): Change the
36672 preprocessor macro that controls conditional compilation from
36673 _ARCH_PWR9 to __POWER9_VECTOR__.
36674 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36675 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36676 control (instead of _ARCH_PWR9 control) so that template
36677 definition uses power9-specific function.
36678 (vec_any_eq): Likewise.
36679 (vec_all_ne): Change macro definition to use a power9-specific
36680 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36681 _ARCH_PWR9 control).
36682 (vec_any_eq) Likewise.
36683 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36684 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36685 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36686 support for xvcmpnedp instruction.
36687 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36688 macro expansion so that Power9 implementation of vec_all_ne does
36689 not use the AltiVec predicate framework.
36690 (VCMPNEH_P): Likewise.
36691 (VCMPNEW_P): Likewise.
36692 (VCMPNED_P): Likewise.
36693 (VCMPNEFP_P): Likewise.
36694 (VCMPNEDP_P): Likewise.
36695 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36696 implementation of vec_any_eq to not use AltiVec predicate
36698 (VCMPAEH_P): Likewise.
36699 (VCMPAEW_P): Likewise.
36700 (VCMPAED_P): Likewise.
36701 (VCMPAEFP_P): Likewise.
36702 (VCMPAEDP_P): Likewise.
36703 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36704 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36705 not use the AltiVec predicate framework.
36706 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36707 of vec_any_eq to not use AltiVec predicate framework.
36708 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36709 support for predefined __POWER9_VECTOR__ macro to indicate that
36710 Power9 instruction selection is enabled.
36711 (altivec_overloaded_builtins): Remove extraneous
36712 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36713 function argument types RS6000_BTI_bool_V16QI and
36714 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36715 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36716 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36717 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36718 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36719 Power9 for implementations of vec_cmpne. Change the signature for
36720 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36721 (representing vec_all_ne) to remove the previously described first
36722 argument of type RS6000_BTI_INTSI, as this was an artifact of
36723 reliance on the AltiVec predicate framework, which is no longer
36724 used in the implementation of these functions. Add
36725 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36726 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36727 since, unlike the AltiVec predicate framework implementation, we
36728 do not share function descriptors between vec_alle and vec_anyeq.
36729 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36730 set of modes that receive special treatment even when
36731 TARGET_P9_VECTOR is true. The special treatment emits code that
36732 does not depend on Power9 instructions.
36733 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36734 define_expand to not rely on AltiVec predicate framework.
36735 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36737 (vector_ne_v2di_p): Change this define_expand to not rely on
36738 AltiVec predicate framework.
36739 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36741 (vector_ne_<mode>_p): Change this define_expand to not rely on
36742 AltiVec predicate framework.
36743 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36745 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36746 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36747 define_insn pattern.
36748 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36749 define_insn pattern because the xvcmpne<VSs>. instruction is not
36751 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36752 instruction is not supported.
36754 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36756 * config/nvptx/nvptx.c: Include intl.h.
36758 2017-03-01 Martin Jambor <mjambor@suse.cz>
36761 * ipa-prop.h (ipa_bits): Removed field known.
36762 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36763 to pointers. Adjusted their comments to warn about their sharing.
36764 (ipcp_transformation_summary): Change bits to a vector of pointers.
36765 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36766 (ipa_get_ipa_bits_for_value): Declare.
36767 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36768 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36769 (ipa_bits_hash_table): Likewise.
36770 (ipa_vr_ggc_hash_traits): Likewise.
36771 (ipa_vr_hash_table): Likewise.
36772 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36773 being pointers and vr_known being removed.
36774 (ipa_set_jf_unknown): Likewise.
36775 (ipa_get_ipa_bits_for_value): New function.
36776 (ipa_set_jfunc_bits): Likewise.
36777 (ipa_get_value_range): New overloaded functions.
36778 (ipa_set_jfunc_vr): Likewise.
36779 (ipa_compute_jump_functions_for_edge): Use the above functions to
36780 construct bits and vr parts of jump functions.
36781 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36782 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36784 (ipcp_grow_transformations_if_necessary): Also allocate
36785 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36787 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36788 them. Fix too long lines.
36789 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36790 vr_known being removed.
36791 (ipa_read_jump_function): Use new setter functions to construct bits
36792 and vr parts of jump functions or set them to NULL.
36793 (write_ipcp_transformation_info): Adjust for bits being pointers.
36794 (read_ipcp_transformation_info): Likewise.
36795 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36797 Include gt-ipa-prop.h.
36798 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36800 (ipcp_store_bits_results): Likewise.
36801 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36802 Do not write to existing jump functions but use a temporary instead.
36804 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36807 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36808 attempt to use its first operand as BIT_FIELD_REF base.
36810 2017-03-01 Richard Biener <rguenther@suse.de>
36812 PR middle-end/79721
36813 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36814 interpolating formula in wrapping arithmetic.
36815 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36817 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36819 PR tree-optimization/79734
36820 * tree-vect-generic.c (expand_vector_condition): Optimize
36821 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36822 Handle VEC_COND_EXPR where comparison has different inner width from
36823 type's inner width.
36825 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36827 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36828 markup, and similar issues. Remove @opindex entries for things
36829 that aren't options. Add missing -mmpy-option entries.
36831 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36833 PR tree-optimization/79737
36834 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36835 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36836 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36837 instead of byte_size. Formatting fix.
36838 (shift_bytes_in_array_right): Formatting fix.
36840 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36843 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36844 condition on optimize for the leaf function test.
36846 2017-02-28 Martin Liska <mliska@suse.cz>
36849 * read-rtl-function.c (function_reader::handle_unknown_directive):
36850 Bail out when one uses -flto.
36852 2017-02-28 Martin Liska <mliska@suse.cz>
36854 * common.opt: Replace space with tabular for options of <number>
36856 * config/i386/i386.opt: Show <number> value for
36857 -mlarge-data-threshold.
36858 * opts.c (print_filtered_help): Do not display number in hexadecimal
36861 2017-02-28 Martin Liska <mliska@suse.cz>
36863 * common.opt: Fix --help=option -Q for options which are of
36866 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36868 * config/i386/i386.c (print_reg): Error out for values
36869 of 8-bit size in invalid integer register.
36871 2017-02-28 Martin Sebor <msebor@redhat.com>
36873 PR tree-optimization/79691
36874 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36876 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36879 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36880 gcc_unreachable with output_operand_lossage.
36882 2017-02-28 Richard Biener <rguenther@suse.de>
36884 PR tree-optimization/79740
36885 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36887 (visit_nary_op): Insert the nary into the hashtable if we
36888 pattern-matched sth.
36889 * tree-ssa-pre.c (eliminate_insert): Robustify.
36891 2017-02-28 Richard Biener <rguenther@suse.de>
36893 PR middle-end/79731
36894 * fold-const.c (decode_field_reference): Reject out-of-bound
36897 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36899 * config/i386/i386.c: Include intl.h.
36900 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36901 instead of just cond ? "..." : "...".
36902 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36903 * coverage.c (read_counts_file): Likewise.
36904 * omp-offload.c: Include intl.h.
36905 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36906 of just cond ? "..." : "...".
36907 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36908 of just cond ? "..." : "...".
36910 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36913 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36915 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36916 'tune for' CPU name.
36917 * config/arm/arm-cpu-data.h: Regenerated.
36919 2017-02-28 Richard Biener <rguenther@suse.de>
36921 PR tree-optimization/79732
36922 * tree-inline.c (expand_call_inline): Do not shadow var.
36924 2017-02-28 Richard Biener <rguenther@suse.de>
36926 PR tree-optimization/79723
36927 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36928 address-space properly.
36930 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36932 * doc/optinfo.texi (Optimization groups): Fix option used for
36934 * doc/invoke.texi (-fopt-info): Document "omp".
36935 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36936 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36937 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36938 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36939 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36941 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36943 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36944 OPTGROUP_OMP with "omp".
36946 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36949 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36950 for arithmetic shift of unsigned V2DI.
36952 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36954 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36955 arc/linux.h headers.
36956 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36957 (LINK_SPEC): Likewise.
36958 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36959 (EXTRA_SPECS): Likewise.
36960 (STARTFILE_SPEC): Likewise.
36961 (ENDFILE_SPEC): Likewise.
36962 (LIB_SPEC): Likewise.
36963 (TARGET_SDATA_DEFAULT): Likewise.
36964 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36965 (MULTILIB_DEFAULTS): Likewise.
36966 (DWARF2_UNWIND_INFO): Likewise.
36967 * config/arc/big.h: New file.
36968 * config/arc/elf.h: Likewise.
36969 * config/arc/linux.h: Likewise.
36970 * config/arc/t-uClibc: Remove.
36972 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36974 PR tree-optimization/77536
36975 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36976 (tree_transform_and_unroll_loop): Use above function to compute the
36977 estimated niter of unrolled loop and use it when scaling profile.
36978 Also use count info rather than frequency if it's non-zero.
36979 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36980 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36981 (vect_transform_loop): Call above function.
36983 2017-02-27 Richard Biener <rguenther@suse.de>
36985 PR tree-optimization/45397
36986 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36987 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36988 (visit_nary_op): Add pattern matching for CSEing sign-changed
36989 or truncated operations with wider ones.
36991 2017-02-27 Richard Biener <rguenther@suse.de>
36993 PR tree-optimization/79690
36994 * tree-vect-stmts.c (vectorizable_store): Use vector type
36995 built from the DR with address-space.
36997 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36999 * doc/invoke.texi (Optimize Options): Refine the description
37000 of asan-use-after-return.
37002 2017-02-25 Alan Modra <amodra@gmail.com>
37004 PR rtl-optimization/79584
37005 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37006 base, not ad->base_term, the reg within base. Remove assertion
37007 that ad->base == ad->base_term. Replace gen_int_mode using
37008 bogus mode with const0_rtx.
37010 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37012 PR middle-end/79396
37013 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37014 FMA_EXPR like tcc_binary or tcc_unary.
37016 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37019 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37021 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37022 (output_loc_operands): Handle DW_OP_call_ref and
37023 DW_OP_GNU_variable_value.
37024 (struct variable_value_struct): New type.
37025 (struct variable_value_hasher): Likewise.
37026 (variable_value_hash): New variable.
37027 (string_types): Remove.
37028 (copy_loc_descr): New function.
37029 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37030 (prepend_loc_descr_to_each): New function.
37031 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37032 instead of add_loc_descr_to_each if the first argument is single
37033 location list and the second has multiple.
37034 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37035 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37036 when looking for variable value which doesn't have other location info.
37037 (loc_list_from_tree): Formatting fix.
37038 (gen_array_type_die): Simplify DW_AT_string_length handling.
37039 (adjust_string_types): Remove.
37040 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37041 string_types. Call resolve_variable_values.
37042 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37043 (resolve_addr_in_expr): Likewise. Add A argument.
37044 (copy_deref_exprloc): Remove deref argument. Adjust for the
37045 original expression being DW_OP_GNU_variable_value with optionally
37046 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37047 optionally after it.
37048 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37049 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37050 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37051 (variable_value_hasher::hash, variable_value_hasher::equal): New
37053 (resolve_variable_value_in_expr, resolve_variable_value,
37054 resolve_variable_values, note_variable_value_in_expr,
37055 note_variable_value): New functions.
37056 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37059 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37062 * opts.h (handle_generated_option): Add GENERATED_P argument.
37063 * opts-common.c (handle_option): Adjust function comment.
37064 (handle_generated_option): Add GENERATED_P argument, pass it to
37066 (control_warning_option): Pass false to handle_generated_option
37068 * opts.c (maybe_default_option): Pass true to handle_generated_option
37070 * optc-gen.awk: Likewise.
37072 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37074 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37075 a REG, look at the REG it is a SUBREG of.
37076 (splitter for cmpeqsi_t): Ditto.
37078 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37080 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37081 the special USEs with the pattern of the insn, not the insn itself.
37083 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37086 * doc/invoke.texi: Document -mload-store-pairs.
37088 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37089 Sandra Loosemore <sandra@codesourcery.com>
37091 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37092 argument isn't a CONST_INT.
37093 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37094 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37095 (nios2_validate_compare): Bypass alternate compare logic if *op2
37096 is not a CONST_INT.
37097 (ldstwm_operation_p): Return false if first_base is not a REG or
37098 if first_offset is not a CONST_INT.
37100 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37102 * config/cris/cris.md: Use correct operand in a define_peephole2.
37104 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37106 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37108 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37110 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37111 this_insn if it is an INSN or JUMP_INSN.
37112 (force_offsettable): Look at base, not at addr.
37113 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37114 on things that aren't necessarily CONST_INTs.
37116 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37118 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37119 -mfpmath=sse is the default also for x86-32 targets with SSE2
37120 instruction set when @option{-ffast-math} is enabled
37122 2017-02-24 Jeff Law <law@redhat.com>
37124 PR rtl-optimizatoin/79286
37125 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37128 2017-02-24 Richard Biener <rguenther@suse.de>
37130 PR tree-optimization/79389
37131 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37134 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37136 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37137 function comment to reflect reality.
37138 (loop_exits_before_overflow): Fix typo in function description.
37140 2017-02-24 Richard Biener <rguenther@suse.de>
37142 PR tree-optimization/79389
37143 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37144 properly that a threading opportunity exists. Detect conditional
37145 copy/constant propagation opportunities.
37147 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37149 * config/visium/visium.md (type): Add trap.
37150 (b): New mode attribute.
37151 (*btst): Rename into...
37152 (*btst<mode>): ...this and adjust.
37153 (*cbranchsi4_btst_insn): Rename into...
37154 (*cbranch<mode>4_btst_insn): ...this and adjust.
37155 (trap): New define_insn.
37157 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37159 PR tree-optimization/79389
37160 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37161 (noce_reversed_cond_code): New function.
37162 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37163 reversed_comparison_code. Formatting fix.
37164 (noce_try_store_flag): Test rev_cond != NULL in addition to
37165 reversed_comparison_code.
37166 (noce_try_store_flag_constants): Likewise.
37167 (noce_try_store_flag_mask): Likewise.
37168 (noce_try_addcc): Use rev_cond if non-NULL instead of
37169 reversed_comparison_code.
37170 (noce_try_cmove_arith): Likewise. Formatting fixes.
37171 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37172 (noce_find_if_block): Initialize rev_cond.
37173 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37174 instead of false as last argument never attempt to reverse it
37177 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37179 PR tree-optimization/79663
37180 * tree-predcom.c (combine_chains): Process refs in reverse order
37181 only for ZERO length chains, and add explaining comment.
37183 2017-02-23 Jeff Law <law@redhat.com>
37185 PR tree-optimization/79578
37186 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37187 in call to operand_equal_p.
37189 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37192 * config/i386/cpuid.h: Fix another undefined behavior.
37194 2017-02-23 Richard Biener <rguenther@suse.de>
37196 PR tree-optimization/79683
37197 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37198 vector types for data-refs.
37200 2017-02-23 Martin Liska <mliska@suse.cz>
37202 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37204 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37206 PR middle-end/79665
37207 * internal-fn.c (get_range_pos_neg): Moved to ...
37208 * tree.c (get_range_pos_neg): ... here. No longer static.
37209 * tree.h (get_range_pos_neg): New prototype.
37210 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37211 are known to be in between 0 and signed maximum inclusive, try to
37212 expand both unsigned and signed divmod and use the cheaper one from
37215 2017-02-22 Jeff Law <law@redhat.com>
37217 PR tree-optimization/79578
37218 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37219 to compare base operands.
37221 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37224 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37225 gpc_reg_operand instead of fpr_reg_operand.
37227 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37229 * config/mips/mips.c (mips_return_in_memory): Force FP
37230 vector types to be returned in memory for o32 ABI.
37232 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37234 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37235 instead of DW_TAG_member for static data member declarations and don't
37236 set no_linkage_name for static inline data members.
37237 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37240 2017-02-22 Martin Liska <mliska@suse.cz>
37242 * doc/invoke.texi: Replace inequality signs with square brackets
37245 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37248 * lra-constraints.c (simplify_operand_subreg): Handle
37249 WORD_REGISTER_OPERATIONS targets.
37251 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37254 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37255 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37256 elimination by swapping fld*.
37258 2017-02-22 Richard Biener <rguenther@suse.de>
37260 PR tree-optimization/79673
37261 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37262 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37263 irrelevant address-space qualifiers and avoiding a
37264 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37266 2017-02-22 Richard Biener <rguenther@suse.de>
37268 PR tree-optimization/79666
37269 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37270 to not symbolically negate if that may introduce undefined
37273 2017-02-22 Martin Liska <mliska@suse.cz>
37276 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37277 * data-streamer-out.c (streamer_write_gcov_count_stream):
37279 * value-prof.c (stream_out_histogram_value): Make assert more
37280 precise based on type of counter.
37282 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37285 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37286 Use nonimmediate_operand instead of memory_operand for operand 1.
37287 (float-extend standard_x87sse_constant_load splitter): Ditto.
37289 2017-02-21 Jeff Law <law@redhat.com>
37291 PR tree-optimization/79621
37292 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37293 blocks with edges to themselves.
37295 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37298 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37299 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37300 Use gimple_call_builtin_p.
37303 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37304 on temporarily removed DEBUG_INSNs.
37306 PR tree-optimization/79649
37307 * tree-loop-distribution.c (classify_partition): Give up on
37308 non-generic address space loads/stores.
37310 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37312 * doc/loop.texi (Loop manipulation): Remove nonexistent
37313 tree_ssa_loop_version from the documentation.
37314 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37316 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37319 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37320 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37321 * config/rs6000/rs6000.c: Include except.h.
37322 (rs6000_expand_split_stack_prologue): Call
37323 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37325 2017-02-21 Martin Jambor <mjambor@suse.cz>
37328 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37329 have been analyzed.
37331 2017-02-21 Martin Jambor <mjambor@suse.cz>
37333 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37334 for backward compatibility only.
37335 * doc/invoke.texi (Option Summary): Remove all references to
37336 -fipa-cp-alignment.
37338 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37342 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37344 * lra-constraints.c (curr_insn_transform): Handle
37345 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37347 2017-02-21 Martin Liska <mliska@suse.cz>
37349 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37351 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37353 PR translation/79638
37354 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37356 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37359 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37360 (arm_function_ok_for_sibcall): Return false for an indirect call by
37361 descriptor if all the argument registers are used.
37362 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37363 alignment of the function.
37365 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37367 PR tree-optimization/61441
37368 * simplify-rtx.c (simplify_const_unary_operation): For
37369 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37370 the sNaN unmodified.
37372 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37374 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37375 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37376 instead of SYSTEM_HEADER_DIR.
37378 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37379 Martin Liška <mliska@suse.cz>
37381 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37382 Fix typos and grammar, use active voice, and clarify.
37384 2017-02-20 Marek Polacek <polacek@redhat.com>
37386 PR middle-end/79537
37387 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37390 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37392 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37395 * config/i386/i386.c (ix86_expand_builtin): Handle
37396 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37397 ix86_builtins_isa[fcode].isa as a requirement of those
37398 flags and any other flag in the bitmask.
37399 (ix86_init_mmx_sse_builtins): Use 0 instead of
37400 ~OPTION_MASK_ISA_64BIT as mask.
37401 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37402 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37403 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37404 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37406 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37409 * lra-constraints.c (split_reg): Check requested split mode
37410 is supported by the register.
37412 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37414 * lra-constraints.c (simplify_operand_subreg): Remove early
37417 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37420 * lra-constraints.c (curr_insn_transform): Tighten condition
37421 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37423 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37426 * lra-constraints.c (curr_insn_transform): Handle
37427 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37429 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37432 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37434 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37436 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37439 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37442 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37444 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37447 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37449 * final.c (last_columnnum, override_columnnum): New variables.
37450 (final_start_function): Set last_columnnum, pass it to begin_prologue
37451 hook and pass 0 to dwarf2out_begin_prologue.
37452 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37453 to source_line debug hook.
37454 (notice_source_line): Compute last_columnnum and for debug_column_info
37455 return true on column changes.
37456 * debug.h (struct gcc_debug_hooks): Add column argument to
37457 source_line and begin_prologue hooks.
37458 (debug_nothing_int_charstar_int_bool): Remove prototype.
37459 (debug_nothing_int_int_charstar,
37460 debug_nothing_int_int_charstar_int_bool): New prototypes.
37461 (dwarf2out_begin_prologue): Add column argument.
37462 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37463 begin_prologue hooks.
37464 (debug_nothing_int_charstar_int_bool): Remove.
37465 (debug_nothing_int_int_charstar,
37466 debug_nothing_int_int_charstar_int_bool): New functions.
37467 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37468 through to dwarf2out_source_line.
37469 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37470 (dwarf2out_source_line): Add column argument, emit it if requested.
37471 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37473 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37474 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37475 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37476 through to dwarf2out_begin_prologue.
37477 (vmsdbgout_source_line): Add column argument, pass it through to
37478 dwarf2out_source_line.
37479 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37480 dbxout_source_line caller.
37481 (dbxout_source_line): Add column argument.
37483 * common.opt (gno-column-info, gcolumn-info): New options.
37484 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37485 (check_die): Also test for multiple DW_AT_decl_column attributes.
37486 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37487 DW_AT_decl_column if requested.
37488 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37490 (gen_variable_die): Likewise.
37491 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37492 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37495 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37496 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37497 (ix86_handle_option): Handle OPT_m3dnowa.
37498 * doc/invoke.texi (-m3dnowa): Document.
37499 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37500 -m3dnowa instead of -m3dnow -march=athlon.
37503 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37504 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37506 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37509 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37510 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37511 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37512 generator for vsx_xxpermdi_<mode>_be.
37513 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37514 force big-endian semantics.
37515 (vsx_xxpermdi_<mode>_be): New define_expand with same
37516 implementation as previous version of vsx_xxpermdi_<mode>.
37518 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37520 PR tree-optimization/79327
37521 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37522 variable, its initialization and use.
37524 2017-02-17 Julia Koval <julia.koval@intel.com>
37526 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37527 (OPTION_MASK_ISA_PKU_UNSET): New.
37528 (ix86_handle_option): Handle -mrdpid.
37529 * config/i386/cpuid.h (bit_RDPID): New.
37530 * config/i386/driver-i386.c (host_detect_local_cpu):
37531 Detect RDPID feature.
37532 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37533 * config/i386/i386-c.c (ix86_target_macros_internal):
37535 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37536 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37537 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37538 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37539 * config/i386/i386.md (define_insn "rdpid"): New.
37540 * config/i386/i386.opt Add -mrdpid.
37541 * config/i386/immintrin.h (_rdpid_u32): New.
37543 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37545 PR rtl-optimization/79541
37546 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37547 instead of transforming it into USE.
37549 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37551 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37552 If HONOR_SNANS (SFmode) force the input to a register.
37553 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37554 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37555 an frsp or similar insn.
37557 2017-02-17 Martin Liska <mliska@suse.cz>
37559 PR rtl-optimization/79577
37560 * params.def (selsched-max-sched-times): Increase minimum to 1.
37562 2017-02-17 Martin Liska <mliska@suse.cz>
37564 PR rtl-optimization/79574
37565 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37567 2017-02-17 Martin Liska <mliska@suse.cz>
37569 PR tree-optimization/79529
37570 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37571 ssa_defined_default_def_p to handle cases which are implicitly
37573 * tree-ssa.c (ssa_defined_default_def_p): New function.
37574 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37575 which are implicitly defined.
37576 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37578 2017-02-17 Richard Biener <rguenther@suse.de>
37580 PR middle-end/79576
37581 * params.def (max-ssa-name-query-depth): Limit to 10.
37583 2017-02-17 Richard Biener <rguenther@suse.de>
37585 PR tree-optimization/79552
37586 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37589 2017-02-17 Richard Biener <rguenther@suse.de>
37592 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37594 2017-02-17 Marek Polacek <polacek@redhat.com>
37596 PR middle-end/79536
37597 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37598 (fold_negate_expr): New wrapper.
37600 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37602 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37603 Correct terminology and de-emphasize pre-standard behavior.
37605 2017-02-16 Alan Modra <amodra@gmail.com>
37607 PR rtl-optimization/79286
37608 * ira.c (def_dominates_uses): New function.
37609 (update_equiv_regs): Don't create an equivalence for insns that
37610 may trap where the register def does not dominate the use.
37612 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37614 PR rtl-optimization/78127
37615 * lra.c (lra): Call lra_eliminate before finish the loop after
37618 2017-02-16 Richard Biener <rguenther@suse.de>
37620 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37622 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37623 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37624 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37625 (isl_val_int_from_wi): New function.
37626 (extract_affine_gmp): Rename to ...
37627 (extract_affine_wi): ... this, take a widest_int.
37628 (extract_affine_int): Just wrap extract_affine_wi.
37629 (add_param_constraints): Use isl_val_int_from_wi.
37630 (add_loop_constraints): Likewise, and extract_affine_wi.
37632 2017-02-15 Jeff Law <law@redhat.com>
37634 PR middle-end/79521
37635 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37636 ira_init_register_move_cost_if_necessary.
37638 2017-02-15 Martin Sebor <msebor@redhat.com>
37640 PR middle-end/32003
37641 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37642 removed in a prior commit.
37644 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37646 PR tree-optimization/79347
37647 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37648 counters during peeling.
37650 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37652 * Makefile.in (site.exp): Remove "set ISLVER".
37654 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37657 * real.c (real_from_integer): Call real_convert even for decimal.
37659 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37662 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37664 2017-02-14 Andrew Pinski <apinski@cavium.com>
37666 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37667 cores and change the partno/implementer to be correct.
37668 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37669 the 'B" as the implementer.
37670 * config/aarch64/aarch64-tune.md: Regenerate.
37672 2017-02-14 Carl Love <cel@us.ibm.com>
37674 * config/rs6000/rs6000.c: Add case statement entry to make the
37675 xvcvuxdsp built-in argument unsigned.
37676 * config/rs6000/vsx.md: Fix the source and return operand types so they
37677 match the instruction definitions from the ISA document. Fix typo
37678 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37681 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37684 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37685 member early_clobber_alts.
37686 * lra-lives.c (reg_early_clobber_p): New.
37687 (process_bb_lives): Use it.
37688 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37689 (debug_operand_data): Initialize early_clobber_alts.
37690 (setup_operand_alternative): Set up early_clobber_alts.
37691 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37692 alternatives to new_insn_reg.
37693 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37695 (lra_update_insn_regno_info): Pass the new arg.
37697 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37699 PR middle-end/79505
37700 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37701 (new_oacc_loop_raw): Don't clear already cleared fields.
37704 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37705 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37706 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37708 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37711 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37713 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37716 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37717 the extra instruction to the right place to store 128-bit constant
37720 2017-02-14 Martin Sebor <msebor@redhat.com>
37722 PR middle-end/79448
37723 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37724 warning for strings of unknown length.
37726 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37728 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37730 2017-02-14 Jeff Law <law@redhat.com>
37733 * ira-costs.c (scan_one_insn): Initialize register move costs
37734 for pseudos seen in USE/CLOBBER insns.
37736 PR tree-optimization/79095
37737 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37738 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37739 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37740 if the operands are known to be not equal, then the resulting range
37742 (intersect_ranges): If the new range is ~[0,0] and the old range is
37743 wide, then prefer ~[0,0].
37744 * tree-vrp.c (overflow_comparison_p_1): New function.
37745 (overflow_comparison_p): New function.
37746 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37747 if NAME is used in an overflow test.
37748 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37749 overflow check that can be expressed as an equality test, then adjust
37750 ops to be that equality test.
37752 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37754 * config/s390/s390-builtin-types.def: Remove flags argument.
37755 * config/s390/s390.c (s390_init_builtins): Likewise.
37757 2017-02-14 Martin Liska <mliska@suse.cz>
37759 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37760 vector. Fix trailing white spaces.
37762 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37764 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37767 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37769 PR rtl-optimization/68664
37770 * config/arm/arm.c (arm_sched_can_speculate_insn):
37771 New function. Declare prototype.
37772 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37774 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37776 PR rtl-optimization/68664
37777 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37779 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37781 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37783 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37784 max skip bytes for function, loop and jump.
37786 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37788 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37789 ABS_EXPR for gimple dump.
37791 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37794 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37796 PR tree-optimization/79408
37797 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37798 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37799 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37800 also if rhs1 is INTEGER_CST.
37802 2017-02-14 Richard Biener <rguenther@suse.de>
37804 PR middle-end/79432
37805 * tree-into-ssa.c (insert_phi_nodes): When the function can
37806 have abnormal edges rewrite SSA names with broken use-def
37807 dominance out of SSA and register them for PHI insertion.
37809 2017-02-13 Martin Sebor <msebor@redhat.com>
37811 PR middle-end/79496
37812 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37813 clearing info.nowrite flag when snprintf size argument is a range.
37815 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37817 * cprop.c (cprop_jump): Add missing space in string literal.
37818 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37819 (get_constraint_for_component_ref): Likewise.
37820 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37821 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37822 * lra-constraints.c (process_alt_operands): Likewise.
37823 * ipa-inline.c (inline_small_functions): Likewise.
37824 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37825 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37826 * trans-mem.c (diagnose_tm_1_op): Likewise.
37827 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37828 (grid_parallel_clauses_gridifiable): Likewise.
37830 * config/nvptx/mkoffload.c (process): Add space in between
37833 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37834 "MOD4_SSE_REGS" and "ALL_REGS".
37836 * spellcheck.c (test_data): Add , in between "foo" and "food".
37838 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37841 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37842 boundary crossing check and subsequent code generation agree.
37844 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37846 * config/aarch64/aarch64.c (has_memory_op): Delete.
37847 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37850 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37852 PR rtl-optimization/79388
37853 PR rtl-optimization/79450
37854 * combine.c (distribute_notes): When removing TEM_INSN for which
37855 corresponding dest has last value recorded, invalidate that last
37858 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37860 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37861 of explicit '@'. Add missing assembly comment marker on branch costs
37864 2017-02-13 Nathan Sidwell <nathan@acm.org>
37866 * gengtype-lex.l (<in_struct>): Add '/'.
37868 2017-02-13 Martin Liska <mliska@suse.cz>
37871 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37873 2017-02-13 Richard Biener <rguenther@suse.de>
37875 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37877 * configure: Re-generate.
37878 * config.in: Likewise.
37879 * graphite-dependences.c: Simplify as if
37880 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37881 * graphite-isl-ast-to-gimple.c: Likewise.
37882 * graphite-optimize-isl.c: Likewise.
37883 * graphite-poly.c: Likewise.
37884 * graphite-sese-to-poly.c: Likewise.
37885 * graphite.h: Likewise.
37886 * toplev.c: Include isl/version.h and use isl_version () for
37887 printing the ISL version.
37888 * doc/install.texi: Update ISL requirement.
37890 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37892 * doc/standards.texi (Standards): Update reference to
37895 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37897 * doc/extend.texi (Named Address Spaces): sourceware.org now
37899 * doc/install.texi (Binaries): Ditto.
37902 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37904 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37905 standard terminology "stringize"/"stringizing" throughout.
37906 * doc/cppinternals.texi: Likewise.
37908 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37910 * doc/extend.texi: Fix some spelling mistakes and typos.
37911 * doc/invoke.texi: Likewise.
37913 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37916 * params.def (inline-min-speedup) Change from 10 to 8.
37918 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37920 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37923 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37926 * ipa-inline-analysis.c (get_minimal_bb): New function.
37927 (record_modified): Use it.
37928 (remap_edge_change_prob): Handle also ancestor functions.
37930 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37932 * doc/contrib.texi (Contributors): Remove broken link into
37933 the Mauve CVS repository.
37935 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37937 PR middle-end/79454
37938 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37939 result computation whenever lhs doesn't have vector mode, not
37940 just when it has BLKmode.
37942 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37944 * doc/makefile.texi (profiledbootstrap): Refer to the
37945 installation instructions only in textual form.
37947 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37950 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37952 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37954 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37955 (Specific): Update mingw-w64 reference.
37957 (Specific): Remove broken link to Renesas RX processor.
37959 2017-02-10 Richard Biener <rguenther@suse.de>
37961 * toplev.c (process_options): Do not mention obsolete graphite
37962 options when printing sorry message about missing graphite support.
37963 Mention -floop-nest-optimize.
37965 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37967 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37968 (vtst_p16): Likewise.
37969 (vtstq_p8): Likewise.
37970 (vtstq_p16): Likewise.
37972 (vtstq_p64): Likewise.
37973 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37974 (vset_lane_p64): New.
37975 (vsetq_lane_p64): New.
37977 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37979 PR tree-optimization/79411
37980 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37981 stmt operands are SSA_NAMEs used in abnormal phis.
37982 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37985 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37988 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37991 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37993 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37995 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37997 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38001 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38002 not arbitrary TREE_CONSTANT.
38005 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38006 "omp declare target link" attribute unless is_global_var.
38007 * omp-offload.c (find_link_var_op): Likewise.
38009 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38010 Chung-Lin Tang <cltang@codesourcery.com>
38012 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38014 (gimplify_adjust_omp_clauses): Don't delete TILE.
38015 (gimplify_omp_for): Deal with TILE.
38016 * internal-fn.c (expand_GOACC_TILE): New function.
38017 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38019 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38020 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38022 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38023 avoid DIV for outermost collapse var.
38024 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38025 Remove out of date comments, fix whitespace.
38026 * omp-general.c (omp_extract_for_data): Deal with tiling.
38027 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38028 adjust OLF_DIM_BASE value.
38029 (struct omp_for_data): Add tiling field.
38030 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38031 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38032 for auto loops. Remove default auto determining, moved to
38033 oacc_loop_fixed_partitions.
38034 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38035 stmts, add e_mask field.
38036 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38037 (oacc_thread_numbers): Use oacc_dim_call.
38038 (oacc_xform_tile): New.
38039 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38040 (finish_oacc_loop): Adjust for ifns vector.
38041 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38042 add case for GOACC_TILE fns.
38043 (oacc_loop_xform_loop): Delete.
38044 (oacc_loop_process): Iterate over call list directly, and add
38045 handling for GOACC_TILE fns.
38046 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38048 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38049 vector partitioning to outer loops. Assign 2 partitions to loops
38050 when available. Add TILE handling.
38051 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38052 (execite_oacc_device_lower): Process GOACC_TILE fns,
38053 ignore unknown specs.
38054 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38055 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38056 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38058 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38060 * configure.ac (ACX_BUGURL): Update.
38061 * configure: Regenerate.
38063 2017-02-09 Richard Biener <rguenther@suse.de>
38065 PR tree-optimization/69823
38066 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38067 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38069 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38071 * config/arc/arc-c.def: Add __NPS400__ definition.
38072 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38073 (TARGET_NPS400): Define.
38075 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38077 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38079 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38080 pointer, arch_info.
38081 (arc_cpu_types): Fill the arch_info field with a pointer into the
38082 arc_arch_types table.
38083 (arc_selected_cpu): Declare.
38084 * config/arc/arc.c (arc_selected_cpu): Make global.
38085 (arc_selected_arch): Delete.
38086 (arc_base_cpu): Delete.
38087 (arc_override_options): Remove references to deleted variables,
38088 update access to arch information.
38089 (ARC_OPT): Update access to arch information.
38090 (ARC_OPTX): Likewise.
38091 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38092 (TARGET_ARC600): Update access to arch information.
38093 (TARGET_ARC601): Likewise.
38094 (TARGET_ARC700): Likewise.
38095 (TARGET_EM): Likewise.
38096 (TARGET_HS): Likewise.
38097 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38100 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38103 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38104 condition/operands for integer GE/LE/GEU/LEU operations.
38106 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38108 PR translation/79397
38109 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38112 2017-02-08 Martin Jambor <mjambor@suse.cz>
38115 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38116 whether allocation happened.
38117 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38118 nothing was allocated.
38120 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38122 PR tree-optimization/79408
38123 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38124 constant, but SSA_NAME with a known integer range, use the minimum
38125 of that range instead of op1 to determine if modulo can be replaced
38126 with its first operand.
38128 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38130 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38132 2017-02-08 Richard Biener <rguenther@suse.de>
38134 PR tree-optimization/71824
38135 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38136 Check all loops contained in the merged region.
38138 2017-02-07 Andrew Pinski <apinski@cavium.com>
38140 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38142 2017-02-07 Andrew Pinski <apinski@cavium.com>
38144 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38145 (thunderxt88): Likewise.
38146 (thunderxt81): Disable LSE and change v8.1 to v8.
38147 (thunderxt83): Likewise.
38149 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38150 Richard Biener <rguenther@suse.de>
38152 PR middle-end/79399
38153 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38154 type from int to size_t.
38155 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38157 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38159 PR rtl-optimization/79386
38160 * cprop.c (bypass_conditional_jumps): Initialize
38161 bypass_last_basic_block already before splitting bbs after
38162 unconditional traps...
38163 (bypass_conditional_jumps): ... rather than here.
38166 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38167 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38168 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38169 fix -masm=intel patterns.
38171 2017-02-07 Richard Biener <rguenther@suse.de>
38173 PR tree-optimization/79256
38174 PR middle-end/79278
38175 * builtins.c (get_object_alignment_2): Use min_align_of_type
38176 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38177 and ADJUST_FIELD_ALIGN.
38179 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38181 * doc/tm.texi: Regenerate.
38182 * stor-layout.c (layout_decl): Adjust.
38183 (update_alignment_for_field): Likewise.
38184 (place_field): Likewise.
38185 (min_align_of_type): Likewise.
38186 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38187 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38188 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38189 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38190 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38191 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38192 * config/i386/i386.c (x86_field_alignment): Likewise.
38193 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38194 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38195 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38196 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38197 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38198 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38202 2017-01-30 Richard Biener <rguenther@suse.de>
38204 PR tree-optimization/79256
38205 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38206 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38209 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38211 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38212 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38213 builtins to SImode and emit a zero-extend, if necessary.
38215 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38217 * docs/invoke.texi (RISC-V Options): Alphabetize.
38219 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38221 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38224 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38226 * config/riscv/riscv.c: New file.
38227 * common/config/riscv/riscv-common.c: Likewise.
38228 * config.gcc: Likewise.
38229 * config/riscv/constraints.md: Likewise.
38230 * config/riscv/elf.h: Likewise.
38231 * config/riscv/generic.md: Likewise.
38232 * config/riscv/linux.h: Likewise.
38233 * config/riscv/multilib-generator: Likewise.
38234 * config/riscv/peephole.md: Likewise.
38235 * config/riscv/pic.md: Likewise.
38236 * config/riscv/predicates.md: Likewise.
38237 * config/riscv/riscv-builtins.c: Likewise.
38238 * config/riscv/riscv-c.c: Likewise.
38239 * config/riscv/riscv-ftypes.def: Likewise.
38240 * config/riscv/riscv-modes.def: Likewise.
38241 * config/riscv/riscv-opts.h: Likewise.
38242 * config/riscv/riscv-protos.h: Likewise.
38243 * config/riscv/riscv.h: Likewise.
38244 * config/riscv/riscv.md: Likewise.
38245 * config/riscv/riscv.opt: Likewise.
38246 * config/riscv/sync.md: Likewise.
38247 * config/riscv/t-elf-multilib: Likewise.
38248 * config/riscv/t-linux: Likewise.
38249 * config/riscv/t-linux-multilib: Likewise.
38250 * config/riscv/t-riscv: Likewise.
38251 * configure.ac: Likewise.
38252 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38253 Waterman as RISC-V maintainers.
38254 * doc/install.texi: Add RISC-V entries.
38255 * doc/invoke.texi: Add RISC-V options section.
38256 * doc/md.texi: Add RISC-V constraints section.
38257 * configure: Regenerated.
38259 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38262 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38263 false values to be constant vectors with all 0 or all 1 bits set.
38264 (vcondu<mode><mode>): Likewise.
38265 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38267 (fpmask_comparison_operator): Update comment.
38268 (vecint_comparison_operator): New predicate.
38269 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38270 vector conditionals when the true and false values are constant
38271 vectors with all 0 bits or all 1 bits set.
38273 2017-02-06 Martin Sebor <msebor@redhat.com>
38275 PR tree-optimization/79376
38276 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38278 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38280 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38281 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38282 to simplify split condition.
38284 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38286 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38287 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38290 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38292 PR rtl-optimization/68664
38293 * target.def (can_speculate_insn): New hook.
38294 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38295 * doc/tm.texi: Regenerate.
38296 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38297 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38298 (rs6000_sched_can_speculate_insn): New function.
38300 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38302 PR tree-optimization/79284
38303 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38304 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38305 vectorizable_mask_load_store, vectorizable_operation,
38306 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38307 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38308 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38309 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38310 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38311 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38312 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38313 is_gimple_assign (stmt). Replace another such test with
38314 is_gimple_assign (stmt).
38316 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38319 * config/avr/avr.c (rtl-iter.h): Include it.
38320 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38321 (avr_legitimate_combined_insn): ...and implementation.
38323 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38325 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38326 * config/s390/s390.c (s390_const_operand_ok)
38327 (s390_canonicalize_comparison, s390_extract_part)
38328 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38329 (s390_contiguous_bitmask_p, s390_rtx_costs)
38330 (legitimize_pic_address): Likewise.
38331 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38332 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38333 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38334 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38335 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38337 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38339 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38340 REGNO($0) == REGNO($1).
38342 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38344 * config/s390/linux.h(SIZE_TYPE): Add comment.
38346 2017-02-06 Julian Brown <julian@codesourcery.com>
38347 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38348 Virendra Pathak <virendra.pathak@broadcom.com>
38350 * config/aarch64/aarch64-cores.def: Change the scheduler
38352 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38353 * config/aarch64/thunderx2t99.md: New file.
38355 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38357 * doc/standards.texi (Go Language): Update link to language
38360 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38362 * tree-eh.c (lower_resx): Sanitize profile.
38363 (cleanup_empty_eh_move_lp): Likewise.
38365 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38368 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38370 * cfgloopmanip.h (loop_version): Update prototype.
38371 * modulo-sched.c (sms_schedule): Update call of loop_version.
38372 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38373 * tree-parloops.c (gen_parallel_loop): Likewise.
38374 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38375 * tree-ssa-loop-split.c (split_loop): Likewise.
38376 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38377 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38379 2017-02-05 Martin Liska <mliska@suse.cz>
38382 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38384 (print_operand_address): Initialize a struct to zero.
38386 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38388 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38389 garbage collector only in textual form.
38391 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38393 * doc/extend.texi (x86 specific memory model extensions for
38394 transactional memory): Simplify a phrase.
38396 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38399 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38400 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38401 (atomic_storedi_1): Likewise.
38403 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38405 PR tree-optimization/79338
38406 * tree-parloops.c (gather_scalar_reductions): Don't call
38407 vect_analyze_loop_form for loop->inner before destroying loop's
38410 2017-02-03 Martin Sebor <msebor@redhat.com>
38412 PR tree-optimization/79327
38413 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38414 when precision has resulted in leading zeros.
38415 (format_integer): Adjust the likely counter to assume an unknown
38416 argument that may be zero is non-zero.
38418 2017-02-03 Jason Merrill <jason@redhat.com>
38421 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38422 avoid copying non-taken branch.
38424 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38426 PR tree-optimization/79340
38427 * tree-vect-loop.c (vectorizable_reduction): Release
38428 vec_defs elements after safe_splicing them into other vectors.
38431 PR tree-optimization/79327
38432 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38433 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38435 (format_integer): Use wide_int_to_tree instead of build_int_cst
38436 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38437 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38438 of shortest and longest sequence.
38440 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38442 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38443 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38445 2017-02-03 Walter Lee <walt@tilera.com>
38448 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38449 after initial stackframe link reg save.
38450 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38452 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38455 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38456 wu for stxssp alternative.
38458 2017-02-03 Martin Sebor <msebor@redhat.com>
38460 PR tree-optimization/79352
38461 * gimple-fold.c (get_range_strlen): Add argument.
38462 (get_range_strlen): Change return type to bool.
38463 (get_maxval_strlen): Pass in a dummy argument.
38464 * gimple-fold.h (get_range_strlen): Change return type to bool.
38465 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38466 * tree.h (array_at_struct_end_p): Add argument.
38467 * tree.c (array_at_struct_end_p): Handle it.
38469 2017-02-03 Martin Liska <mliska@suse.cz>
38472 * multiple_target.c (create_dispatcher_calls): Redirect edge
38473 from a caller of a dispatcher.
38474 (expand_target_clones): Make the clones local.
38475 (ipa_target_clone): Do both target clones and resolvers.
38476 (ipa_dispatcher_calls): Remove the pass.
38477 (pass_dispatcher_calls::gate): Likewise.
38478 (make_pass_dispatcher_calls): Likewise.
38479 * passes.def (pass_target_clone): Put as very first IPA early
38482 2017-02-03 Martin Liska <mliska@suse.cz>
38484 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38485 in case of a function with ifunc attribute.
38487 2017-02-03 Martin Liska <mliska@suse.cz>
38489 * cgraph.c (cgraph_node::dump): Dump function version info.
38490 * symtab.c (symtab_node::dump_base): Add missing new line.
38492 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38494 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38495 (ifcombine_ifandif): Use it.
38497 2017-02-03 Martin Liska <mliska@suse.cz>
38499 * doc/invoke.texi: Document default value for
38500 use-after-scope-direct-emission-threshold.
38502 2017-02-03 Martin Liska <mliska@suse.cz>
38504 PR tree-optimization/79339
38505 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38506 (format_floating): Likewise.
38508 2017-02-03 Martin Liska <mliska@suse.cz>
38511 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38513 (ipa_node_params_t::remove): Likewise.
38514 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38515 initialization from removed ipa_node_params_t::insert.
38516 (ipa_node_params::~ipa_node_params): Move from removed
38517 ipa_node_params_t::release.
38518 * symbol-summary.h (symbol_summary::m_released): New member.
38519 Do not release a summary twice. Do not allow to call finalizer
38520 for types of a summary that live in GGC memory.
38522 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38524 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38527 2017-02-02 Martin Sebor <msebor@redhat.com>
38529 PR middle-end/79275
38530 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38531 (format_string): Tighten up the range of output for non-constant
38532 strings and correct the expected range for wide non-constant strings.
38534 2017-02-02 Martin Sebor <msebor@redhat.com>
38536 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38538 PR middle-end/32003
38539 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38541 (-fdump-tree-@var): Add to index and document how to come up
38542 with pass-specific option and dump file names.
38543 (-fdump-passes): Clarify where to look for output.
38545 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38547 PR middle-end/77445
38548 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38549 statistics of the analyzed path; allow threading for speed when
38550 any of BBs along the path are optimized for speed.
38552 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38554 PR middle-end/78468
38555 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38556 settings of the virtual registers.
38559 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38561 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38562 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38565 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38567 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38568 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38570 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38572 * config/s390/s390.md: Add missing comments with the expanded
38574 * config/s390/vector.md: Likewise.
38575 * config/s390/vx-builtins.md: Likewise.
38577 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38580 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38581 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38582 conditions on a single line.
38584 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38586 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38587 __S390_VX__ to __VX__.
38589 2017-02-01 Andrew Pinski <apinski@cavium.com>
38591 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38592 stmt_info to record_stmt_cost.
38593 (vect_get_known_peeling_cost): Pass stmt_info if known to
38595 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38596 cpu_vector_cost field into
38597 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38598 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38599 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38600 splitting of scalar_stmt_cost and vec_stmt_cost.
38601 (thunderx_vector_cost): Likewise.
38602 (cortexa57_vector_cost): LIkewise.
38603 (exynosm1_vector_cost): Likewise.
38604 (xgene1_vector_cost): Likewise.
38605 (thunderx2t99_vector_cost): Improve after the splitting of the two
38607 (aarch64_builtin_vectorization_cost): Update for the splitting of
38608 scalar_stmt_cost and vec_stmt_cost.
38610 2017-02-01 Torvald Riegel <triegel@redhat.com>
38611 Richard Henderson <rth@redhat.com>
38613 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38614 conditional on existance of a fast atomic load.
38615 * optabs-query.c (can_atomic_load_p): New function.
38616 * optabs-query.h (can_atomic_load_p): Declare it.
38617 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38618 no fast atomic load is available for the particular size of access.
38619 (expand_atomic_compare_and_swap): Likewise.
38620 (expand_atomic_load): Likewise.
38621 (expand_atomic_store): Likewise.
38622 (expand_atomic_fetch_op): Likewise.
38623 * testsuite/lib/target-supports.exp
38624 (check_effective_target_sync_int_128): Remove x86 because it provides
38625 no fast atomic load.
38626 (check_effective_target_sync_int_128_runtime): Likewise.
38628 2017-02-01 Richard Biener <rguenther@suse.de>
38630 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38631 (graphite_transform_loops): Provide opt-info for optimized nests.
38632 * tree-parloop.c (parallelize_loops): Provide opt-info for
38633 parallelized loops.
38635 2017-02-01 Richard Biener <rguenther@suse.de>
38637 PR middle-end/79315
38638 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38639 was not set before.
38641 2017-02-01 Richard Biener <rguenther@suse.de>
38643 PR tree-optimization/71824
38644 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38645 Verify the loops are valid in the merged SESE region.
38646 (scop_detection::can_represent_loop_1): Check analyzing the
38647 evolution of the number of iterations in the region succeeds.
38649 2017-01-31 Ian Lance Taylor <iant@golang.org>
38651 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38652 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38654 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38656 PR preprocessor/79210
38657 * input.c (get_substring_ranges_for_loc): Replace line_width
38658 assertion with error-handling.
38660 2017-01-31 Richard Biener <rguenther@suse.de>
38662 PR tree-optimization/77318
38663 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38664 (create_pw_aff_from_tree): Take loop parameter.
38665 (add_condition_to_pbb): Pass loop of the condition to
38666 create_pw_aff_from_tree.
38668 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38670 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38671 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38673 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38677 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38679 (convert_int_to_float128): Likewise.
38680 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38681 (convert_int_to_float128): Likewise.
38682 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38683 (UNSPEC_IEEE128_CONVERT): Likewise.
38684 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38685 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38686 Use local variables for IBM extended format.
38687 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38688 (fix_trunc<mode>si2_fprs): Likewise.
38689 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38690 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38691 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38692 to know that we can now have integers of all sizes in vector
38694 (fix<uns>_<mode>di2_hw): Likewise.
38695 (float<uns>_<mode>si2_hw): Likewise.
38696 (fix_<mode>si2_hw): Likewise.
38697 (fixuns_<mode>si2_hw): Likewise.
38698 (float<uns>_<mode>di2_hw): Likewise.
38699 (float_<mode>di2_hw): Likewise.
38700 (float_<mode>si2_hw): Likewise.
38701 (floatuns_<mode>di2_hw): Likewise.
38702 (floatuns_<mode>si2_hw): Likewise.
38703 (xscvqp<su>wz_<mode>): Delete, no longer used.
38704 (xscvqp<su>dz_<mode>): Likewise.
38705 (xscv<su>dqp_<mode>): Likewise.
38706 (ieee128_mfvsrd_64bit): Likewise.
38707 (ieee128_mfvsrd_32bit): Likewise.
38708 (ieee128_mfvsrwz): Likewise.
38709 (ieee128_mtvsrw): Likewise.
38710 (ieee128_mtvsrd_64bit): Likewise.
38711 (ieee128_mtvsrd_32bit): Likewise.
38713 2017-01-31 Martin Liska <mliska@suse.cz>
38716 * ipa-prop.c (ipa_free_all_node_params): Call release method
38717 instead of ~sumbol_summary to not to trigger double times
38720 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38722 PR tree-optimization/71691
38723 * bitmap.h (class auto_bitmap): New.
38724 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38725 is_maybe_undefined instead of ssa_undefined_value_p.
38727 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38729 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38730 __S390_ARCH_LEVEL__ to __ARCH__.
38732 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38734 PR tree-optimization/79267
38735 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38736 if should_remove_lhs_p is true.
38738 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38741 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38742 (add_alignment_attribute): New.
38743 (base_type_die): Add alignment attribute.
38744 (subrange_type_die): Likewise.
38745 (modified_type_die): Likewise.
38746 (gen_array_type_die): Likewise.
38747 (gen_descr_array_type_die: Likewise.
38748 (gen_enumeration_type_die): Likewise.
38749 (gen_subprogram_die): Likewise.
38750 (gen_variable_die): Likewise.
38751 (gen_field_die): Likewise.
38752 (gen_ptr_to_mbr_type_die): Likewise.
38753 (gen_struct_or_union_type_die): Likewise.
38754 (gen_subroutine_type_die): Likewise.
38755 (gen_typedef_die): Likewise.
38756 (base_type_cmp): Compare alignment attribute.
38758 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38761 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38762 (setb_unsigned) New pattern for setb with CCUNS.
38763 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38764 subfc./subfe sequence to avoid overflow problems. Generate a
38765 shorter sequence with cmpld/setb for power9.
38766 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38767 for generating subfc. instruction.
38768 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38769 now uses this instruction.
38771 2017-01-30 Ian Lance Taylor <iant@google.com>
38774 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38775 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38777 2017-01-30 Martin Sebor <msebor@redhat.com>
38779 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38780 Move constant to the right of a relational operator.
38781 (get_mpfr_format_length, format_character, format_string): Ditto.
38782 (should_warn_p, maybe_warn): Same.
38784 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38786 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38789 * asan.c (get_translation_unit_decl): Remove function.
38790 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38792 2017-01-30 Martin Liska <mliska@suse.cz>
38794 PR gcov-profile/79259
38795 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38796 -fprofile-generate.
38798 2017-01-30 Martin Liska <mliska@suse.cz>
38801 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38802 Initialize variables with NULL value.
38804 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38807 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38809 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38811 2017-01-30 Richard Biener <rguenther@suse.de>
38813 PR tree-optimization/79276
38814 * tree-vrp.c (process_assert_insertions): Properly adjust common
38815 when removing a duplicate.
38817 2017-01-30 Richard Biener <rguenther@suse.de>
38819 PR tree-optimization/79256
38820 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38821 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38823 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38825 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38828 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38829 ("*r<noxa>sbg_<mode>_sll_bitmask")
38830 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38831 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38832 Use contiguous_bitmask_nowrap_operand.
38834 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38837 * config/rs6000/altivec.h (vec_xl): Revise #define.
38838 (vec_xst): Likewise.
38840 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38842 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38844 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38846 PR rtl-optimization/79194
38847 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38848 traps before call to bypass_conditional_jumps.
38850 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38852 PR tree-optimization/71374
38853 * lra-constraints.c (check_conflict_input_operands): New.
38854 (match_reload): Use it.
38856 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38859 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38860 account to calculate conflict_set.
38862 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38864 PR rtl-optimization/78559
38865 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38866 other_insn in combine.
38868 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38870 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38871 uint16_type_node for BT_UINT16.
38873 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38875 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38876 "RTL Tests" to menu.
38877 (GIMPLE Tests): New node.
38878 (RTL Tests): New node.
38880 2017-01-27 Richard Biener <rguenther@suse.de>
38882 PR tree-optimization/79245
38883 * tree-loop-distribution.c (distribute_loop): Apply cost
38884 modeling also to detected patterns.
38886 2017-01-27 Richard Biener <rguenther@suse.de>
38888 PR tree-optimization/71433
38889 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38890 (compare_assert_loc): New function.
38891 (process_assert_insertions): Sort and optimize assert locations
38892 to remove duplicates and push down identical assertions on
38893 edges to their destination block.
38895 2017-01-27 Richard Biener <rguenther@suse.de>
38897 PR tree-optimization/79244
38898 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38899 out SSA names even if abnormal.
38901 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38903 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38904 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38905 instead of MPFR_RNDN.
38907 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38910 * arm.c (arm_option_override): Don't call build_target_option_node
38911 until after doing all option overrides.
38912 (arm_valid_target_attribute_tree): Likewise.
38914 2017-01-27 Martin Liska <mliska@suse.cz>
38916 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38917 for {cd}tors and C++ {cd}tors.
38919 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38921 * config/s390/s390.md ("*setmem_long_and")
38922 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38924 2017-01-26 Martin Sebor <msebor@redhat.com>
38926 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38929 2017-01-26 Martin Sebor <msebor@redhat.com>
38931 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38932 HAVE_DFmode before using XFmode or DFmode.
38933 (parse_directive): Avoid using the z length modifier to avoid
38934 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38936 PR middle-end/78703
38937 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38938 to accept adjustment as an array.
38939 (get_int_range): New function.
38940 (struct directive): Make width and prec arrays.
38941 (directive::set_width, directive::set_precision): Call get_int_range.
38942 (format_integer, format_floating): Handle width and precision ranges.
38943 (format_string, parse_directive): Same.
38945 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38948 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38949 comdat_type_p set, just clone them, but keep the children in the
38953 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38954 which have direct callers with -fvar-tracking-assignments enabled
38956 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38957 inside of type units.
38959 2017-01-26 Martin Sebor <msebor@redhat.com>
38961 PR middle-end/78703
38962 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38964 (struct format_result): Replace number_chars, number_chars_min,
38965 and number_chars_max with a single member of struct result_range.
38967 (format_result::operator+=): Adjust.
38968 (struct fmtresult): Remove bounded. Handle likely and unlikely
38970 (fmtresult::adjust_for_width_or_precision): New function.
38971 (fmtresult:type_max_digits): New function.
38972 (bytes_remaining): Handle likely and unlikely counters.
38973 (min_bytes_remaining): Remove.
38974 (format_percent): Simplify.
38975 (format_integer, format_floating): Set likely and unlikely counters.
38976 (get_string_length, format_character, format_string): Same.
38977 (format_plain, should_warn_p): New function.
38978 (maybe_warn): Call should_warn_p. Update diagnostic messages
38979 and handle those for all directives, including plain strings.
38980 (format_directive): Handle likely and unlikely counters.
38981 Remove unnecessary quoting from diagnostics. Add an informational
38983 (add_bytes): Remove.
38984 (pass_sprintf_length::compute_format_length): Simplify.
38985 (try_substitute_return_value): Handle likely and unlikely counters.
38987 2017-01-26 Carl Love <cel@us.ibm.com>
38989 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38990 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38992 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38995 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38996 endianess for subregs into account.
38997 * lra-constraints.c (lra_constraints): Do risky transformations
38998 always on the first iteration.
38999 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39000 last_call_used_reg_set.
39001 (process_bb_lives): Define and use last_call_used_reg_set.
39002 * lra.c (lra): Always continue after lra_constraints on the first
39005 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39007 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39009 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39011 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39013 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39014 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39015 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39016 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39017 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39018 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39019 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39020 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39021 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39023 2017-01-26 Marek Polacek <polacek@redhat.com>
39026 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39027 for the third operand.
39029 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39031 PR middle-end/79236
39032 * omp-low.c (struct omp_context): Add simt_stmt field.
39033 (scan_omp_for): Return omp_context *.
39034 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39035 context to the _simt_ SIMD stmt.
39036 (lower_omp_for): For combined SIMD with sibling _simt_
39037 SIMD, make sure to use the same decls in _looptemp_
39038 clauses as in the sibling.
39040 2017-01-26 David Sherwood <david.sherwood@arm.com>
39042 PR middle-end/79212
39043 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39046 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39049 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39050 emit fld b; fld a; if possible.
39052 * brig-builtins.def: Update copyright years.
39053 * config/arm/arm_acle_builtins.def: Update copyright years.
39055 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39058 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39059 constraint instead of o for the stxsd instruction.
39061 2017-01-25 Carl Love <cel@us.ibm.com>
39063 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39064 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39066 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39068 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39070 2017-01-25 Richard Biener <rguenther@suse.de>
39072 PR tree-optimization/69264
39073 * target.def (vector_alignment_reachable): Improve documentation.
39074 * doc/tm.texi: Regenerate.
39075 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39077 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39078 earlier changes with respect to TYPE_USER_ALIGN.
39079 (vector_alignment_reachable_p): Likewise. Improve dumping.
39081 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39084 * config/arm/arm.md (xordi3): Force constant operand into a register
39087 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39089 * doc/invoke.texi (-fstore-merging): Correct default optimization
39090 levels at which it is enabled.
39091 (-O): Move -fstore-merging from list to...
39094 2017-01-25 Richard Biener <rguenther@suse.de>
39097 * omp-expand.c: Include debug.h.
39098 (expand_omp_taskreg): Make sure to generate early debug before
39099 outlining anything from a function.
39100 (expand_omp_target): Likewise.
39101 (grid_expand_target_grid_body): Likewise.
39103 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39106 * asan.c (get_translation_unit_decl): New function.
39107 (asan_add_global): Extract modules file name from globals
39108 TRANSLATION_UNIT_DECL name.
39110 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39113 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39114 for long calls with APCS frame and VFP.
39116 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39118 * cfg.c (original_copy_tables_initialized_p): New function.
39119 * cfg.h (original_copy_tables_initialized_p): New decl.
39120 * cfgrtl.c (relink_block_chain): Guard the call to
39121 free_original_copy_tables with a call to
39122 original_copy_tables_initialized_p.
39123 * cgraph.h (symtab_node::native_rtl_p): New decl.
39124 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39125 (symtab_node::needed_p): Don't assert for early assembly output
39126 for __RTL functions.
39127 (cgraph_node::finalize_function): Set "force_output" for __RTL
39129 (cgraph_node::analyze): Bail out early for __RTL functions.
39130 (analyze_functions): Update assertion to support __RTL functions.
39131 (cgraph_node::expand): Bail out early for __RTL functions.
39132 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39134 * function.h (struct function): Update comment for field
39136 * gimple-expr.c: Include "tree-pass.h".
39137 (gimple_has_body_p): Return false for __RTL functions.
39138 * Makefile.in (OBJS): Add run-rtl-passes.o.
39139 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39141 (gcc::pass_manager::get_clean_slate): New accessor.
39142 * passes.c: Include "insn-addr.h".
39143 (should_skip_pass_p): Add logging. Update logic for running
39144 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39145 property-provider override so it is only done for gimple passes.
39147 (skip_pass): New function.
39148 (execute_one_pass): Call skip_pass when skipping passes.
39149 * read-md.c (md_reader::read_char): Support filtering
39150 the input to a subset of line numbers.
39151 (md_reader::md_reader): Initialize fields
39152 m_first_line and m_last_line.
39153 (md_reader::read_file_fragment): New function.
39154 * read-md.h (md_reader::read_file_fragment): New decl.
39155 (md_reader::m_first_line): New field.
39156 (md_reader::m_last_line): New field.
39157 * read-rtl-function.c (function_reader::create_function): Only
39158 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39159 curr_properties. Set DECL_INITIAL to a dummy block.
39160 (read_rtl_function_body_from_file_range): New function.
39161 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39163 * run-rtl-passes.c: New file.
39164 * run-rtl-passes.h: New file.
39166 2017-01-24 Jeff Law <law@redhat.com>
39168 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39171 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39173 PR tree-optimization/79159
39174 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39175 (record_nonwrapping_iv): Improve boundary using above function if no
39176 value range information.
39178 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39179 Martin Jambor <mjambor@suse.cz>
39181 * brig-builtins.def: New file.
39182 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39183 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39184 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39185 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39186 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39187 * builtin-types.def (BT_INT8): New.
39188 (BT_INT16): Likewise.
39189 (BT_UINT8): Likewise.
39190 (BT_UINT16): Likewise.
39191 (BT_FN_ULONG): Likewise.
39192 (BT_FN_UINT_INT): Likewise.
39193 (BT_FN_UINT_ULONG): Likewise.
39194 (BT_FN_UINT_LONG): Likewise.
39195 (BT_FN_UINT_PTR): Likewise.
39196 (BT_FN_ULONG_PTR): Likewise.
39197 (BT_FN_INT8_FLOAT): Likewise.
39198 (BT_FN_INT16_FLOAT): Likewise.
39199 (BT_FN_UINT32_FLOAT): Likewise.
39200 (BT_FN_UINT16_FLOAT): Likewise.
39201 (BT_FN_UINT8_FLOAT): Likewise.
39202 (BT_FN_UINT64_FLOAT): Likewise.
39203 (BT_FN_UINT16_UINT32): Likewise.
39204 (BT_FN_UINT32_UINT16): Likewise.
39205 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39206 (BT_FN_INT_PTR_INT): Likewise.
39207 (BT_FN_UINT_PTR_UINT): Likewise.
39208 (BT_FN_LONG_PTR_LONG): Likewise.
39209 (BT_FN_ULONG_PTR_ULONG): Likewise.
39210 (BT_FN_VOID_UINT64_UINT64): Likewise.
39211 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39212 (BT_FN_INT8_INT8_INT8): Likewise.
39213 (BT_FN_INT16_INT16_INT16): Likewise.
39214 (BT_FN_INT_INT_INT): Likewise.
39215 (BT_FN_UINT_FLOAT_UINT): Likewise.
39216 (BT_FN_FLOAT_UINT_UINT): Likewise.
39217 (BT_FN_ULONG_UINT_UINT): Likewise.
39218 (BT_FN_ULONG_UINT_PTR): Likewise.
39219 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39220 (BT_FN_UINT_UINT_UINT): Likewise.
39221 (BT_FN_VOID_UINT_PTR): Likewise.
39222 (BT_FN_UINT_UINT_PTR: Likewise.
39223 (BT_FN_UINT32_UINT64_PTR): Likewise.
39224 (BT_FN_INT_INT_UINT_UINT): Likewise.
39225 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39226 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39227 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39228 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39229 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39230 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39231 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39232 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39233 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39234 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39235 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39236 * doc/frontends.texi: List BRIG FE.
39237 * doc/install.texi (Testing): Add BRIG tesring requirements.
39238 * doc/invoke.texi (Overall Options): Mention BRIG.
39239 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39241 2017-01-24 Richard Biener <rguenther@suse.de>
39243 PR translation/79208
39244 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39246 2017-01-24 Martin Jambor <mjambor@suse.cz>
39249 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39250 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39251 and known_contexts.
39253 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39255 PR middle-end/79123
39256 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39257 casts from signed to unsigned really don't have a range.
39259 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39261 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39262 GMP_RNDx for compatiblity.
39264 2017-01-24 Martin Liska <mliska@suse.cz>
39267 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39268 that would prevent us to call alloca with -1 as argument.
39270 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39272 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39273 Avoid -Wformat-security warning.
39275 2017-01-23 Andrew Pinski <apinski@cavium.com>
39277 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39280 2017-01-23 Martin Sebor <msebor@redhat.com>
39282 PR middle-end/78703
39283 * gimple-ssa-sprintf.c (warn_level): New global.
39284 (format_integer): Use it here and throughout the rest of the file.
39285 Use the same switch to compute sign as base.
39286 (maybe_warn): New function.
39287 (format_directive): Factor out warnings into maybe_warn.
39288 Add debugging output. Use warn_level.
39289 (add_bytes): Use warn_level.
39290 (pass_sprintf_length::compute_format_length): Add debugging output.
39291 (try_substitute_return_value): Same.
39292 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39294 PR middle-end/78703
39295 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39296 (struct fmtresult, format_integer, format_floating): Adjust.
39297 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39298 (get_string_length, format_string,format_directive): Same.
39299 (pass_sprintf_length::compute_format_length): Same.
39300 (try_substitute_return_value): Simplify slightly.
39302 PR middle-end/78703
39303 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39304 (fmtresult::operator+=): Outlined.
39305 (struct fmtresult): Add ctors.
39306 (struct conversion_spec): Rename...
39307 (struct directive): ...to this. Add and remove data members.
39308 (directive::set_width, directive::set_precision): New functions.
39309 (format_percent): Use fmtresult ctor.
39310 (get_width_and_precision): Remove.
39311 (format_integer): Make naming changes. Avoid computing width and
39313 (format_floating): Same. Adjust indentation.
39314 (format_character, format_none): New functions.
39315 (format_string): Moved character handling to format_character.
39316 (format_directive): Remove arguments, change return type.
39317 (parse_directive): New function.
39318 (pass_sprintf_length::compute_format_length): Move directive
39319 parsing to parse_directive.
39321 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39323 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39324 (assign_assembler_name_if_needed): ... this.
39325 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39326 (assign_assembler_name_if_needed): ... this.
39327 (free_lang_data_in_cgraph): Adjust callers.
39328 * cgraphunit.c (cgraph_node::analyze): Likewise.
39329 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39332 2017-01-23 Richard Biener <rguenther@suse.de>
39334 PR tree-optimization/79088
39335 PR tree-optimization/79188
39336 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39337 resetting loop bounds after last path deletion. Reset loop
39338 bounds of the target loop, make code match the comments.
39339 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39340 Make sure loops need no fixups.
39342 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39344 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39345 exponent support with double type for first argument.
39346 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39347 type returned by __builtin_vec_extract_sig,
39348 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39349 functions from "vector int" to "vector unsigned int" or from
39350 "vector long long int" to "vector unsigned long long int".
39351 Changed type returned by __builtin_vec_extract_exp,
39352 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39353 functions from "vector int" to "vector unsigned int" or from
39354 "vector long long int" to "vector unsigned long long int".
39355 Changed return type of __builtin_vec_test_data_class,
39356 __builtin_vec_test_data_class_sp, and
39357 __builtin_vec_test_data_class_dp from "vector int" to
39358 "vector bool int" or from "vector long long int" to "vector bool
39359 long long int" and changed second argument type from "unsigned
39360 int" to "int". Added new overloaded function forms "vector float
39361 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39362 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39363 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39364 double, vector unsigned long long int)" and "vector double
39365 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39366 long int)". Changed return type of
39367 __builtin_scalar_test_data_class and
39368 __builtin_scalar_test_data_class_sp and
39369 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39370 int" and changed second argument from "unsigned int" to "int".
39371 Changed type returned by __builtin_scalar_test_neg,
39372 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39373 from "int" to "bool int". Added new overloaded function form
39374 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39375 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39376 exponent double-precision with floating point first argument.
39377 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39378 documentation of scalar_test_data_class, scalar_test_neg,
39379 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39380 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39381 vec_test_data_class built-in functions to reflect refinements in
39382 their type signatures.
39384 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39386 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39388 (aarch64_elf_asm_destructor): Likewise.
39390 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39392 PR rtl-optimization/78634
39393 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39394 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39395 * ifcvt.c (noce_try_cmove): Add missing cost check.
39397 PR rtl-optimization/71724
39398 * combine.c (if_then_else_cond): Look for situations where it is
39399 beneficial to undo the work of one of the recursive calls.
39401 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39403 PR tree-optimization/70754
39404 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39405 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39406 combined stmt before it if not NULL.
39407 (combine_chains): Process refs reversely and compute dominance point
39410 2017-01-23 Martin Liska <mliska@suse.cz>
39412 PR tree-optimization/79196
39413 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39414 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39416 (strlen_optimize_stmt): Call the renamed function.
39418 2017-01-23 Michael Matz <matz@suse.de>
39420 PR tree-optimization/78384
39421 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39423 2017-01-23 Richard Biener <rguenther@suse.de>
39425 PR tree-optimization/79186
39426 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39427 both incoming edges before moving an assert.
39429 2017-01-23 Martin Jambor <mjambor@suse.cz>
39431 * ipa-prop.c (load_from_param_1): Removed.
39432 (load_from_unmodified_param): Bits from load_from_param_1 put back
39434 (load_from_param): Removed.
39435 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39436 with stmt. Reverted back to use of load_from_unmodified_param.
39438 2017-01-23 Martin Jambor <mjambor@suse.cz>
39441 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39442 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39443 field a pointer to garbage collected vector, mark lattices and
39444 ipcp_orig_node with GTY((skip)).
39445 (ipa_get_param_count): Adjust to descriptors being a pointer.
39446 (ipa_get_param): Likewise.
39447 (ipa_get_type): Likewise.
39448 (ipa_get_param_move_cost): Likewise.
39449 (ipa_set_param_used): Likewise.
39450 (ipa_get_controlled_uses): Likewise.
39451 (ipa_set_controlled_uses): Likewise.
39452 (ipa_is_param_used): Likewise.
39453 (ipa_node_params_t): Move into garbage collector. New methods insert
39455 (ipa_node_params_sum): Annotate wth GTY(()).
39456 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39458 (ipa_load_from_parm_agg): Adjust declaration.
39459 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39460 * ipa-profile.c (ipa_profile): Likewise.
39461 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39462 (ipa_populate_param_decls): Make descriptors parameter garbage
39464 (ipa_dump_param): Adjust to descriptors being a pointer.
39465 (ipa_alloc_node_params): Likewise.
39466 (ipa_initialize_node_params): Likewise.
39467 (load_from_param_1): Make descriptors parameter garbage collected.
39468 (load_from_unmodified_param): Likewise.
39469 (load_from_param): Likewise.
39470 (ipa_load_from_parm_agg): Likewise.
39471 (ipa_node_params::~ipa_node_params): Removed.
39472 (ipa_free_all_node_params): Remove call to delete operator.
39473 (ipa_node_params_t::insert): New.
39474 (ipa_node_params_t::remove): Likewise.
39475 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39476 copy known_csts and known_contexts vectors.
39477 (ipa_read_node_info): Adjust to descriptors being a pointer.
39478 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39480 (ipcp_transform_function): Make descriptors variable garbage
39483 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39485 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39486 * config/i386/avx512dqintrin.h: Ditto.
39487 * config/i386/avx512fintrin.h: Ditto.
39488 * config/i386/i386.c: Handle new builtins.
39489 * config/i386/i386-builtin.def: Add new builtins.
39490 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39491 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39493 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39494 Martin Liska <mliska@suse.cz>
39496 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39497 * asan.c (asan_expand_poison_ifn): Support stores and use
39498 appropriate ASAN report function.
39499 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39500 * internal-fn.def (ASAN_POISON_USE): Declare.
39501 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39502 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39503 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39504 ASAN_POISON calls w/o LHS.
39505 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39506 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39507 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39508 * gimplify.c (asan_poison_variables): Add attribute
39509 use_after_scope_memory to variables that really needs to live
39511 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39512 having the attribute.
39514 2017-01-23 Martin Liska <mliska@suse.cz>
39516 * asan.c (create_asan_shadow_var): New function.
39517 (asan_expand_poison_ifn): Likewise.
39518 * asan.h (asan_expand_poison_ifn): New declaration.
39519 * internal-fn.c (expand_ASAN_POISON): Likewise.
39520 * internal-fn.def (ASAN_POISON): New builtin.
39521 * sanopt.c (pass_sanopt::execute): Expand
39522 asan_expand_poison_ifn.
39523 * tree-inline.c (copy_decl_for_dup_finish): Make function
39525 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39526 * tree-ssa.c (is_asan_mark_p): New function.
39527 (execute_update_addresses_taken): Rewrite local variables
39528 (identified just by use-after-scope as addressable) into SSA.
39530 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39532 * doc/install.texi (Specific): opensource.apple.com uses https
39533 now. Remove trailing slash.
39535 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39537 * README.Portability: Remove note on an Irix compatibility issue.
39539 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39541 * gcov.c (INCLUDE_ALGORITHM): Define.
39542 (INCLUDE_VECTOR): Define.
39543 No longer include <vector> and <algorithm> directly.
39545 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39547 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39549 * doc/invoke.texi (Code Gen Options): Ditto.
39551 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39554 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39556 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39558 rtl-optimization/79125
39559 * cprop.c (local_cprop_pass): Handle cases where we make an
39560 unconditional trap.
39562 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39566 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39567 read from, for big endian.
39569 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39571 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39572 register pauth builtins for LP64 only.
39574 2017-01-20 Marek Polacek <polacek@redhat.com>
39577 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39580 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39582 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39584 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39585 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39586 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39588 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39591 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39592 in big-endian mode when they are not a single duplicated value.
39594 2017-01-20 Richard Biener <rguenther@suse.de>
39596 * BASE-VER: Bump to 7.0.1.
39598 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39600 * omp-low.c (omplow_simd_context): New struct. Use it...
39601 (lower_rec_simd_input_clauses): ...here and...
39602 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39603 references to idx, lane, max_vf, is_simt.
39605 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39607 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39610 2017-01-20 Martin Jambor <mjambor@suse.cz>
39612 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39613 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39615 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39616 (GTFILES): Rename hsa.c to hsa-common.c.
39617 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39618 * hsa-dump.c: Likewise.
39619 * hsa-gen.c: Likewise.
39620 * hsa-regalloc.c: Likewise.
39621 * ipa-hsa.c: Likewise.
39622 * omp-expand.c: Likewise.
39623 * omp-low.c: Likewise.
39624 * toplev.c: Likewise.
39626 2017-01-20 Marek Polacek <polacek@redhat.com>
39629 * doc/invoke.texi: Document -Wduplicated-branches.
39630 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39631 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39632 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39633 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39634 return 0 only when not OEP_LEXICOGRAPHIC.
39635 (fold_build_cleanup_point_expr): Use the expression
39636 location when building CLEANUP_POINT_EXPR.
39637 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39638 * tree.c (add_expr): Handle error_mark_node.
39640 2017-01-20 Martin Liska <mliska@suse.cz>
39643 * tree-profile.c (init_ic_make_global_vars): Do not call
39645 (gimple_init_gcov_profiler): Likewise.
39647 2017-01-20 Martin Liska <mliska@suse.cz>
39650 * cgraph.h (maybe_create_reference): Remove argument and
39652 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39654 * ipa-cp.c (create_specialized_node): Likewise.
39655 * symtab.c (symtab_node::maybe_create_reference): Handle
39656 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39658 2017-01-20 Martin Liska <mliska@suse.cz>
39660 * read-rtl-function.c (function_reader::create_function): Use
39661 build_decl instread of build_decl_stat.
39663 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39665 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39666 * config/i386/avx512dqintrin.h: Ditto.
39667 * config/i386/avx512fintrin.h: Ditto.
39668 * config/i386/i386-builtin-types.def: Add new types.
39669 * config/i386/i386.c: Handle new types.
39670 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39671 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39672 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39673 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39674 (__builtin_ia32_kshiftridi): New.
39675 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39677 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39681 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39682 define to rs6000_init_stack_protect_guard.
39683 (rs6000_init_stack_protect_guard): New function.
39685 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39686 Yunqiang Su <yunqiang.su@imgtec.com>
39688 * config.gcc (supported_defaults): Add madd4.
39689 (with_madd4): Add validation.
39690 (all_defaults): Add madd4.
39691 * config/mips/mips.opt (mmadd4): New option.
39692 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39694 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39696 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39697 (ISA_HAS_FUSED_MADD4): Likewise.
39698 * doc/invoke.texi (-mmadd4): Document the new option.
39699 * doc/install.texi (--with-madd4): Document the new option.
39701 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39703 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39704 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39705 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39706 (aarch64_init_pauth_hint_builtins): New.
39707 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39708 (aarch64_expand_builtin): Expand new builtins.
39710 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39712 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39713 * combine-stack-adj.c (no_unhandled_cfa): Handle
39714 REG_CFA_TOGGLE_RA_MANGLE.
39715 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39716 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39717 info for return address signing.
39718 (aarch64_expand_epilogue): Likewise.
39720 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39722 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39723 * config/aarch64/aarch64-protos.h
39724 (aarch64_return_address_signing_enabled): New declaration.
39725 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39727 (aarch64_expand_prologue): Sign return address before it's pushed onto
39729 (aarch64_expand_epilogue): Authenticate return address fetched from
39731 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39732 (aarch64_attributes): New function attributes for "sign-return-address".
39733 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39734 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39735 ("*do_return"): Generate combined instructions according to key index.
39736 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39737 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39739 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39740 * config/aarch64/aarch64.opt (msign-return-address=): New.
39741 * doc/extend.texi (AArch64 Function Attributes): Documents
39742 "sign-return-address=".
39743 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39745 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39747 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39748 overall option summary.
39750 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39752 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39753 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39754 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39755 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39757 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39759 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39760 -mpower9-minmax by default for -mcpu=power9.
39761 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39762 128-bit floating point.
39764 2017-01-20 Alan Modra <amodra@gmail.com>
39766 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39767 optimizing for size.
39769 2017-01-20 Alan Modra <amodra@gmail.com>
39772 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39773 for strcmp and strncmp from corresponding builtin decl.
39775 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39777 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39778 instead of i386/rtems-64.h.
39779 * config/i386/rtems-64.h: Remove.
39781 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39785 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39787 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39789 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39791 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39792 Change int to HOST_WIDE_INT.
39793 * config/aarch64/aarch64-protos.h
39794 (aarch64_simd_gen_const_vector_dup): Likewise.
39795 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39797 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39799 * langhooks-def.h (lhd_type_for_size): New decl.
39800 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39801 * langhooks.c (lhd_type_for_size): New function, taken from
39804 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39806 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39807 define_bypass for CR latency.
39808 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39809 (power9-alu2): Add define_bypass for CR latency.
39811 (power9-mul): Update insn latency.
39812 (power9-mul-compare): Update insn latency, bypass latency and remove
39815 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39817 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39819 * config/aarch64/aarch64.md
39820 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39821 aarch64_nopcrelative_literal_loads.
39822 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39824 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39826 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39827 TARGET_LOONGSON_3A.
39828 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39830 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39833 * config.gcc (supported_defaults): Add lxc1-sxc1.
39834 (with_lxc1_sxc1): Add validation.
39835 (all_defaults): Add lxc1-sxc1.
39836 * config/mips/mips.opt (mlxc1-sxc1): New option.
39837 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39839 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39840 __mips_no_lxc1_sxc1.
39841 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39842 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39843 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39845 2017-01-19 Richard Biener <rguenther@suse.de>
39847 PR tree-optimization/72488
39848 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39849 sure to restore SSA info.
39850 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39852 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39854 PR rtl-optimization/79121
39855 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39856 of the inner type when shifting an extended value.
39858 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39861 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39862 interposable aliases.
39864 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39867 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39868 Use the evmergelohi instruction.
39869 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39870 (mov_sitf_e500_subreg8_2_be): Likewise.
39871 (mov_sitf_e500_subreg12_2_le): Likewise.
39872 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39873 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39874 (mov_sitf_e500_subreg8_2_le): Likewise.
39875 (mov_sitf_e500_subreg12_2_be): Likewise.
39877 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39879 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39880 attribute from vecsimple to vecperm.
39881 (altivec_vbpermq2): Likewise.
39883 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39886 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39888 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39889 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39890 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39891 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39892 case where N arg is SIZE_MAX.
39893 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39894 (cmpstrsi): Add pattern.
39896 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39898 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39899 __builtin_vec_revb builtins.
39900 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39901 built-in functions to support generation of the ISA 3.0 XXBR<x>
39902 vector byte reverse instructions.
39903 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39904 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39905 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39906 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39907 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39908 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39909 (P9V_BUILTIN_VEC_REVB): Likewise.
39910 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39911 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39912 (p9_xxbrq_v16qi): Likewise.
39913 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39914 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39915 (p9_xxbrh_v8hi): Likewise.
39916 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39917 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39918 vec_revb built-in functions.
39920 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39922 PR rtl-optimization/78952
39923 * config/i386/i386.md (any_extract): New code iterator.
39924 (*insvqi_2): Use any_extract for source operand.
39925 (*insvqi_3): Use any_shiftrt for source operand.
39927 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39929 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39931 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39933 2017-01-18 Matthias Klose <doko@ubuntu.com>
39935 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39937 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39939 * config/rs6000/altivec.h (vec_bperm): Change #define.
39940 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39941 (altivec_vbpermq2): New define_insn.
39942 (altivec_vbpermd): Likewise.
39943 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39944 function interface.
39945 (VBPERMD): Likewise.
39946 (VBPERM): New polymorphic function interface.
39947 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39948 Add entries for P9V_BUILTIN_VEC_VBPERM.
39949 * doc/extend.texi: Add interfaces for vec_bperm.
39951 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39953 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39954 first letter of error messages.
39955 (s390_resolve_overloaded_builtin): Likewise.
39956 * config/s390/s390.c (s390_expand_builtin): Likewise.
39957 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39958 (s390_valid_target_attribute_inner_p): Likewise.
39959 * config/s390/s390.md ("tabort"): Likewise.
39961 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39963 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39964 (ISA_AVOID_DIV_HILO): New macro.
39965 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39966 (ISA_HAS_DDIV): Likewise.
39968 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39970 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39972 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39974 * doc/invoke.texi (fabi-version): Spelling fix.
39976 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39979 * doc/invoke.texi (fabi-version): Mention mangling fix for
39982 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39985 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39987 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39990 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39991 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39993 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39995 (stack_protect_test): Ditto.
39996 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39997 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39999 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40000 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40001 -mstack-protector-guard-offset=.
40002 (RS/6000 and PowerPC Options): Ditto.
40004 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40006 * config/i386/i386.h (MASK_CLASS_P): New define.
40007 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40008 there are no registers from different register sets also when
40009 mask registers are used. Update function comment.
40010 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40011 to (*k/*r) and (*k/*km) alternatives.
40013 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40015 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40016 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40017 (EH_RETURN_HANDLER_RTX): New define.
40018 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40019 Force frame pointer in EH return functions.
40020 (aarch64_expand_epilogue): Add barrier for eh_return.
40021 (aarch64_final_eh_return_addr): Remove.
40022 (aarch64_eh_return_handler_rtx): New function.
40023 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40025 (aarch64_eh_return_handler_rtx): New prototype.
40027 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40029 * config/rs6000/altivec.h (vec_rlmi): New #define.
40030 (vec_vrlnm): Likewise.
40031 (vec_rlnm): Likewise.
40032 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40033 (UNSPEC_VRLNM): Likewise.
40034 (VIlong): New mode iterator.
40035 (altivec_vrl<VI_char>mi): New define_insn.
40036 (altivec_vrl<VI_char>nm): Likewise.
40037 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40039 (VRLDNM): Likewise.
40040 (RLNM): New polymorphic function entry.
40041 (VRLWMI): New monomorphic function entry.
40042 (VRLDMI): Likewise.
40043 (RLMI): New polymorphic function entry.
40044 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40045 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40046 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40049 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40052 * dwarf2out.c (field_byte_offset): Restore the
40053 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40054 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40055 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40058 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40061 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40063 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40066 * dwarf2out.c (add_data_member_location_attribute): For constant
40067 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40068 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40069 DW_AT_byte_size attributes.
40071 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40073 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40074 after forcing to constant memory when the code model is medium.
40076 2017-01-17 Julia Koval <julia.koval@intel.com>
40079 * config/i386/avx512fintrin.h
40080 (_mm512_i32gather_ps): Change __addr type to void const*.
40081 (_mm512_mask_i32gather_ps): Ditto.
40082 (_mm512_i32gather_pd): Ditto.
40083 (_mm512_mask_i32gather_pd): Ditto.
40084 (_mm512_i64gather_ps): Ditto.
40085 (_mm512_mask_i64gather_ps): Ditto.
40086 (_mm512_i64gather_pd): Ditto.
40087 (_mm512_mask_i64gather_pd): Ditto.
40088 (_mm512_i32gather_epi32): Ditto.
40089 (_mm512_mask_i32gather_epi32): Ditto.
40090 (_mm512_i32gather_epi64): Ditto.
40091 (_mm512_mask_i32gather_epi64): Ditto.
40092 (_mm512_i64gather_epi32): Ditto.
40093 (_mm512_mask_i64gather_epi32): Ditto.
40094 (_mm512_i64gather_epi64): Ditto.
40095 (_mm512_mask_i64gather_epi64): Ditto.
40096 (_mm512_i32scatter_ps): Change __addr type to void*.
40097 (_mm512_mask_i32scatter_ps): Ditto.
40098 (_mm512_i32scatter_pd): Ditto.
40099 (_mm512_mask_i32scatter_pd): Ditto.
40100 (_mm512_i64scatter_ps): Ditto.
40101 (_mm512_mask_i64scatter_ps): Ditto.
40102 (_mm512_i64scatter_pd): Ditto.
40103 (_mm512_mask_i64scatter_pd): Ditto.
40104 (_mm512_i32scatter_epi32): Ditto.
40105 (_mm512_mask_i32scatter_epi32): Ditto.
40106 (_mm512_i32scatter_epi64): Ditto.
40107 (_mm512_mask_i32scatter_epi64): Ditto.
40108 (_mm512_i64scatter_epi32): Ditto.
40109 (_mm512_mask_i64scatter_epi32): Ditto.
40110 (_mm512_i64scatter_epi64): Ditto.
40111 (_mm512_mask_i64scatter_epi64): Ditto.
40112 * config/i386/avx512pfintrin.h
40113 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40114 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40115 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40116 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40117 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40118 (_mm512_prefetch_i32scatter_ps): Ditto.
40119 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40120 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40121 (_mm512_prefetch_i64scatter_pd): Ditto.
40122 (_mm512_prefetch_i64scatter_ps): Ditto.
40123 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40124 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40125 * config/i386/avx512vlintrin.h
40126 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40127 (_mm_mmask_i32gather_ps): Ditto.
40128 (_mm256_mmask_i32gather_pd): Ditto.
40129 (_mm_mmask_i32gather_pd): Ditto.
40130 (_mm256_mmask_i64gather_ps): Ditto.
40131 (_mm_mmask_i64gather_ps): Ditto.
40132 (_mm256_mmask_i64gather_pd): Ditto.
40133 (_mm_mmask_i64gather_pd): Ditto.
40134 (_mm256_mmask_i32gather_epi32): Ditto.
40135 (_mm_mmask_i32gather_epi32): Ditto.
40136 (_mm256_mmask_i32gather_epi64): Ditto.
40137 (_mm_mmask_i32gather_epi64): Ditto.
40138 (_mm256_mmask_i64gather_epi32): Ditto.
40139 (_mm_mmask_i64gather_epi32): Ditto.
40140 (_mm256_mmask_i64gather_epi64): Ditto.
40141 (_mm_mmask_i64gather_epi64): Ditto.
40142 (_mm256_i32scatter_ps): Change __addr type to void*.
40143 (_mm256_mask_i32scatter_ps): Ditto.
40144 (_mm_i32scatter_ps): Ditto.
40145 (_mm_mask_i32scatter_ps): Ditto.
40146 (_mm256_i32scatter_pd): Ditto.
40147 (_mm256_mask_i32scatter_pd): Ditto.
40148 (_mm_i32scatter_pd): Ditto.
40149 (_mm_mask_i32scatter_pd): Ditto.
40150 (_mm256_i64scatter_ps): Ditto.
40151 (_mm256_mask_i64scatter_ps): Ditto.
40152 (_mm_i64scatter_ps): Ditto.
40153 (_mm_mask_i64scatter_ps): Ditto.
40154 (_mm256_i64scatter_pd): Ditto.
40155 (_mm256_mask_i64scatter_pd): Ditto.
40156 (_mm_i64scatter_pd): Ditto.
40157 (_mm_mask_i64scatter_pd): Ditto.
40158 (_mm256_i32scatter_epi32): Ditto.
40159 (_mm256_mask_i32scatter_epi32): Ditto.
40160 (_mm_i32scatter_epi32): Ditto.
40161 (_mm_mask_i32scatter_epi32): Ditto.
40162 (_mm256_i32scatter_epi64): Ditto.
40163 (_mm256_mask_i32scatter_epi64): Ditto.
40164 (_mm_i32scatter_epi64): Ditto.
40165 (_mm_mask_i32scatter_epi64): Ditto.
40166 (_mm256_i64scatter_epi32): Ditto.
40167 (_mm256_mask_i64scatter_epi32): Ditto.
40168 (_mm_i64scatter_epi32): Ditto.
40169 (_mm_mask_i64scatter_epi32): Ditto.
40170 (_mm256_i64scatter_epi64): Ditto.
40171 (_mm256_mask_i64scatter_epi64): Ditto.
40172 (_mm_i64scatter_epi64): Ditto.
40173 (_mm_mask_i64scatter_epi64): Ditto.
40174 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40175 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40176 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40177 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40178 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40179 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40180 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40181 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40182 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40183 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40184 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40185 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40186 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40187 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40188 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40189 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40190 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40191 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40192 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40193 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40194 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40195 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40196 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40197 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40198 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40199 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40200 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40201 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40202 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40203 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40204 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40205 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40206 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40207 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40208 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40209 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40210 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40211 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40212 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40213 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40214 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40215 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40216 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40217 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40218 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40219 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40220 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40221 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40222 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40223 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40224 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40225 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40226 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40227 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40228 definitions accordingly.
40230 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40231 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40234 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40237 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40240 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40241 hard regs for inner regno.
40243 2017-01-17 Martin Liska <mliska@suse.cz>
40246 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40247 assumption and add comment.
40249 2017-01-17 Nathan Sidwell <nathan@acm.org>
40251 * ipa-visibility.c (localize_node): New function, broken out of ...
40252 (function_and_variable_visibility): ... here. Call it.
40254 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40256 PR middle-end/77445
40257 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40258 correctly set frequency of oudgoing edge.
40259 (duplicate_thread_path): Fix profile updating.
40261 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40264 * configure.ac: Add GCC_BASE_VER.
40265 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40266 version from BASE-VER file.
40267 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40268 (gcc.o): Depend on $(BASEVER).
40269 * common.opt (dumpfullversion): New option.
40270 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40271 * doc/invoke.texi: Document -dumpfullversion.
40272 * doc/install.texi: Document --with-gcc-major-version-only.
40273 * configure: Regenerated.
40275 2017-01-17 Richard Biener <rguenther@suse.de>
40277 PR tree-optimization/71433
40278 * tree-vrp.c (register_new_assert_for): Merge same asserts
40279 on all incoming edges.
40280 (process_assert_insertions_for): Handle insertions at the
40283 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40285 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40286 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40288 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40291 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40294 2017-01-17 Alan Modra <amodra@gmail.com>
40297 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40298 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40299 symbolic stack limit when pic.
40301 2017-01-16 Martin Sebor <msebor@redhat.com>
40303 PR tree-optimization/78608
40304 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40306 2017-01-16 Jeff Law <law@redhat.com>
40309 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40310 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40311 for several include directories that may be relative to sysroot.
40312 * config/i386/x-mingw32 (gplus_includedir): Define.
40313 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40314 (native_system_includedir): Likewise.
40315 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40316 override if TARGET_SYSTEM_ROOT is defined.
40317 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40319 PR tree-optimization/79090
40320 PR tree-optimization/33562
40321 PR tree-optimization/61912
40322 PR tree-optimization/77485
40323 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40324 and computed trims into the dump file.
40326 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40328 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40330 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40333 * gimplify.c (gimplify_init_constructor): If want_value and
40334 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40338 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40339 sequence. Formatting fixes.
40340 (doloop_optimize): Formatting fixes.
40343 * gcc.c (debug_level_greater_than_spec_func): New function.
40344 (static_spec_functions): Add debug-level-gt spec function.
40345 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40347 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40348 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40349 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40350 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40351 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40352 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40354 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40356 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40357 QImode fixups to general and mask registers only.
40359 2017-01-16 Carl Love <cel@us.ibm.com>
40361 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40362 for built-in functions
40363 vector signed char vec_nabs (vector signed char)
40364 vector signed short vec_nabs (vector signed short)
40365 vector signed int vec_nabs (vector signed int)
40366 vector signed long long vec_nabs (vector signed long long)
40367 vector float vec_nabs (vector float)
40368 vector double vec_nabs (vector double)
40369 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40371 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40372 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40373 * doc/extend.texi: Update the documentation file for the new built-in
40376 2017-01-16 Martin Sebor <msebor@redhat.com>
40378 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40381 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40383 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40384 UNSPEC_VSX__XXSPLTD to require special splat handling.
40386 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40389 * system.h: Poison strndup.
40391 2017-01-16 Alan Modra <amodra@gmail.com>
40394 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40397 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40399 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40401 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40403 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40404 call recog here. Assert that INSN_CODE (insn) is non-negative.
40406 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40409 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40411 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40412 in the currently scheduled RTL fragment.
40414 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40416 PR rtl-optimization/78751
40417 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40420 2017-01-14 Jeff Law <law@redhat.com>
40422 PR tree-optimization/79090
40423 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40424 variable length stores.
40425 (compute_trims): Delete dead assignment to *trim_tail.
40426 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40429 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40431 PR rtl-optimization/78626
40432 PR rtl-optimization/78727
40433 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40434 of a block, and split such blocks after everything else is finished.
40436 2017-01-14 Alan Modra <amodra@gmail.com>
40439 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40440 target legitimate_combined_insn.
40441 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40442 (rs6000_legitimate_combined_insn): New function.
40443 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40445 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40446 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40447 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40449 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40451 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40453 2017-01-13 Jeff Law <law@redhat.com>
40455 PR tree-optimization/33562
40456 PR tree-optimization/61912
40457 PR tree-optimization/77485
40458 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40460 (delete_dead_assignment): Likewise.
40461 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40462 statement to delete_dead_call and delete_dead_assignment.
40464 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40467 * substring-locations.c (format_warning_va): Strengthen case 1 so
40468 that both endpoints of the substring must be within the format
40469 range for just the substring to be printed.
40471 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40473 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40474 * config/i386/i386.c (ix86_target_string): Add missing options
40475 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40476 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40477 flags_other and ix86_target_other to flags2_other. Display unknown
40479 (ix86_valid_target_attribute_inner_p): Add missing options and
40480 reorder options by implied ISAs, as in ix86_target_string.
40482 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40484 * hash-table.h (hash_table::too_empty_p): New function.
40485 (hash_table::expand): Use it.
40486 (hash_table::traverse): Likewise.
40487 (hash_table::empty_slot): Use sizeof (value_type) instead of
40488 sizeof (PTR) to convert bytes to elements. Shrink the table
40489 if the current size is excessive for the current number of
40492 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40494 * ira-costs.c (record_reg_classes): Break from the inner loop
40495 early once alt_fail is known to be true. Update outer loop
40496 handling accordingly.
40498 2017-01-13 Jeff Law <law@redhat.com>
40500 * tree-ssa-dse.c (decrement_count): New function.
40501 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40502 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40503 when we know the partially dead statement is a mem* function.
40505 PR tree-optimization/61912
40506 PR tree-optimization/77485
40507 * tree-ssa-dse.c: Include expr.h.
40508 (maybe_trim_constructor_store): New function.
40509 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40511 PR tree-optimization/33562
40512 PR tree-optimization/61912
40513 PR tree-optimization/77485
40514 * doc/invoke.texi: Document new dse-max-object-size param.
40515 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40516 * tree-ssa-dse.c: Include params.h.
40517 (dse_store_status): New enum.
40518 (initialize_ao_ref_for_dse): New, partially extracted from
40520 (valid_ao_ref_for_dse, normalize_ref): New.
40521 (setup_live_bytes_from_ref, compute_trims): Likewise.
40522 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40523 (maybe_trim_partially_dead_store): Likewise.
40524 (maybe_trim_complex_store): Likewise.
40525 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40526 Track what bytes live from the original store. Return tri-state
40527 for dead, partially dead or live.
40528 (dse_dom_walker): Add constructor, destructor and new private members.
40529 (delete_dead_call, delete_dead_assignment): New extracted from
40531 (dse_optimize_stmt): Make a member of dse_dom_walker.
40532 Use initialize_ao_ref_for_dse.
40534 PR tree-optimization/33562
40535 PR tree-optimization/61912
40536 PR tree-optimization/77485
40537 * sbitmap.h (bitmap_count_bits): Prototype.
40538 (bitmap_clear_range, bitmap_set_range): Likewise.
40539 * sbitmap.c (bitmap_clear_range): New function.
40540 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40542 2017-01-13 Martin Liska <mliska@suse.cz>
40545 * function.c (set_cfun): Add new argument force.
40546 * function.h (set_cfun): Likewise.
40547 * ipa-inline-transform.c (inline_call): Use the function when
40548 strict alising from is dropped for function we inline to.
40550 2017-01-13 Richard Biener <rguenther@suse.de>
40552 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40553 for dumping GIMPLE INTEGER_CSTs.
40555 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40557 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40558 to 201112L since C++17.
40560 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40563 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40564 if -fsanitize=kernel-address is present.
40566 2017-01-13 Richard Biener <rguenther@suse.de>
40568 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40569 as _Literal ( type ) number in case usual suffixes do not
40570 preserve all information.
40572 2017-01-13 Richard Biener <rguenther@suse.de>
40574 PR tree-optimization/77283
40575 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40576 and ssa-iterators.h.
40577 (is_feasible_trace): Implement a cost model based on joiner
40580 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40583 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40584 char or short to __float128/_Float128 directly.
40586 2017-01-12 Martin Sebor <msebor@redhat.com>
40588 to -Wformat-overflow.
40589 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40590 (min_bytes_remaining): Same.
40591 (get_string_length): Same.
40592 (format_string): Same.
40593 (format_directive): Same.
40595 (pass_sprintf_length::handle_gimple_call): Same.
40597 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40599 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40600 info.nowrite calls with no lhs that can't throw. Return bool
40601 whether gsi_remove has been called or not.
40602 (pass_sprintf_length::handle_gimple_call): Return bool whether
40603 try_substitute_return_value called gsi_remove. Formatting fix.
40604 (pass_sprintf_length::execute): Don't use gsi_remove if
40605 handle_gimple_call returned true.
40608 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40609 be removed due to side-effects, don't remove following barrier nor
40610 turn the successor edge into fallthru edge.
40612 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40615 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40616 element-reversing loads and stores as not swappable.
40618 2017-01-12 Nathan Sidwell <nathan@acm.org>
40619 Nicolai Stange <nicstange@gmail.com>
40621 * combine.c (try_combine): Don't ignore result of overlap checking
40622 loop. Combine overlap & asm check into single loop.
40624 2017-01-12 Richard Biener <rguenther@suse.de>
40626 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40627 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40629 2017-01-12 Richard Biener <rguenther@suse.de>
40631 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40632 and TS_TARGET_OPTION directly derive from TS_BASE.
40633 * tree-core.h (tree_optimization_option): Derive from tree_base.
40634 (tree_target_option): Likewise.
40636 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40638 * config/i386/i386.c (memory_address_length): Increase len
40639 only when rip_relative_addr_p returns false.
40641 2017-01-11 Julia Koval <julia.koval@intel.com>
40643 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40644 (OPTION_MASK_ISA_SGX_SET): New.
40645 (ix86_handle_option): Handle OPT_msgx.
40646 * config.gcc: Added sgxintrin.h.
40647 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40648 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40649 * config/i386/i386.c (ix86_target_string): Add -msgx.
40651 (ix86_option_override_internal): Handle new options.
40652 (ix86_valid_target_attribute_inner_p): Add sgx.
40653 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40654 * config/i386/i386.opt: Add msgx.
40655 * config/i386/sgxintrin.h: New file.
40656 * config/i386/x86intrin.h: Add sgxintrin.h.
40658 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40661 * fold-const.c (maybe_nonzero_address): Return 1 for function
40663 (tree_single_nonzero_warnv_p): Don't handle function local objects
40667 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40670 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40673 * opts.c: Include "spellcheck.h"
40674 (struct string_fragment): New struct.
40675 (struct edit_distance_traits<const string_fragment &>): New
40677 (get_closest_sanitizer_option): New function.
40678 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40680 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40682 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40684 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40685 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40686 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40687 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40688 for initial die_offset if dwarf_split_debug_info.
40689 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40690 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40692 (output_skeleton_debug_sections): Formatting fix. Use
40693 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40694 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40696 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40698 * config/arm/cortex-a53.md: Add bypasses for
40699 cortex_a53_r2f_cvt.
40700 (cortex_a53_r2f): Only use for transfers.
40701 (cortex_a53_f2r): Likewise.
40702 (cortex_a53_r2f_cvt): Add reservation for conversions.
40703 (cortex_a53_f2r_cvt): Likewise.
40705 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40707 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40708 to all inlined functions, change static to extern.
40710 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40713 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40715 (arm_assemble_integer): Likewise.
40717 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40719 * config.gcc: Use new awk script to check CPU, FPU and architecture
40720 parameters for --with-... options.
40721 * config/arm/parsecpu.awk: New file
40722 * config/arm/arm-cpus.in: New file.
40723 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40725 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40727 * config/arm/t-arm: Update dependency rules.
40728 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40729 of processing .def files.
40730 * config/arm/genopt.sh: Deleted.
40731 * config/arm/gentune.sh: Deleted.
40732 * config/arm/arm-cores.def: Deleted.
40733 * config/arm/arm-arches.def: Deleted.
40734 * config/arm/arm-fpus.def: Deleted.
40735 * config/arm/arm-tune.md: Regenerated.
40736 * config/arm/arm-tables.opt: Regenerated.
40737 * config/arm/arm-cpu.h: New generated file.
40738 * config/arm/arm-cpu-data.h: New generated file.
40739 * config/arm/arm-cpu-cdata.h: New generated file.
40741 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40744 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40746 (input_varpool_node): Unpack dynamically_initialized bit.
40748 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40750 PR rtl-optimization/79032
40751 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40752 the alignment of the adjusted memory reference against that of MODE,
40753 instead of the alignment of the original memory reference.
40755 2017-01-11 Martin Jambor <mjambor@suse.cz>
40757 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40759 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40760 decorated functions.
40762 2017-01-11 Richard Biener <rguenther@suse.de>
40764 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40765 set range/nonnull info for PHI results. Do not set it on
40766 stmts marked for removal.
40768 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40770 * expr.c (store_field): In the bitfield case, fetch the return value
40771 from the registers before applying a single big-endian adjustment.
40772 Always do a final load for a BLKmode value not larger than a word.
40774 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40777 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40778 that we correctly handle column numbers greater than
40779 LINE_MAP_MAX_COLUMN_NUMBER.
40781 2017-01-10 Martin Sebor <msebor@redhat.com>
40783 PR middle-end/78245
40784 * gimple-ssa-sprintf.c (get_destination_size): Call
40785 {init,fini}object_sizes.
40786 * tree-object-size.c (addr_object_size): Adjust.
40787 (pass_through_call): Adjust.
40788 (pass_object_sizes::execute): Adjust.
40789 * tree-object-size.h (fini_object_sizes): Declare.
40791 2017-01-10 Martin Sebor <msebor@redhat.com>
40793 PR tree-optimization/78775
40794 * builtins.c (get_size_range): Move...
40795 * calls.c: ...to here.
40796 (alloc_max_size): Accept zero argument.
40797 (operand_signed_p): Remove.
40798 (maybe_warn_alloc_args_overflow): Call get_size_range.
40799 * calls.h (get_size_range): Declare.
40801 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40803 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40804 from TI's devices.csv file as of September 2016.
40805 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40807 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40809 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40810 * doc/invoke.texi: Likewise.
40811 * doc/md.texi: Likewise.
40812 * doc/objc.texi: Likewise.
40814 2017-01-10 Joshua Conner <joshconner@google.com>
40816 * config/arm/fuchsia-elf.h: New file.
40817 * config/fuchsia.h: New file.
40818 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40819 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40821 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40823 2016-01-10 Richard Biener <rguenther@suse.de>
40825 PR tree-optimization/79034
40826 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40827 Propagate out degenerate PHIs in the joiner.
40829 2017-01-10 Martin Liska <mliska@suse.cz>
40831 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40832 (sort_congruence_classes_by_decl_uid): Likewise.
40833 (sort_congruence_class_groups_by_decl_uid): Likewise.
40834 (sem_item_optimizer::merge_classes): Sort class, groups in these
40835 classes and members in the groups by DECL_UID of declarations.
40836 This would make merge operations stable.
40838 2017-01-10 Martin Liska <mliska@suse.cz>
40840 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40841 usage of m_classes_vec.
40842 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40843 (sem_item_optimizer::get_group_by_hash): Likewise.
40844 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40845 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40846 (sem_item_optimizer::verify_classes): Likewise.
40847 (sem_item_optimizer::process_cong_reduction): Likewise.
40848 (sem_item_optimizer::dump_cong_classes): Likewise.
40849 (sem_item_optimizer::merge_classes): Likewise.
40850 * ipa-icf.h (congruence_class_hash): Rename from
40851 congruence_class_group_hash. Remove declaration of m_classes_vec.
40853 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40855 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40856 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40857 * config.gcc: Add avx512vpopcntdqintrin.h.
40858 * config/i386/avx512vpopcntdqintrin.h: New.
40859 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40860 * config/i386/i386-builtin-types.def: Add new types.
40861 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40862 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40863 __builtin_ia32_vpopcountq_v8di_mask): New.
40864 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40865 __AVX512VPOPCNTDQ__.
40866 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40867 (PTA_AVX512VPOPCNTDQ): Define.
40868 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40869 TARGET_AVX512VPOPCNTDQ_P): Define.
40870 * config/i386/i386.opt: Add mavx512vpopcntdq.
40871 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40872 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40874 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40876 PR middle-end/77484
40877 * predict.def (PRED_CALL): Set to 67.
40879 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40881 * expr.c (store_field): In the bitfield case, if the value comes from
40882 a function call and is of an aggregate type returned in registers, do
40883 not modify the field mode; extract the value in all cases if the mode
40884 is BLKmode and the size is not larger than a word.
40886 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40889 * config/i386/cpuid.h: Fix undefined behavior.
40891 2017-01-04 Jeff Law <law@redhat.com>
40893 PR tree-optimization/79007
40894 PR tree-optimization/67955
40895 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40896 conservative for pt.null when flag_non_call_exceptions is on.
40898 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40900 PR translation/79019
40901 PR translation/79020
40902 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40903 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40904 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40906 * config/avr/avr.opt (maccumulate-args): Likewise.
40907 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40908 * common.opt (freport-bug): Likewise.
40909 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40910 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40911 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40912 translatable string.
40913 * config/i386/i386.c (function_value_32): Likewise.
40914 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40915 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40917 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40918 * common/config/msp430/msp430-common.c (msp430_handle_option):
40920 * symtab.c (symtab_node::verify_base): Likewise.
40921 * opts.c (set_debug_level): Likewise.
40922 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40923 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40924 missing whitespace to translatable strings.
40925 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40926 * config/sh/superh.h: Likewise.
40927 * config/i386/xopintrin.h: Likewise.
40928 * config/i386/znver1.md: Likewise.
40929 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40930 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40931 * double-int.h (struct double_int): Likewise.
40932 * double-int.c (div_and_round_double): Likewise.
40933 * wide-int.cc: Likewise.
40934 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40935 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40936 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40937 (crossjumps_occurred): ... this.
40938 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40941 PR tree-optimization/78899
40942 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40943 returning bool return struct loop *, NULL for failure and the new
40945 (versionable_outer_loop_p): Don't version outer loop if it has
40946 dont_vectorized bit set.
40947 (tree_if_conversion): When versioning outer loop, ensure
40948 tree_if_conversion is performed also on the inner loop of the
40949 non-vectorizable outer loop copy.
40950 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40951 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40952 prevent vectorization of it.
40953 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40954 the outer loop vectorization of the non-scalar version is attempted
40955 before vectorization of the inner loop in scalar version. If
40956 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40957 vectorization of its inner loop.
40958 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40959 has 2 inner loops, rename also on edges from bb whose single pred
40960 is outer_loop->header. Fix typo in function comment.
40962 2017-01-09 Martin Sebor <msebor@redhat.com>
40965 * asan.c (asan_emit_stack_protection): Increase local buffer size
40966 to avoid snprintf truncation warning.
40968 2017-01-09 Andrew Pinski <apinski@cavium.com>
40970 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40971 to reference thunderx2t99 for the tuning structure
40972 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40974 (thunderx2t99_extra_costs): This.
40975 * config/aarch64/aarch64-tune.md: Regenerate.
40976 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40977 (vulcan_addrcost_table): This.
40978 (vulcan_regmove_cost): Rename to ...
40979 (thunderx2t99_regmove_cost): This.
40980 (vulcan_vector_cost): Rename to ...
40981 (thunderx2t99_vector_cost): this.
40982 (vulcan_branch_cost): Rename to ...
40983 (thunderx2t99_branch_cost): This.
40984 (vulcan_tunings): Rename to ...
40985 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40986 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40988 2017-01-09 Martin Jambor <mjambor@suse.cz>
40992 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40993 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40994 (propagate_vr_accross_jump_function): Use the above function for all
40995 value range computations for pass-through jump functions and type
40996 converasion from explicit value range values.
40997 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40998 parameters from TYPE_ARG_TYPES.
40999 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41000 (ipa_write_node_info): Stream type of the actual argument.
41001 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41003 2017-01-09 Martin Liska <mliska@suse.cz>
41006 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41007 (lookup_compiler): Do not show error message with have_E.
41009 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41011 PR tree-optimization/78938
41012 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41013 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41014 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41015 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41018 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41020 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41023 2017-01-09 Richard Biener <rguenther@suse.de>
41025 PR tree-optimization/78997
41026 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41027 name condition properly.
41029 2017-01-09 Richard Biener <rguenther@suse.de>
41032 * dwarf2out.c (is_cxx): New overload with context.
41033 (is_naming_typedef_decl): Use it.
41035 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41037 * invoke.texi (Option Summary): Correct spacing in option lists
41038 and add line breaks to fix over-long lines.
41040 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41042 PR middle-end/17660
41044 * extend.texi (Common Variable Attributes): Add xref to GCC
41045 Internals manual to explain mode attribute keywords.
41047 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41050 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41051 and Preprocessor Options.
41052 (Options for Linking): Document -pthread here....
41053 (RS/6000 and PowerPC Options): ...not here.
41054 (Solaris 2 Options): ...or here.
41055 * doc/cppopts.texi: Document -pthread.
41057 2017-01-08 Martin Sebor <msebor@redhat.com>
41059 PR middle-end/77708
41060 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41061 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41062 New member functions.
41063 (format_directive): Used them.
41065 (pass_sprintf_length::handle_gimple_call): Same.
41066 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41067 to avoid truncation for any argument.
41068 (extract_affine_mul): Same.
41069 * tree.c (get_file_function_name): Same.
41071 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41073 PR middle-end/77484
41074 * predict.def (PRED_INDIR_CALL): Set to 86.
41076 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41078 PR preprocessor/54124
41079 * doc/cppopts.texi: Reformat -d subtable to list the full name
41080 of the options. Add cross-reference to the docs for the general
41081 compiler -d options.
41082 * doc/invoke.texi (Developer Options): Add cross-reference to the
41083 preprocessor-specific -d option documentation.
41085 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41087 PR preprocessor/13498
41088 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41089 redudant material, and reflect new command-line options.
41090 (System Headers): Likewise.
41092 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41094 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41095 -isystem, and -idirafter. Copy-edit.
41096 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41097 default for -ftrack-macro-expansion. Delete obsolete and
41098 badly-formatted implementation details about -fdebug-cpp output.
41099 * doc/cppwarnopts.texi: Copy-edit.
41101 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41104 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41105 that the transition from a max line width >= 1<<10 to narrower
41106 lines works correctly.
41108 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41110 * doc/options.texi (PerFunction): New.
41111 * opt-functions.awk (switch_flags): Map both Optimization and
41112 PerFunction to CL_OPTIMIZATION.
41113 * opth-gen.awk: Test for PerFunction flag along with
41115 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41116 it only when the latter is present. Skip those that don't in
41117 the hash function generator.
41118 * common.opt (fvar-tracking): Mark as PerFunction instead of
41120 (fvar-tracking-assignments): Likewise.
41121 (fvar-tracking-assignments-toggle): Likewise.
41122 (fvar-tracking-uninit): Likewise.
41124 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41126 PR translation/79018
41127 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41130 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41133 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41134 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41135 TARGET_LONG_JUMP_TABLE_OFFSETS.
41136 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41137 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41138 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41139 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41140 * config/m68k/m68k.md (tablejump expander): Likewise.
41141 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41142 TARGET_LONG_JUMP_TABLE_OFFSETS.
41143 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41144 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41146 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41147 David Holsgrove <david.holsgrove@xilinx.com>
41149 * common/config/microblaze/microblaze-common.c
41150 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41151 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41153 * config/microblaze/microblaze.c (microblaze_must_save_register)
41154 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41156 (microblaze_eh_return): New function.
41157 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41158 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41159 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41160 * config/microblaze/microblaze.md (eh_return): New pattern.
41162 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41164 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41165 GCC_DIAGNOSTIC_STRINGIFY): Define.
41167 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41169 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41171 * config/arm/arm.md (<mcrr>): New.
41173 * config/arm/arm.c (arm_arch5te): New.
41174 (arm_option_override): Set arm_arch5te.
41175 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41177 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41178 (arm_mcrr_qualifiers): ... this. New.
41179 (MRRC_QUALIFIERS): Define to...
41180 (arm_mrrc_qualifiers): ... this. New.
41181 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41183 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41184 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41185 (MRRCI, mrrc, MRRC): New.
41186 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41187 VUNSPEC_MRRC2): New.
41189 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41191 * config/arm/arm.md (<mcr>): New.
41193 * config/arm/arm.c (arm_coproc_builtin_available): Add
41194 support for mcr, mrc, mcr2 and mrc2.
41195 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41196 (arm_mcr_qualifiers): ... this. New.
41197 (MRC_QUALIFIERS): Define to ...
41198 (arm_mrc_qualifiers): ... this. New.
41199 (MCR_QUALIFIERS): Define to ...
41200 (arm_mcr_qualifiers): ... this. New.
41201 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41203 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41204 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41205 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41206 VUNSPEC_MRC2): New.
41208 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41210 * config/arm/arm.md (*ldc): New.
41214 * config/arm/arm.c (arm_coproc_builtin_available): Add
41215 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41216 (arm_coproc_ldc_stc_legitimate_address): New.
41217 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41218 'qualifier_const_pointer'.
41219 (LDC_QUALIFIERS): Define to...
41220 (arm_ldc_qualifiers): ... this. New.
41221 (STC_QUALIFIERS): Define to...
41222 (arm_stc_qualifiers): ... this. New.
41223 * config/arm/arm-protos.h
41224 (arm_coproc_ldc_stc_legitimate_address): New.
41225 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41226 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41227 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41228 stc2, stcl, stc2l): New.
41229 * config/arm/constraints.md (Uz): New.
41230 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41231 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41232 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41233 VUNSPEC_STC2L): New.
41235 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41237 * config/arm/arm.md (<cdp>): New.
41238 * config/arm/arm.c (neon_const_bounds): Rename this ...
41239 (arm_const_bounds): ... this.
41240 (arm_coproc_builtin_available): New.
41241 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41242 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41243 (CDP_QUALIFIERS): Define to...
41244 (arm_cdp_qualifiers): ... this. New.
41246 (arm_expand_builtin_args): Add case for 6 arguments.
41247 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41248 (arm_const_bounds): ... this.
41249 (arm_coproc_builtin_available): New.
41250 * config/arm/arm_acle.h (__arm_cdp): New.
41252 * config/arm/arm_acle_builtins.def (cdp): New.
41254 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41255 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41256 'arm_const_bounds'.
41257 * config/arm/types.md (coproc): New.
41258 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41259 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41260 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41261 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41263 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41265 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41266 (UBINOP_QUALIFIERS): New.
41268 (acle_builtin_data): New. Change comment.
41269 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41270 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41271 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41272 arm_acle_builtins.def.
41273 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41274 (arm_init_acle_builtins): New.
41275 (CRC32_BUILTIN): Remove.
41276 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41277 crc32cb, crc32ch and crc32cw.
41278 (arm_init_crc32_builtins): Remove.
41279 (arm_init_builtins): Use arm_init_acle_builtins rather
41280 than arm_init_crc32_builtins.
41281 (arm_expand_acle_builtin): New.
41282 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41283 * config/arm/arm_acle_builtins.def: New.
41285 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41287 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41288 (arm_builtin_datum): ... this.
41289 (arm_init_neon_builtin): Rename to ...
41290 (arm_init_builtin): ... this. Add a new parameters PREFIX
41291 and USE_SIG_IN_NAME.
41292 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41293 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41294 'arm_builtin_datum'.
41295 (arm_init_vfp_builtins): Likewise.
41296 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41297 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41298 (arm_expand_neon_args): Rename to ...
41299 (arm_expand_builtin_args): ... this. Rename builtin_arg
41300 enum values and differentiate between ARG_BUILTIN_MEMORY
41301 and ARG_BUILTIN_NEON_MEMORY.
41302 (arm_expand_neon_builtin_1): Rename to ...
41303 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41304 values, arm_expand_builtin_args and add bool parameter NEON.
41305 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41306 (arm_expand_vfp_builtin): Likewise.
41307 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41309 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41311 PR middle-end/77484
41312 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41313 * predict.c (tree_estimate_probability_bb): Reverse direction of
41314 polymorphic call predictor.
41316 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41318 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41319 (determine_pass_name_match): ...this new function and...
41320 (should_skip_pass_p): ...this new function.
41322 2017-01-06 Nathan Sidwell <nathan@acm.org>
41324 * ipa-visibility.c (function_and_variable_visibility): Reformat
41325 comments and long lines. Remove extrneous if.
41326 * symtab.c (symtab_node::make_decl_local): Fix code format.
41327 (symtab_node::set_section_for_node): Fix comment typo.
41329 2017-01-06 Martin Liska <mliska@suse.cz>
41332 * lra-constraints.c: Rename invariant to lra_invariant.
41333 * predict.c (set_even_probabilities): Initialize e to NULL.
41335 2017-01-05 Martin Sebor <msebor@redhat.com>
41337 PR tree-optimization/78910
41338 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41339 (format_integer): Correct off-by-one error in the handling
41340 of precision with negative numbers in signed conversions..
41342 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41344 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41346 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41348 PR tree-optimization/71016
41349 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41350 factor_out_conditional_conversion. Formatting fix.
41351 (factor_out_conditional_conversion): Add cond_stmt argument.
41352 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41353 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41356 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41358 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41359 read-rtl-function.o, and selftest-rtl.o.
41360 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41361 (selftest::aarch64_test_loading_full_dump): New function.
41362 (selftest::aarch64_run_selftests): New function.
41363 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41364 selftest::aarch64_run_selftests.
41365 * config/i386/i386.c
41366 (selftest::ix86_test_loading_dump_fragment_1): New function.
41367 (selftest::ix86_test_loading_call_insn): New function.
41368 (selftest::ix86_test_loading_full_dump): New function.
41369 (selftest::ix86_test_loading_unspec): New function.
41370 (selftest::ix86_run_selftests): Call the new functions.
41371 * emit-rtl.c (maybe_set_max_label_num): New function.
41372 * emit-rtl.h (maybe_set_max_label_num): New decl.
41373 * function.c (instantiate_decls): Guard call to
41374 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41375 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41377 * gensupport.c (gen_reader::gen_reader): Pass "false"
41378 for new "compact" param of rtx_reader.
41379 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41380 rather than an empty string for NULL strings.
41381 * read-md.c: Potentially include config.h rather than bconfig.h.
41382 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41383 (have_error): New global, copied from errors.c.
41384 (md_reader::read_name): Rename to...
41385 (md_reader::read_name_1): ...this, adding "out_loc" param,
41386 and converting "missing name or number" to returning false, rather
41388 (md_reader::read_name): Reimplement in terms of read_name_1.
41389 (md_reader::read_name_or_nil): New function.
41390 (md_reader::read_string): Handle "(nil)" by returning NULL.
41391 (md_reader::md_reader): Add new param "compact".
41392 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41393 (md_reader::read_file): New method.
41394 * read-md.h (md_reader::md_reader): Add new param "compact".
41395 (md_reader::read_file): New method.
41396 (md_reader::is_compact): New accessor.
41397 (md_reader::read_name): Convert return type from void to file_location.
41398 (md_reader::read_name_or_nil): New decl.
41399 (md_reader::read_name_1): New decl.
41400 (md_reader::m_compact): New field.
41401 (noop_reader::noop_reader): Pass "false" for new "compact" param
41403 (rtx_reader::rtx_reader): Add new "compact" param.
41404 (rtx_reader::read_rtx_operand): Make virtual and convert return
41405 type from void to rtx.
41406 (rtx_reader::read_until): New decl.
41407 (rtx_reader::handle_any_trailing_information): New virtual function.
41408 (rtx_reader::postprocess): New virtual function.
41409 (rtx_reader::finalize_string): New virtual function.
41410 (rtx_reader::m_in_call_function_usage): New field.
41411 (rtx_reader::m_reuse_rtx_by_id): New field.
41412 * read-rtl-function.c: New file.
41413 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41414 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41415 (selftest::verify_three_block_rtl_cfg): New decl.
41416 * read-rtl-function.h: New file.
41417 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41418 For host, include function.h, memmodel.h, and emit-rtl.h.
41419 (one_time_initialization): New function.
41420 (struct compact_insn_name): New struct.
41421 (compact_insn_names): New array.
41422 (find_code): Handle insn codes in compact dumps.
41423 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41424 (bind_subst_iter_and_attr): Likewise.
41425 (add_condition_to_string): Likewise.
41426 (add_condition_to_rtx): Likewise.
41427 (apply_attribute_uses): Likewise.
41428 (add_current_iterators): Likewise.
41429 (apply_iterators): Likewise.
41430 (initialize_iterators): Guard usage of apply_subst_iterator with
41431 #ifdef GENERATOR_FILE.
41432 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41433 (md_reader::read_mapping): Likewise.
41434 (add_define_attr_for_define_subst): Likewise.
41435 (add_define_subst_attr): Likewise.
41436 (read_subst_mapping): Likewise.
41437 (check_code_iterator): Likewise.
41438 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41440 (one_time_initialization): New function.
41441 (rtx_reader::read_until): New method.
41442 (read_flags): New function.
41443 (parse_reg_note_name): New function.
41444 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41445 Handle reuse_rtx ids.
41446 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41447 Add parsing support for RTL dumps, mirroring the special-cases in
41448 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41449 values, and calling handle_any_trailing_information.
41450 (rtx_reader::read_rtx_operand): Convert return type from void
41451 to rtx, returning return_rtx. Handle case 'e'. Call
41452 finalize_string on XSTR and XTMPL fields.
41453 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41454 "(nil)" values were omitted. Call the postprocess vfunc on the
41456 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41457 class ctor. Initialize m_in_call_function_usage. Call
41458 one_time_initialization.
41459 * rtl-tests.c (selftest::test_uncond_jump): Call
41460 set_new_first_and_last_insn.
41461 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41462 * selftest-rtl.c: New file.
41463 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41464 (selftest::get_insn_by_uid): New decl.
41465 * selftest-run-tests.c (selftest::run_tests): Call
41466 read_rtl_function_c_tests.
41467 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41468 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41471 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41473 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41474 operands in a special way. Assert that pos+len <= mode precision.
41476 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41478 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41479 3 argument Alias with unlimited for the negative form.
41480 (fno-vect-cost-model): Removed.
41482 2017-01-05 Martin Liska <mliska@suse.cz>
41484 * hsa-gen.c (gen_hsa_divmod): New function.
41485 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41487 2017-01-05 Martin Liska <mliska@suse.cz>
41490 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41493 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41495 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41496 small constant length operands.
41498 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41500 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41501 between loop iterations.
41503 2017-01-05 Martin Liska <mliska@suse.cz>
41506 * gimplify.c (gimplify_decl_expr): Compare to
41507 asan_poisoned_variables instread of checking flags.
41508 (gimplify_target_expr): Likewise.
41509 (gimplify_expr): Likewise.
41510 (gimplify_function_tree): Conditionally initialize
41511 asan_poisoned_variables.
41513 2017-01-04 Jeff Law <law@redhat.com>
41515 PR tree-optimizatin/78812
41516 * rtl.h (contains_mem_rtx_p): Prototype.
41517 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41518 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41519 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41520 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41521 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41523 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41525 * input.c (assert_char_at_range): Default-initialize actual_range.
41527 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41529 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41530 to match the caller.
41532 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41534 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41535 insns after final jump in test to emit dummy move.
41537 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41539 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41540 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41542 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41544 * multiple_target.c (create_dispatcher_calls): Init e_next.
41545 * tree-ssa-loop-split.c (split_loop): Init border.
41546 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41549 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41554 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41555 (altivec_register_operand): Do not return true if the operand
41556 contains a SUBREG mixing SImode and SFmode.
41557 (vsx_register_operand): Likewise.
41558 (vsx_reg_sfsubreg_ok): New predicate.
41559 (vfloat_operand): Do not return true if the operand contains a
41560 SUBREG mixing SImode and SFmode.
41561 (vint_operand): Likewise.
41562 (vlogical_operand): Likewise.
41563 (gpc_reg_operand): Likewise.
41564 (int_reg_operand): Likewise.
41565 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41566 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41567 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41569 (rs6000_emit_move_si_sf_subreg): New helper function.
41570 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41571 fixup SUBREGs involving SImode and SFmode.
41572 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41573 numbers for the new peephole2 optimization.
41574 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41575 the GLIBC math library that do AND/IOR/XOR operations on single
41576 precision floating point.
41577 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41578 target macros to say whether we need to avoid SUBREGs mixing
41580 (TARGET_ALLOW_SF_SUBREG): Likewise.
41581 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41582 (UNSPEC_SI_FROM_SF): Likewise.
41583 (iorxor): Change spacing.
41584 (and_ior_xor): New iterator for AND, IOR, and XOR.
41585 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41586 (movdi_from_sf_zero_ext): Likewise.
41587 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41588 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41589 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41590 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41591 (fms<mode>4): Likewise.
41592 (fnma<mode>4): Likewise.
41593 (fnms<mode>4): Likewise.
41594 (nfma<mode>4): Likewise.
41595 (nfms<mode>4): Likewise.
41597 2017-01-04 Marek Polacek <polacek@redhat.com>
41600 * doc/invoke.texi: Document -Wpointer-compare.
41602 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41604 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41607 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41608 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41611 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41614 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41615 documentation of the powerpc_popcntb_ok attribute.
41616 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41617 code to issue warning messages if a requested CPU configuration is
41618 not supported by the binary (assembler and loader) toolchain.
41619 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41620 made to define a built-in function that has been disabled.
41621 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41622 made to define a built-in function that has been disabled.
41623 (altivec_init_builtins): Add comment explaining why definition
41624 of the DST built-in functions is not preceded by an assertion
41625 check. Add assertions to prevent ICE if attempts are made to
41626 define an altivec predicate or an abs* built-in function that has
41628 (htm_init_builtins): Add comment explaining why definition of the
41629 htm built-in functions is not preceded by an assertion check.
41631 2017-01-04 Jeff Law <law@redhat.com>
41633 PR tree-optimizatin/67955
41634 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41635 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41636 the points-to solution does not include pt_null. Use DECL_PT_UID
41639 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41641 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41642 Use gen_int_mode instead of gen_lopwart for const_int operands.
41644 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41646 PR tree-optimization/71563
41647 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41648 out of range value - has low bits known to be zero.
41650 2017-01-04 Alan Modra <amodra@gmail.com>
41652 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41653 * configure: Regenerate.
41654 * config.in: Regenerate.
41656 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41659 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41660 a substring of the message, but strcmp with the whole message. Ifdef
41661 ENABLE_NLS, translate the message first using dgettext.
41663 2017-01-03 Jeff Law <law@redhat.com>
41665 PR tree-optimizatin/78856
41666 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41667 (mark_threaded_blocks): Remove code to truncate thread paths that
41668 cross multiple loop headers. Instead invalidate the cached loop
41669 iteration information and handle case of a thread path walking
41670 into an irreducible region.
41672 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41675 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41676 assertions. Add support for doing the signbit if the IEEE 128-bit
41677 floating point value is in a GPR.
41678 * config/rs6000/rs6000.md (Fsignbit): Delete.
41679 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41680 Update the length attribute if the value is in a GPR.
41681 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41682 the sign or zero extension instruction, since the value is always 0/1.
41683 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41686 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41687 extracting SImode to a GPR register so that we can generate a
41688 store, limit the vector to be in a traditional Altivec register
41689 for the vextuwrx instruction.
41691 2017-01-03 Ian Lance Taylor <iant@google.com>
41693 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41695 2017-01-03 Martin Sebor <msebor@redhat.com>
41697 PR tree-optimization/78696
41698 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41699 precision. Use MPFR for %f for greater fidelity. Correct handling
41701 (pass_sprintf_length::compute_format_length): Set width and precision
41702 specified by asrerisk to void_node for vararg functions.
41703 (try_substitute_return_value): Adjust dump output.
41705 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41707 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41709 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41711 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41712 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41713 -mlra/-mno-lra was passed to the compiler.
41715 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41717 PR rtl-optimization/65618
41718 * emit-rtl.c (try_split): Move initialization of "before" and
41719 "after" to just before the call to emit_insn_after_setloc.
41721 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41723 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41725 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41727 * dwarf2out.c (gen_enumeration_type_die): When
41728 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41730 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41732 PR tree-optimization/78965
41733 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41734 Change first argument from const call_info & to call_info &. For %n
41735 set info.nowrite to false.
41737 PR middle-end/78901
41738 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41739 possibly throwing calls.
41741 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41742 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41743 and fns handling, rather than in a separate case SSA_NAME.
41745 2017-01-02 Jeff Law <law@redhat.com>
41747 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41748 fixes for first_period and second_period variables.
41750 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41753 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41754 (*insvqi_1): New insn pattern.
41755 (*insvqi_1_mem_rex64): Ditto.
41756 (*insvqi_2): Ditto.
41757 (*insvqi_3): Rename from *insvqi.
41759 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41761 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41763 * doc/cfg.texi (Edges): Remove reference to Java.
41764 (Maintaining the CFG): Ditto.
41766 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41768 PR middle-end/77674
41769 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41770 transparent aliases.
41772 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41774 PR middle-end/77484
41775 * predict.def (PRED_CALL): Update hitrate.
41776 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41777 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41778 into direct/indirect/polymorphic variants.
41780 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41782 Update copyright years.
41784 * gcc.c (process_command): Update copyright notice dates.
41785 * gcov-dump.c (print_version): Ditto.
41786 * gcov.c (print_version): Ditto.
41787 * gcov-tool.c (print_version): Ditto.
41788 * gengtype.c (create_file): Ditto.
41789 * doc/cpp.texi: Bump @copying's copyright year.
41790 * doc/cppinternals.texi: Ditto.
41791 * doc/gcc.texi: Ditto.
41792 * doc/gccint.texi: Ditto.
41793 * doc/gcov.texi: Ditto.
41794 * doc/install.texi: Ditto.
41795 * doc/invoke.texi: Ditto.
41797 Copyright (C) 2017 Free Software Foundation, Inc.
41799 Copying and distribution of this file, with or without modification,
41800 are permitted in any medium without royalty provided the copyright
41801 notice and this notice are preserved.