1 2018-01-01 Jakub Jelinek <jakub@redhat.com>
4 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
5 last argument when extracting from CONCAT. If either from_real or
6 from_imag is NULL, use expansion through memory. If result is not
7 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
8 the parts directly to inner mode, if even that fails, use expansion
12 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
13 check for bswap in mode rather than HImode and use that in expand_unop
16 2017-12-31 Jakub Jelinek <jakub@redhat.com>
19 * config/i386/i386.c (ix86_attribute_table): Remove excess
20 initializer for "shared" attribute.
22 2017-12-30 Tom de Vries <tom@codesourcery.com>
25 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
27 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
28 offload_funcs entries that no longer have a corresponding cgraph_node.
29 Mark the remaining ones as DECL_PRESERVE_P.
30 (output_lto): Call prune_offload_funcs.
32 2017-12-30 Jakub Jelinek <jakub@redhat.com>
34 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
35 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
36 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
37 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
38 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
39 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
40 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
41 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
42 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
43 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
44 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
45 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
46 vaesenc_<mode>, vpclmulqdq_<mode>,
47 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
49 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
51 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
52 _Float<N> and _Float<N>X variants for rounding built-in
54 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
55 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
56 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
57 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
58 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
59 * builtins.c (mathfn_built_in_2): Likewise.
60 * internal-fn.def (CEIL): Likewise.
62 (NEARBYINT): Likewise.
66 * convert.c (convert_to_integer_1): Likewise.
67 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
68 (integer_valued_real_call_p): Likewise.
69 * fold-const-call.c (fold_const_call_ss): Likewise.
70 * gencfn-macros.c (print_case_cfn): Change CFN and operator
71 printers to take a const char * suffix instead of a bool.
72 (print_define_operator_list): Likewise.
73 (fltall_suffixes): New list of suffixes, that include the
74 traditional suffixes as well as all of the _Float<N> and
76 (main): For _Float<N> and _Float<N>X functions, emit both
77 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
78 has the _Float<N> and _Float<N>X case names or operators. The
79 <name>_ALL variant has both the traditional and the
80 _Float<N>/_Float<N>X case names or operators.
81 * match.pd (COPYSIGN optimizations): Provide optimizations for
82 _Float<N> and _Float<N>X types where possible.
83 (MIN/MAX optimizations): Likewise.
84 (sqrt optimizations): Likewise.
85 (rounding optimizations): Likewise.
87 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
89 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
90 gen_int_mode rather than GEN_INT.
91 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
93 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
95 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
96 instead of CONSTANT_P.
97 (gen_vec_series): Likewise.
98 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
100 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
102 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
105 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
107 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
108 (aarch64_ld1x2<VDC:mode>): Likewise.
109 (aarch64_simd_ld1<mode>_x2): Likewise.
110 (aarch64_simd_ld1<mode>_x2): Likewise.
111 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
112 (vld1_s8_x2): Likewise.
113 (vld1_u16_x2): Likewise.
114 (vld1_s16_x2): Likewise.
115 (vld1_u32_x2): Likewise.
116 (vld1_s32_x2): Likewise.
117 (vld1_u64_x2): Likewise.
118 (vld1_s64_x2): Likewise.
119 (vld1_f16_x2): Likewise.
120 (vld1_f32_x2): Likewise.
121 (vld1_f64_x2): Likewise.
122 (vld1_p8_x2): Likewise.
123 (vld1_p16_x2): Likewise.
124 (vld1_p64_x2): Likewise.
125 (vld1q_u8_x2): Likewise.
126 (vld1q_s8_x2): Likewise.
127 (vld1q_u16_x2): Likewise.
128 (vld1q_s16_x2): Likewise.
129 (vld1q_u32_x2): Likewise.
130 (vld1q_s32_x2): Likewise.
131 (vld1q_u64_x2): Likewise.
132 (vld1q_s64_x2): Likewise.
133 (vld1q_f16_x2): Likewise.
134 (vld1q_f32_x2): Likewise.
135 (vld1q_f64_x2): Likewise.
136 (vld1q_p8_x2): Likewise.
137 (vld1q_p16_x2): Likewise.
138 (vld1q_p64_x2): Likewise.
140 2017-12-27 Martin Liska <mliska@suse.cz>
142 PR tree-optimization/83552
143 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
144 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
146 2017-12-27 Tom de Vries <tom@codesourcery.com>
148 * config/nvptx/nvptx.c (nvptx_option_override): Disable
149 -gstatement-frontiers.
151 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
153 PR rtl-optimization/83513
154 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
155 before priority comparison.
157 2017-12-25 Jakub Jelinek <jakub@redhat.com>
160 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
161 ix86_isa_flags2 to ix86_isa_flags.
162 * config/i386/i386-c.c (ix86_target_macros_internal): Test
163 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
164 isa_flags rather than isa_flags2.
165 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
166 and -mavx512bitalg from isa2_opts to isa_opts.
167 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
168 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
169 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
171 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
172 OPTION_MASK_ISA_AVX512F ored with another option similarly to
173 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
174 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
175 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
176 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
177 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
178 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
179 builtins from bdesc_args2 to bdesc_args section.
180 (__builtin_ia32_compressstoreuqi512_mask,
181 __builtin_ia32_compressstoreuhi512_mask,
182 __builtin_ia32_compressstoreuqi256_mask,
183 __builtin_ia32_expandloadqi512_mask,
184 __builtin_ia32_expandloadqi512_maskz,
185 __builtin_ia32_expandloadhi512_mask,
186 __builtin_ia32_expandloadhi512_maskz,
187 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
188 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
189 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
190 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
191 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
192 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
193 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
194 __builtin_ia32_vpshldv_v32hi_maskz,
195 __builtin_ia32_vpopcountb_v64qi_mask,
196 __builtin_ia32_vpopcountw_v32hi_mask,
197 __builtin_ia32_vpshufbitqmb512_mask,
198 __builtin_ia32_vpshufbitqmb256_mask): Add
199 " | OPTION_MASK_ISA_AVX512BW".
200 (__builtin_ia32_expandloadqi256_mask,
201 __builtin_ia32_expandloadqi256_maskz,
202 __builtin_ia32_vpopcountb_v32qi_mask): Add
203 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
204 (__builtin_ia32_expandloadhi256_mask,
205 __builtin_ia32_expandloadhi256_maskz,
206 __builtin_ia32_expandloadqi128_mask,
207 __builtin_ia32_expandloadqi128_maskz,
208 __builtin_ia32_expandloadhi128_mask,
209 __builtin_ia32_expandloadhi128_maskz,
210 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
211 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
212 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
213 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
214 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
215 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
216 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
217 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
218 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
219 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
220 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
221 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
222 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
223 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
224 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
225 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
226 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
227 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
228 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
229 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
230 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
231 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
232 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
233 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
234 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
235 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
236 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
237 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
238 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
239 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
240 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
241 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
242 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
243 __builtin_ia32_vpopcountw_v8hi_mask): Add
244 " | OPTION_MASK_ISA_AVX512VL".
245 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
246 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
247 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
248 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
249 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
250 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
251 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
252 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
253 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
254 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
255 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
256 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
257 _mm_mask_bitshuffle_epi64_mask): Likewise.
258 * common/config/i386/i386-common.c
259 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
260 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
261 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
262 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
263 OPTION_MASK_ISA_AVX512BITALG_UNSET.
264 (OPTION_MASK_ISA2_AVX512F_UNSET,
265 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
266 (ix86_handle_option): For -mno-general-regs-only, clear from
267 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
268 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
269 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
270 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
271 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
272 for moving from ix86_isa_flags2 to ix86_isa_flags.
274 2017-12-23 Jakub Jelinek <jakub@redhat.com>
277 * fold-const.c (struct contains_label_data): New type.
278 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
279 inside of a SWITCH_BODY seen during the walk.
280 (contains_label_p): Use walk_tree instead of
281 walk_tree_without_duplicates, prepare data for contains_label_1 and
284 2017-12-22 Jakub Jelinek <jakub@redhat.com>
287 * tree-iterator.c (alloc_stmt_list): Start with cleared
288 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
292 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
293 enable avx512f explicitly in #pragma GCC target.
294 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
295 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
296 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
297 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
298 __builtin_ia32_vpdpbusds_v8si_mask,
299 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
300 __builtin_ia32_vpdpbusds_v4si_mask,
301 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
302 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
303 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
304 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
305 __builtin_ia32_vpdpwssds_v8si_mask,
306 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
307 __builtin_ia32_vpdpwssds_v4si_mask,
308 __builtin_ia32_vpdpwssds_v4si_maskz): Use
309 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
310 just OPTION_MASK_ISA_AVX512VNNI.
312 2017-12-22 Martin Jambor <mjambor@suse.cz>
315 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
318 2017-12-22 Julia Koval <julia.koval@intel.com>
319 Sebastian Peryt <sebastian.peryt@intel.com>
321 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
322 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
323 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
324 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
325 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
326 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
327 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
328 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
329 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
330 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
331 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
332 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
333 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
334 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
335 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
336 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
337 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
338 _mm256_maskz_popcnt_epi64): New intrinsics.
339 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
340 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
341 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
342 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
343 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
344 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
345 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
346 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
347 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
348 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
349 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
350 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
351 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
352 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
353 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
354 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
355 __builtin_ia32_vpshufbitqmb256_mask,
356 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
357 * config/i386/i386-c.c (__AVX512BITALG__): New.
358 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
359 (ix86_valid_target_attribute_inner_p): Ditto.
360 (ix86_expand_args_builtin): Handle new types.
361 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
362 * config/i386/i386.opt: Add -mavx512bitalg.
363 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
364 avx512bitalgintrin.h.
365 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
366 (vpopcount<mode><mask_name>): Add more types.
367 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
368 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
370 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
372 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
373 Or in OPTION_MASK_ISA_AVX512F_SET.
374 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
375 OPTION_MASK_ISA_AVX512VNNI_UNSET.
376 (ix86_handle_option): Adjust for
377 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
378 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
380 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
381 isa_flag instead of isa_flag2.
382 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
383 isa_opts2 to isa_opts.
384 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
387 2017-12-22 Mike Stump <mikestump@comcast.net>
388 Eric Botcazou <ebotcazou@adacore.com>
390 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
392 2017-12-21 Jakub Jelinek <jakub@redhat.com>
395 * config/i386/i386.c (ix86_function_arg_boundary): Return
396 PARM_BOUNDARY for TYPE_EMPTY_P types.
399 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
400 if navail is >= dir.len.
402 2017-12-21 Steve Ellcey <sellcey@cavium.com>
404 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
407 2017-12-21 Jakub Jelinek <jakub@redhat.com>
409 PR rtl-optimization/80747
410 PR rtl-optimization/83512
411 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
412 succ edge from ENTRY, copy partition from e->dest to the newly
414 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
415 ENTRY, use BB_PARTITION of its successor block as current_partition.
416 Don't copy partition when splitting succ edge from ENTRY.
418 PR tree-optimization/83523
419 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
420 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
421 (convert_mult_to_fma): Likewise.
423 PR tree-optimization/83521
424 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
425 gimple_build_assign without code on result of
426 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
429 2017-12-21 Andrew Pinski <apinski@cavium.com>
430 Steve Ellcey <sellcey@cavium.com>
432 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
433 multi-arch for ilp32.
435 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
438 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
439 constraints to operand 2.
440 (*ashl<mode>3_mask_1): Ditto.
441 (*<shift_insn><mode>3_mask): Ditto.
442 (*<shift_insn><mode>3_mask_1): Ditto.
443 (*<rotate_insn><mode>3_mask): Ditto.
444 (*<rotate_insn><mode>3_mask_1): Ditto.
446 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
448 * reorg.c (make_return_insns): Reemit each insn with its own location.
450 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
452 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
453 generation for cases where splatting a value is not useful.
454 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
455 across a vec_duplicate and a paradoxical subreg forming a vector
456 mode to a vec_concat.
458 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
460 * combine.c (simplify_set): Do not transform subregs to zero_extends
461 if the destination is not a scalar int mode.
463 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
466 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
467 shared zero if the input has overflowed.
469 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
471 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
472 support for Cortex-A55 and Cortex-A75.
474 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
476 * common/config/arm/arm-common.c (compare_opt_names): Add function
477 comment. Use strcmp instead of manual loop.
479 2017-12-21 Martin Liska <mliska@suse.cz>
481 PR gcov-profile/83509
482 * gcov-dump.c (dump_gcov_file): Do not read info about
483 support_unexecuted_blocks for gcda files.
485 2017-12-21 Jakub Jelinek <jakub@redhat.com>
487 PR rtl-optimization/82973
488 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
489 (valid_for_const_vector_p): ... this.
490 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
491 (valid_for_const_vector_p): ... this. Adjust function comment.
492 (gen_vec_duplicate): Adjust caller.
493 * optabs.c (expand_vector_broadcast): Likewise.
494 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
495 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
497 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
498 instead of GET_CODE == CONST_FIXED.
499 (simplify_subreg): Use CONST_FIXED_P macro instead of
500 GET_CODE == CONST_FIXED.
503 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
504 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
505 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
506 (ix86_option_override_internal): Adjust for
507 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
508 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
509 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
510 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
511 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
512 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
513 def_builtin2 for CET builtins.
514 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
515 up formatting in IX86_BUILTIN_RDPID code.
516 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
517 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
518 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
519 ix86_isa_flags2 to ix86_isa_flags.
520 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
522 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
523 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
524 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
526 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
527 Or in OPTION_MASK_ISA_AVX512F_SET.
528 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
529 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
530 (ix86_handle_option): Adjust for
531 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
532 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
534 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
535 Alan Hayward <alan.hayward@arm.com>
536 David Sherwood <david.sherwood@arm.com>
538 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
539 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
540 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
541 Track polynomial offsets.
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-vect-data-refs.c (vect_compute_data_ref_alignment):
548 Treat drb->init as a poly_int. Fail if its misalignment wrt
549 vector_alignment isn't known.
551 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
552 Alan Hayward <alan.hayward@arm.com>
553 David Sherwood <david.sherwood@arm.com>
555 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
557 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
558 (vect_loop_versioning): Take the loop versioning threshold as a
560 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
561 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
562 versioning_threshold.
563 (vect_analyze_loop_2): Compute the loop versioning threshold
564 whenever loop versioning is needed, and store it in the new
565 field rather than combining it with the cost model threshold.
566 (vect_transform_loop): Update call to vect_loop_versioning.
567 Try to combine the loop versioning and cost thresholds here.
569 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
570 Alan Hayward <alan.hayward@arm.com>
571 David Sherwood <david.sherwood@arm.com>
573 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
574 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
575 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
576 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
577 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
578 (group_compare_offset): Update accordingly.
579 (split_small_address_groups_p): Likewise.
580 (record_use): Take addr_offset as a poly_uint64 rather than
581 an unsigned HOST_WIDE_INT.
582 (strip_offset): Return the offset as a poly_uint64 rather than
583 an unsigned HOST_WIDE_INT.
584 (record_group_use, split_address_groups): Track polynomial offsets.
585 (add_iv_candidate_for_use): Likewise.
586 (addr_offset_valid_p): Take the offset as a poly_int64 rather
587 than a HOST_WIDE_INT.
588 (strip_offset_1): Return the offset as a poly_int64 rather than
591 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
592 Alan Hayward <alan.hayward@arm.com>
593 David Sherwood <david.sherwood@arm.com>
595 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
596 rather than a HOST_WIDE_INT.
597 * tree.c (get_binfo_at_offset): Likewise.
599 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
600 Alan Hayward <alan.hayward@arm.com>
601 David Sherwood <david.sherwood@arm.com>
603 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
604 rather than a HOST_WIDE_INT.
605 * tree-sra.c (build_ref_for_offset): Likewise.
607 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
608 Alan Hayward <alan.hayward@arm.com>
609 David Sherwood <david.sherwood@arm.com>
611 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
613 * tree.c (mem_ref_offset): Likewise.
614 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
615 * builtins.c (get_object_alignment_2): Likewise.
616 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
617 * gimple-fold.c (get_base_constructor): Likewise.
618 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
619 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
621 * ipa-polymorphic-call.c
622 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
623 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
624 (get_ancestor_addr_info): Likewise.
625 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
626 * match.pd: Likewise.
627 * tree-data-ref.c (dr_analyze_innermost): Likewise.
628 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
629 * tree-eh.c (tree_could_trap_p): Likewise.
630 * tree-object-size.c (addr_object_size): Likewise.
631 * tree-ssa-address.c (copy_ref_info): Likewise.
632 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
633 (indirect_refs_may_alias_p): Likewise.
634 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
635 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
636 (non_rewritable_mem_ref_base): Likewise.
637 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
638 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
639 * varasm.c (decode_addr_const): Likewise.
641 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
642 Alan Hayward <alan.hayward@arm.com>
643 David Sherwood <david.sherwood@arm.com>
645 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
646 offsets for MEM_REFs.
648 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
649 Alan Hayward <alan.hayward@arm.com>
650 David Sherwood <david.sherwood@arm.com>
652 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
653 as a poly_uint64 rather than an unsigned int.
654 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
656 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
657 Alan Hayward <alan.hayward@arm.com>
658 David Sherwood <david.sherwood@arm.com>
660 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
662 (decode_addr_const): Update accordingly.
664 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
665 Alan Hayward <alan.hayward@arm.com>
666 David Sherwood <david.sherwood@arm.com>
668 * tree.h (bit_field_size, bit_field_offset): New functions.
669 * hsa-gen.c (gen_hsa_addr): Use them.
670 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
671 (simplify_vector_constructor): Likewise.
672 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
673 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
674 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
675 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
678 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
679 Alan Hayward <alan.hayward@arm.com>
680 David Sherwood <david.sherwood@arm.com>
682 * expr.h (emit_group_load, emit_group_load_into_temps)
683 (emit_group_store): Take the size as a poly_int64 rather than an int.
684 * expr.c (emit_group_load_1, emit_group_load): Likewise.
685 (emit_group_load_into_temp, emit_group_store): Likewise.
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 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
692 an unsigned int to a poly_uint64.
693 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
694 sizes as poly_uint64s rather than unsigned ints.
695 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
698 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
699 Alan Hayward <alan.hayward@arm.com>
700 David Sherwood <david.sherwood@arm.com>
702 * lra-constraints.c (emit_inc): Change inc_amount from an int
705 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
706 Alan Hayward <alan.hayward@arm.com>
707 David Sherwood <david.sherwood@arm.com>
709 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
711 (add_stack_var, stack_var_cmp, partition_stack_vars)
712 (dump_stack_var_partition): Update accordingly.
713 (alloc_stack_frame_space): Take the size as a poly_int64 rather
714 than a HOST_WIDE_INT.
715 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
716 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
717 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
718 rather than a HOST_WIDE_INT.
720 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
721 Alan Hayward <alan.hayward@arm.com>
722 David Sherwood <david.sherwood@arm.com>
724 * target.def (return_pops_args): Treat both the input and output
725 sizes as poly_int64s rather than HOST_WIDE_INTS.
726 * targhooks.h (default_return_pops_args): Update accordingly.
727 * targhooks.c (default_return_pops_args): Likewise.
728 * doc/tm.texi: Regenerate.
729 * emit-rtl.h (incoming_args): Change pops_args, size and
730 outgoing_args_size from int to poly_int64_pod.
731 * function.h (expr_status): Change x_pending_stack_adjust and
732 x_stack_pointer_delta from int to poly_int64.
733 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
734 (ARGS_SIZE_RTX): Update accordingly.
735 * calls.c (highest_outgoing_arg_in_use): Change from int to
737 (stack_usage_watermark, stored_args_watermark): New variables.
738 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
739 (emit_call_1): Change the stack_size and rounded_stack_size
740 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
742 (save_fixed_argument_area): Check stack_usage_watermark.
743 (initialize_argument_information): Change old_pending_adj from
744 a HOST_WIDE_INT * to a poly_int64_pod *.
745 (compute_argument_block_size): Return the size as a poly_int64
747 (finalize_must_preallocate): Track polynomial argument sizes.
748 (compute_argument_addresses): Likewise.
749 (internal_arg_pointer_based_exp): Track polynomial offsets.
750 (mem_overlaps_already_clobbered_arg_p): Rename to...
751 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
752 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
753 Check stored_args_used_watermark.
754 (load_register_parameters): Update accordingly.
755 (check_sibcall_argument_overlap_1): Likewise.
756 (combine_pending_stack_adjustment_and_call): Take the unadjusted
757 args size as a poly_int64 rather than an int. Return a bool
758 indicating whether the optimization was possible and return
759 the new adjustment by reference.
760 (check_sibcall_argument_overlap): Track polynomail argument sizes.
761 Update stored_args_watermark.
762 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
763 (expand_call): Likewise. Maintain stack_usage_watermark and
764 stored_args_watermark. Update calls to
765 combine_pending_stack_adjustment_and_call.
766 (emit_library_call_value_1): Handle polynomial argument sizes.
767 Call stack_region_maybe_used_p and mark_stack_region_used.
768 Maintain stack_usage_watermark.
769 (store_one_arg): Likewise. Update call to
770 mem_overlaps_already_clobbered_arg_p.
771 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
773 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
774 * config/microblaze/microblaze.c (microblaze_function_prologue):
776 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
777 TARGET_RETURN_POPS_ARGS interface.
778 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
780 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
781 * config/i386/i386.c (ix86_return_pops_args): Update for new
782 TARGET_RETURN_POPS_ARGS interface.
783 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
784 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
785 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
786 TARGET_RETURN_POPS_ARGS interface.
787 * config/vax/vax.c (vax_return_pops_args): Likewise.
788 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
789 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
790 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
791 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
792 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
793 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
794 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
795 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
796 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
797 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
798 and x_stack_pointer_delta from int to poly_int64.
799 * dojump.c (do_pending_stack_adjust): Update accordingly.
800 * explow.c (allocate_dynamic_stack_space): Handle polynomial
801 stack_pointer_deltas.
802 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
803 (pad_to_arg_alignment): Track polynomial offsets.
804 (assign_parm_find_stack_rtl): Likewise.
805 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
806 * toplev.c (output_stack_usage): Update reference to
807 current_function_pushed_stack_size.
809 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
810 Alan Hayward <alan.hayward@arm.com>
811 David Sherwood <david.sherwood@arm.com>
813 * function.c (in_arg_offset, var_offset, dynamic_offset)
814 (out_arg_offset, cfa_offset): Change from int to poly_int64.
815 (instantiate_new_reg): Return the new offset as a poly_int64_pod
816 rather than a HOST_WIDE_INT.
817 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
818 (instantiate_virtual_regs_in_insn): Likewise.
820 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
821 Alan Hayward <alan.hayward@arm.com>
822 David Sherwood <david.sherwood@arm.com>
824 * rtl.h (get_args_size, add_args_size_note): New functions.
825 (find_args_size_adjust): Return a poly_int64 rather than a
827 (fixup_args_size_notes): Likewise. Make the same change to the
828 end_args_size parameter.
829 * rtlanal.c (get_args_size, add_args_size_note): New functions.
830 * builtins.c (expand_builtin_trap): Use add_args_size_note.
831 * calls.c (emit_call_1): Likewise.
832 * explow.c (adjust_stack_1): Likewise.
833 * cfgcleanup.c (old_insns_match_p): Update use of
834 find_args_size_adjust.
835 * combine.c (distribute_notes): Track polynomial arg sizes.
836 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
837 end_true_args_size, beg_delay_args_size and end_delay_args_size
838 from HOST_WIDE_INT to poly_int64.
839 (add_cfi_args_size): Take the args_size as a poly_int64 rather
840 than a HOST_WIDE_INT.
841 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
842 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
843 polynomial arg sizes.
844 * emit-rtl.c (try_split): Use get_args_size.
845 * recog.c (peep2_attempt): Likewise.
846 * reload1.c (reload_as_needed): Likewise.
847 * expr.c (find_args_size_adjust): Return the adjustment as a
848 poly_int64 rather than a HOST_WIDE_INT.
849 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
850 to a poly_int64 and change the return type in the same way.
851 (emit_single_push_insn): Track polynomial arg sizes.
853 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
854 Alan Hayward <alan.hayward@arm.com>
855 David Sherwood <david.sherwood@arm.com>
857 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
858 from HOST_WIDE_INT to poly_int64.
859 * expr.c (push_block, emit_push_insn): Likewise.
861 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
862 Alan Hayward <alan.hayward@arm.com>
863 David Sherwood <david.sherwood@arm.com>
865 * function.h (frame_space): Change start and length from HOST_WIDE_INT
867 (get_frame_size): Return the size as a poly_int64 rather than a
869 (frame_offset_overflow): Take the offset as a poly_int64 rather
870 than a HOST_WIDE_INT.
871 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
872 (assign_stack_temp): Likewise for the size.
873 * function.c (get_frame_size): Return a poly_int64 rather than
875 (frame_offset_overflow): Take the offset as a poly_int64 rather
876 than a HOST_WIDE_INT.
877 (try_fit_stack_local): Take the start, length and size as poly_int64s
878 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
879 rather than a HOST_WIDE_INT.
880 (add_frame_space): Take the start and end as poly_int64s rather than
882 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
883 (assign_stack_temp): Likewise for the size.
884 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
886 (find_temp_slot_from_address): Handle polynomial offsets.
887 (combine_temp_slots): Likewise.
888 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
890 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
891 poly_int64 rather than a HOST_WIDE_INT.
892 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
893 than a HOST_WIDE_INT.
894 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
895 polynomial frame offsets.
896 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
897 as a poly_int64 rather than an int.
898 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
899 * config/v850/v850-protos.h (compute_frame_size): Likewise.
900 * config/v850/v850.c (compute_frame_size): Likewise.
901 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
902 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
903 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
904 * config/pa/pa.c (pa_compute_frame_size): Likewise.
905 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
906 rather than a HOST_WIDE_INT.
907 * explow.c (get_dynamic_stack_base): Likewise.
908 * final.c (final_start_function): Use the constant lower bound
909 of the frame size for -Wframe-larger-than.
910 * ira.c (do_reload): Adjust for new get_frame_size return type.
911 * lra.c (lra): Likewise.
912 * reload1.c (reload): Likewise.
913 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
914 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
915 * rtlanal.c (get_initial_register_offset): Return the offset as
916 a poly_int64 rather than a HOST_WIDE_INT.
918 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
919 Alan Hayward <alan.hayward@arm.com>
920 David Sherwood <david.sherwood@arm.com>
922 * reload1.c (elim_table): Change initial_offset, offset and
923 previous_offset from HOST_WIDE_INT to poly_int64_pod.
924 (offsets_at): Change the target array's element type from
925 HOST_WIDE_INT to poly_int64_pod.
926 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
927 (elimination_costs_in_insn, update_eliminable_offsets)
928 (verify_initial_elim_offsets, set_offsets_for_label)
929 (init_eliminable_invariants): Update after above changes.
931 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
932 Alan Hayward <alan.hayward@arm.com>
933 David Sherwood <david.sherwood@arm.com>
935 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
936 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
937 (decomposition): Change start and end from HOST_WIDE_INT
939 (decompose, immune_p): Update accordingly.
940 (find_inc_amount): Return a poly_int64 rather than an int.
941 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
944 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
945 Alan Hayward <alan.hayward@arm.com>
946 David Sherwood <david.sherwood@arm.com>
948 * tree.h (get_inner_reference): Return the bitsize and bitpos
949 as poly_int64_pods rather than HOST_WIDE_INT.
950 * fold-const.h (ptr_difference_const): Return the pointer difference
951 as a poly_int64_pod rather than a HOST_WIDE_INT.
952 * expr.c (get_inner_reference): Return the bitsize and bitpos
953 as poly_int64_pods rather than HOST_WIDE_INT.
954 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
956 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
957 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
958 (optimize_bit_field_compare): Update call to get_inner_reference.
959 (decode_field_reference): Likewise.
960 (fold_unary_loc): Track polynomial offsets and sizes.
961 (split_address_to_core_and_offset): Return the bitpos as a
962 poly_int64_pod rather than a HOST_WIDE_INT.
963 (ptr_difference_const): Likewise for the pointer difference.
964 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
965 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
966 * dbxout.c (dbxout_expand_expr): Likewise.
967 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
968 (loc_list_from_tree_1, fortran_common): Likewise.
969 * gimple-laddress.c (pass_laddress::execute): Likewise.
970 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
971 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
972 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
973 * tree-affine.c (tree_to_aff_combination): Likewise.
974 (get_inner_reference_aff): Likewise.
975 * tree-data-ref.c (split_constant_offset_1): Likewise.
976 (dr_analyze_innermost): Likewise.
977 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
978 * tree-sra.c (ipa_sra_check_caller): Likewise.
979 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
980 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
981 (instrument_bool_enum_load, instrument_object_size): Likewise.
982 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
983 to get_inner_reference.
984 * hsa-gen.c (gen_hsa_addr): Likewise.
985 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
986 * tsan.c (instrument_expr): Likewise.
987 * match.pd: Update call to ptr_difference_const.
989 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
990 Alan Hayward <alan.hayward@arm.com>
991 David Sherwood <david.sherwood@arm.com>
993 * fold-const.c (fold_comparison): Track sizes and offsets as
994 poly_int64s rather than HOST_WIDE_INTs when folding address
997 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
998 Alan Hayward <alan.hayward@arm.com>
999 David Sherwood <david.sherwood@arm.com>
1001 * expr.h (get_bit_range): Return the bitstart and bitend as
1002 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1003 as a poly_int64 rather than a HOST_WIDE_INT.
1004 * expr.c (get_bit_range): Likewise.
1005 (expand_assignment): Update call accordingly.
1006 * fold-const.c (optimize_bit_field_compare): Likewise.
1008 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1009 Alan Hayward <alan.hayward@arm.com>
1010 David Sherwood <david.sherwood@arm.com>
1012 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1013 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1014 as an optional final parameter.
1015 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1016 (aarch64_legitimate_address_p): Likewise.
1017 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1018 instead of an rtx code.
1019 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1020 (aarch64_legitimate_address_hook_p): Likewise.
1021 (aarch64_print_ldpstp_address): Likewise.
1022 (aarch64_print_operand_address): Likewise.
1023 (aarch64_address_cost): Likewise.
1024 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1025 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1026 (aarch64_mem_pair_lanes_operand): Likewise.
1028 2017-12-20 Richard Biener <rguenther@suse.de>
1030 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1031 update_stmt_if_modified.
1033 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1035 PR tree-optimization/83491
1036 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1037 before walking uses. Improve coding style and comments.
1039 2017-12-20 Tom de Vries <tom@codesourcery.com>
1041 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1043 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1046 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1048 (pass_ipa_free_fn_summary): Add small_p private data member,
1049 initialize to false in the ctor.
1050 (pass_ipa_free_fn_summary::clone,
1051 pass_ipa_free_fn_summary::set_pass_param,
1052 pass_ipa_free_fn_summary::gate): New methods.
1053 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1054 | TODO_dump_symtab if small_p.
1055 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1056 entry and add another instance of the pass with false parm after
1058 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1059 ipa_free_fn_summary here.
1061 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1063 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1065 2017-12-20 Martin Sebor <msebor@redhat.com>
1068 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1069 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1070 (expand_builtin_strncmp): Same.
1072 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1075 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1076 allowed debug stmts before labels.
1077 (expand_gimple_basic_block): Likewise.
1078 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1079 * gimple-iterator.h (gsi_after_labels): Likewise.
1080 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1081 rename reused variable, and simplify using gsi_move_before.
1082 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1083 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1084 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1085 (gimple_verify_flow_info, gimple_block_label): Likewise.
1086 (make_blocks): Move debug markers after adjacent labels.
1087 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1088 allowed debug insns outside blocks.
1089 * df-scan.c (df_insn_delete): Likewise.
1090 * lra-constraints.c (update_ebb_live_info): Likewise.
1091 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1092 (vt_initialize, delete_vta_debug_insns): Likewise.
1093 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1095 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1096 Alan Hayward <alan.hayward@arm.com>
1097 David Sherwood <david.sherwood@arm.com>
1099 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1100 (round_up_to_byte_boundary): Likewise.
1101 * expr.h (get_bit_range): Add temporary shim.
1102 * gimple-ssa-store-merging.c (store_operand_info): Change the
1103 bitsize, bitpos, bitregion_start and bitregion_end fields from
1104 unsigned HOST_WIDE_INT to poly_uint64.
1105 (merged_store_group): Likewise load_align_base.
1106 (compatible_load_p, compatible_load_p): Update accordingly.
1107 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1108 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1109 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1110 bitregion_start and bitregion_end as poly_uint64s rather than
1111 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1112 (handled_load): Take the bitsize, bitpos,
1113 bitregion_start and bitregion_end as poly_uint64s rather than
1114 unsigned HOST_WIDE_INTs.
1115 (pass_store_merging::process_store): Update call to
1116 mem_valid_for_store_merging.
1118 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1119 Alan Hayward <alan.hayward@arm.com>
1120 David Sherwood <david.sherwood@arm.com>
1122 * builtins.c (get_object_alignment_2): Track polynomial offsets
1123 and sizes. Update the alignment handling.
1125 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1126 Alan Hayward <alan.hayward@arm.com>
1127 David Sherwood <david.sherwood@arm.com>
1129 * tree.h (get_inner_reference): Add a version that returns the
1130 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1131 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1132 the case in which bitpos is not associated with the first byte.
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-affine.h (get_inner_reference_aff): Return the size as a
1140 * tree-affine.c (get_inner_reference_aff): Likewise.
1141 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1142 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1144 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1145 Alan Hayward <alan.hayward@arm.com>
1146 David Sherwood <david.sherwood@arm.com>
1148 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1149 HOST_WIDE_INT rather than a poly_int64.
1151 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1152 Alan Hayward <alan.hayward@arm.com>
1153 David Sherwood <david.sherwood@arm.com>
1155 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1156 HOST_WIDE_INT to poly_int64_pod.
1157 (perform_symbolic_merge): Update accordingly.
1158 (bswap_replace): Likewise.
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 * tree-affine.h (aff_tree::offset): Change from widest_int
1166 (wide_int_ext_for_comb): Delete.
1167 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1168 constants as poly_widest_int rather than widest_int.
1169 (aff_combination_constant_multiple_p): Return the multiplier
1170 as a poly_widest_int.
1171 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1173 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1174 version static and add an overload for poly_widest_int.
1175 (aff_combination_const, aff_combination_add_cst)
1176 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1177 the constants as poly_widest_int rather than widest_int.
1178 (tree_to_aff_combination): Generalize INTEGER_CST case to
1180 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1181 (aff_combination_add_product, aff_combination_mult): Handle
1183 (aff_combination_constant_multiple_p): Return the multiplier
1184 as a poly_widest_int.
1185 * tree-predcom.c (determine_offset): Return the offset as a
1187 (split_data_refs_to_components, suitable_component_p): Update
1189 (valid_initializer_p): Update call to
1190 aff_combination_constant_multiple_p.
1191 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1192 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1193 as a poly_int64 rather than a HOST_WIDE_INT.
1194 (get_address_cost): Handle polynomial offsets.
1195 (iv_elimination_compare_lt): Likewise.
1196 (rewrite_use_nonlinear_expr): Likewise.
1198 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1199 Alan Hayward <alan.hayward@arm.com>
1200 David Sherwood <david.sherwood@arm.com>
1202 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1203 as a poly_int64_pod rather than a HOST_WIDE_INT.
1204 (get_addr_base_and_unit_offset): Likewise.
1205 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1206 (get_addr_base_and_unit_offset): Likewise.
1207 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1208 to poly_int64 in example.
1209 * fold-const.c (fold_binary_loc): Update call to
1210 get_addr_base_and_unit_offset.
1211 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1212 (maybe_canonicalize_mem_ref_addr): Likewise.
1213 (gimple_fold_stmt_to_constant_1): Likewise.
1214 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1216 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1217 * match.pd: Likewise.
1218 * omp-low.c (lower_omp_target): Likewise.
1219 * tree-sra.c (build_ref_for_offset): Likewise.
1220 (build_debug_ref_for_model): Likewise.
1221 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1222 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1223 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1224 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1225 (constant_pointer_difference): Likewise.
1226 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1227 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1228 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1229 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1230 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1231 (set_ssa_val_to): Likewise.
1232 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1233 (maybe_diag_stxncpy_trunc): Likewise.
1234 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1235 * tree.c (build_simple_mem_ref_loc): Likewise.
1236 (array_at_struct_end_p): Likewise.
1238 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1239 Alan Hayward <alan.hayward@arm.com>
1240 David Sherwood <david.sherwood@arm.com>
1242 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1243 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1244 (get_ref_base_and_extent_hwi): Declare.
1245 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1246 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1247 (get_ref_base_and_extent_hwi): New function.
1248 * cfgexpand.c (expand_debug_expr): Update call to
1249 get_ref_base_and_extent.
1250 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1251 * gimple-fold.c (get_base_constructor): Return the offset as a
1252 poly_int64_pod rather than a HOST_WIDE_INT.
1253 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1254 * ipa-polymorphic-call.c
1255 (ipa_polymorphic_call_context::set_by_invariant)
1256 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1257 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1258 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1259 rather than get_ref_base_and_extent.
1260 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1261 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1262 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1264 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1265 call to get_ref_base_and_extent.
1266 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1267 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1268 (stmt_kills_ref_p): Likewise.
1269 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1270 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1272 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1273 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1274 when calling native_encode_expr.
1275 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1276 call to get_ref_base_and_extent.
1277 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1278 get_ref_base_and_extent.
1279 * var-tracking.c (track_expr_p): Likewise.
1281 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1282 Alan Hayward <alan.hayward@arm.com>
1283 David Sherwood <david.sherwood@arm.com>
1285 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1286 HOST_WIDE_INT to poly_int64_pod.
1287 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1288 polynomail parameter offsets.
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 * gengtype.c (main): Handle poly_int64_pod.
1295 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1296 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1297 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1298 from HOST_WIDE_INT to poly_int64_pod.
1299 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1300 (copy_cfa): New function.
1301 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1302 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1303 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1304 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1305 (initial_return_save): Treat offsets as poly_ints.
1306 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1307 offset is nonconstant.
1308 (reg_save): Take the offset as a poly_int64. Fall back to
1309 DW_CFA_expression for nonconstant offsets.
1310 (queue_reg_save): Take the offset as a poly_int64.
1311 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1313 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1314 Alan Hayward <alan.hayward@arm.com>
1315 David Sherwood <david.sherwood@arm.com>
1317 * rtl.h (operand_subword, operand_subword_force): Take the offset
1318 as a poly_uint64 an unsigned int.
1319 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1321 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1322 Alan Hayward <alan.hayward@arm.com>
1323 David Sherwood <david.sherwood@arm.com>
1325 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1326 'p' format code. Use INT_LIST rather than SUBREG as the example of
1327 a code with an XINT and an XEXP. Remove the implication that
1328 accessing an rtx field using XINT is expected to work.
1329 * rtl.def (SUBREG): Change format from "ei" to "ep".
1330 * rtl.h (rtunion::rt_subreg): New field.
1331 (XCSUBREG): New macro.
1332 (SUBREG_BYTE): Use it.
1333 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1334 Update constructor accordingly.
1335 (subreg_shape::operator ==): Update accordingly.
1336 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1337 than an unsigned int.
1338 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1339 a poly_uint64 rather than an unsigned int.
1340 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1341 than an unsigned int.
1342 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1343 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1344 an unsigned int. Take the sizes as poly_uint64s.
1345 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1346 an unsigned int. Take the shift as a poly_uint64 rather than
1348 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1349 as a poly_uint64 rather than an unsigned int.
1350 (simplify_subreg_regno): Likewise.
1351 (byte_lowpart_offset): Return the memory offset as a poly_int64
1353 (subreg_memory_offset): Likewise. Take the subreg offset as a
1354 poly_uint64 rather than an unsigned int.
1355 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1356 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1357 poly_uint64 rather than an unsigned int.
1358 * rtl.c (rtx_format): Describe 'p' in comment.
1359 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1360 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1361 offset as a poly_uint64 rather than an unsigned int.
1362 (byte_lowpart_offset): Return the memory offset as a poly_int64
1364 (subreg_memory_offset): Likewise. Take the subreg offset as a
1365 poly_uint64 rather than an unsigned int.
1366 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1367 mode sizes as poly_uint64s rather than unsigned ints. Return a
1368 poly_uint64 rather than an unsigned int.
1369 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1370 (copy_insn_1): Handle 'p'.
1371 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1372 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1373 an unsigned int. Return the shift in the same way.
1374 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1376 (subreg_size_offset_from_lsb): Take the sizes and shift as
1377 poly_uint64s rather than unsigned ints. Return the offset as
1379 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1380 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1382 * rtlhash.c (add_rtx): Handle 'p'.
1383 * genemit.c (gen_exp): Likewise.
1384 * gengenrtl.c (type_from_format, gendef): Likewise.
1385 * gensupport.c (subst_pattern_match, get_alternatives_number)
1386 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1387 (subst_dup): Likewise.
1388 * gengtype.c (adjust_field_rtx_def): Likewise.
1389 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1390 (match_pattern_2): Likewise.
1391 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1392 (rtx_test::subreg_field): New function.
1393 (operator ==, safe_to_hoist_p, transition_parameter_type)
1394 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1395 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1396 * genpeep.c (match_rtx): Likewise.
1397 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1398 (rtx_writer::print_rtx_operand): Handle 'p'.
1399 (print_value): Handle SUBREG.
1400 * read-rtl.c (apply_int_iterator): Likewise.
1401 (rtx_reader::read_rtx_operand): Handle 'p'.
1402 * alias.c (rtx_equal_for_memref_p): Likewise.
1403 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1404 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1406 * calls.c (expand_call): Likewise.
1407 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1408 (make_extraction, gen_lowpart_for_combine): Likewise.
1409 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1411 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1412 rather than an unsigned int. Treat subreg offsets as poly_ints.
1413 (exp_equiv_p): Handle 'p'.
1414 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1415 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1416 * dse.c (find_shift_sequence): Likewise.
1417 * dwarf2out.c (rtl_for_decl_location): Likewise.
1418 * expmed.c (extract_low_bits): Likewise.
1419 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1420 (expand_expr_real_2): Likewise.
1421 * final.c (alter_subreg): Likewise.
1422 (leaf_renumber_regs_insn): Handle 'p'.
1423 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1424 Treat subreg offsets as poly_ints.
1425 * fwprop.c (forward_propagate_and_simplify): Likewise.
1426 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1427 * ira.c (get_subreg_tracking_sizes): Likewise.
1428 * ira-conflicts.c (go_through_subreg): Likewise.
1429 * ira-lives.c (process_single_reg_class_operands): Likewise.
1430 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1431 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1432 as a poly_uint64 rather than an unsigned int.
1433 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1434 subreg offsets as poly_ints.
1435 * lra-constraints.c (operands_match_p): Handle 'p'.
1436 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1437 * lra-spills.c (assign_mem_slot): Likewise.
1438 * postreload.c (move2add_valid_value_p): Likewise.
1439 * recog.c (general_operand, indirect_operand): Likewise.
1440 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1441 (copyprop_hardreg_forward_1): Likewise.
1442 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1443 (record_subregs_of_mode): Likewise.
1444 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1445 * reload.c (operands_match_p): Handle 'p'.
1446 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1447 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1448 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1449 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1450 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1451 (simplify_const_poly_int_tests<N>::run): Likewise.
1452 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1453 a poly_uint64 rather than an unsigned int.
1454 * valtrack.c (debug_lowpart_subreg): Likewise.
1455 * var-tracking.c (var_lowpart): Likewise.
1456 (loc_cmp): Handle 'p'.
1458 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1459 Alan Hayward <alan.hayward@arm.com>
1460 David Sherwood <david.sherwood@arm.com>
1462 * ira.c (get_subreg_tracking_sizes): New function.
1463 (init_live_subregs): Take an integer size rather than a register.
1464 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1465 to init_live_subregs.
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 * expr.c (store_constructor_field): Change bitsize from a
1472 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1473 HOST_WIDE_INT to a poly_int64.
1474 (store_constructor): Change size from a HOST_WIDE_INT to
1476 (store_field): Likewise bitsize and bitpos.
1478 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1479 Alan Hayward <alan.hayward@arm.com>
1480 David Sherwood <david.sherwood@arm.com>
1482 * expmed.h (store_bit_field): Change bitregion_start and
1483 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1484 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1485 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1486 (store_fixed_bit_field, store_split_bit_field): Likewise.
1487 * expr.c (store_constructor_field, store_field): Likewise.
1488 (optimize_bitfield_assignment_op): Likewise. Make the same change
1489 to bitsize and bitpos.
1490 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1491 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1492 same change in the constructor arguments.
1493 (get_best_mode): Change bitregion_start and bitregion_end from
1494 unsigned HOST_WIDE_INT to poly_uint64.
1495 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1496 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1498 (bit_field_mode_iterator::next_mode): Update for new types
1499 of m_bitregion_start and m_bitregion_end.
1500 (get_best_mode): Change bitregion_start and bitregion_end from
1501 unsigned HOST_WIDE_INT to poly_uint64.
1503 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1504 Alan Hayward <alan.hayward@arm.com>
1505 David Sherwood <david.sherwood@arm.com>
1507 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1508 accepts poly_uint64 offsets.
1509 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1510 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1511 * expmed.c (lowpart_bit_field_p): Likewise.
1512 (extract_bit_field_as_subreg): New function, split out from...
1513 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1514 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1515 extractions, check that BITSIZE matches the size of the extracted
1516 value and that BITNUM is an exact multiple of that size.
1517 If all else fails, try forcing the value into memory if
1518 BITNUM is variable, and adjusting the address so that the
1519 offset is constant. Split the part that can only handle constant
1520 bitsize and bitnum out into...
1521 (extract_integral_bit_field): ...this new function.
1522 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1523 rather than unsigned HOST_WIDE_INTs.
1525 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1526 Alan Hayward <alan.hayward@arm.com>
1527 David Sherwood <david.sherwood@arm.com>
1529 * expmed.h (store_bit_field): Take bitsize and bitnum as
1530 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1531 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1532 that returns the byte size.
1533 (store_bit_field_1): Take bitsize and bitnum as
1534 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1535 to simple_mem_bitfield_p. Split the part that can only handle
1536 constant bitsize and bitnum out into...
1537 (store_integral_bit_field): ...this new function.
1538 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1539 than unsigned HOST_WIDE_INTs.
1540 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1542 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1543 Alan Hayward <alan.hayward@arm.com>
1544 David Sherwood <david.sherwood@arm.com>
1546 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1547 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1549 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1550 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1551 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1552 offset as a poly_int64 rather than an int.
1553 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1554 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1555 * lra-constraints.c (equiv_address_substitution): Track offsets
1557 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1558 (curr_insn_transform): Handle the new form of sp_offset.
1559 * lra-eliminations.c (lra_elim_table): Change previous_offset
1560 and offset from HOST_WIDE_INT to poly_int64.
1561 (print_elim_table, update_reg_eliminate): Update accordingly.
1562 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1563 (get_elimination): Update accordingly.
1564 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1565 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1566 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1567 poly_int64 offsets generally.
1568 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1569 (mark_not_eliminable, init_elimination): Update accordingly.
1570 (remove_reg_equal_offset_note): Return a bool and pass the new
1571 offset back by pointer as a poly_int64.
1572 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1573 rather than a HOST_WIDE_INT.
1574 (do_remat): Track offsets poly_int64s.
1575 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1577 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1578 Alan Hayward <alan.hayward@arm.com>
1579 David Sherwood <david.sherwood@arm.com>
1581 * rtl.h (mem_attrs): Add a default constructor. Change size and
1582 offset from HOST_WIDE_INT to poly_int64.
1583 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1584 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1585 (widen_memory_access): Take the sizes and offsets as poly_int64s
1586 rather than HOST_WIDE_INTs.
1587 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1588 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1590 (adjust_offset_for_component_ref): Change the offset from a
1591 HOST_WIDE_INT to a poly_int64.
1592 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1593 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1594 * dce.c (find_call_stack_args): Likewise.
1595 * dse.c (record_store): Likewise.
1596 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1597 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1598 * read-rtl-function.c (test_loading_mem): Likewise.
1599 * rtlanal.c (may_trap_p_1): Likewise.
1600 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1601 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1602 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1603 (mem_attrs::mem_attrs): New function.
1604 (set_mem_attributes_minus_bitpos): Change bitpos from a
1605 HOST_WIDE_INT to poly_int64.
1606 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1607 (clear_mem_offset, clear_mem_size, change_address)
1608 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1609 initialize mem_attrs.
1610 (set_mem_offset, set_mem_size, adjust_address_1)
1611 (adjust_automodify_address_1, offset_address, widen_memory_access):
1612 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1614 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1615 Alan Hayward <alan.hayward@arm.com>
1616 David Sherwood <david.sherwood@arm.com>
1618 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1619 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1620 rather than 0 to represent an unknown size. Assert that the size
1621 is known when the mode isn't BLKmode.
1622 (may_trap_p_1): Use -1 for unknown sizes.
1623 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1625 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1626 Alan Hayward <alan.hayward@arm.com>
1627 David Sherwood <david.sherwood@arm.com>
1629 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1630 to poly_int64. Update commentary for positions_needed.large.
1631 (read_info_type): Change offset and width from HOST_WIDE_INT
1633 (set_usage_bits): Likewise.
1634 (canon_address): Return the offset as a poly_int64 rather than
1635 a HOST_WIDE_INT. Use strip_offset_and_add.
1636 (set_all_positions_unneeded, any_positions_needed_p): Use
1637 positions_needed.large to track stores with non-constant widths.
1638 (all_positions_needed_p): Likewise. Take the offset and width
1639 as poly_int64s rather than ints. Assert that rhs is nonnull.
1640 (record_store): Cope with non-constant offsets and widths.
1641 Nullify the rhs of an earlier store if we can't tell which bytes
1643 (find_shift_sequence): Take the access_size and shift as poly_int64s
1645 (get_stored_val): Take the read_offset and read_width as poly_int64s
1646 rather than HOST_WIDE_INTs.
1647 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1648 non-constant offsets and widths.
1650 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1651 Alan Hayward <alan.hayward@arm.com>
1652 David Sherwood <david.sherwood@arm.com>
1654 * inchash.h (inchash::hash::add_poly_int): New function.
1655 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1656 Use poly_int64 rather than HOST_WIDE_INT.
1657 (ao_ref::max_size_known_p): New function.
1658 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1659 rather than HOST_WIDE_INT.
1660 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1661 to temporaries until its interface is adjusted to match.
1662 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1663 (aliasing_component_refs_p, decl_refs_may_alias_p)
1664 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1665 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1666 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1668 * alias.c (ao_ref_from_mem): Likewise.
1669 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1670 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1671 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1672 (maybe_trim_complex_store, maybe_trim_constructor_store)
1673 (live_bytes_read, dse_classify_store): Likewise.
1674 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1675 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1676 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1677 (vn_reference_lookup_3): Likewise.
1678 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1680 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1681 Alan Hayward <alan.hayward@arm.com>
1682 David Sherwood <david.sherwood@arm.com>
1684 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1685 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1686 instead of ranges_overlap_p.
1688 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1689 Alan Hayward <alan.hayward@arm.com>
1690 David Sherwood <david.sherwood@arm.com>
1692 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1693 sizes as poly_int64s rather than HOST_WIDE_INTs.
1695 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1696 Alan Hayward <alan.hayward@arm.com>
1697 David Sherwood <david.sherwood@arm.com>
1699 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1700 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1701 * gimple-fold.c (fold_ctor_reference): Likewise.
1703 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1704 Alan Hayward <alan.hayward@arm.com>
1705 David Sherwood <david.sherwood@arm.com>
1707 * target.def (dwarf_poly_indeterminate_value): New hook.
1708 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1709 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1710 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1711 * doc/tm.texi: Regenerate.
1712 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1713 offset as a poly_int64.
1714 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1715 offset as a poly_int64.
1716 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1717 Take the offset as a poly_int64.
1718 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1719 (frame_pointer_fb_offset): Change to a poly_int64.
1720 (int_loc_descriptor): Take the offset as a poly_int64. Use
1721 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1722 (based_loc_descr): Take the offset as a poly_int64.
1723 Use strip_offset_and_add to handle (plus X (const)).
1724 Use new_reg_loc_descr instead of an open-coded version of the
1725 previous implementation.
1726 (mem_loc_descriptor): Handle CONST_POLY_INT.
1727 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1728 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1730 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1731 Alan Hayward <alan.hayward@arm.com>
1732 David Sherwood <david.sherwood@arm.com>
1734 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1735 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1736 * inchash.h (inchash::hash::add_poly_hwi): New function.
1737 * gengtype.c (main): Register poly_int64.
1738 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1739 offset as a poly_int.
1740 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1741 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1742 offset as a poly_int64.
1743 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1744 * print-rtl.c (print_poly_int): New function.
1745 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1747 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1749 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1750 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1751 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1752 Enforce track_offset_p for parts of a PARALLEL.
1753 (vt_add_function_parameter): Use const_offset for the final
1754 offset to track. Use get_tracked_reg_offset for the parts
1757 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1758 Alan Hayward <alan.hayward@arm.com>
1759 David Sherwood <david.sherwood@arm.com>
1761 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1762 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1763 * doc/tm.texi: Regenerate.
1764 * hooks.h (hook_bool_uint_uint_true): Delete.
1765 (hook_bool_puint64_puint64_true): Declare.
1766 * hooks.c (hook_bool_uint_uint_true): Delete.
1767 (hook_bool_puint64_puint64_true): New function.
1768 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1769 instead of unsigned ints.
1770 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1771 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1773 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1774 Alan Hayward <alan.hayward@arm.com>
1775 David Sherwood <david.sherwood@arm.com>
1777 * optabs.h (expand_operand): Add an int_value field.
1778 (create_expand_operand): Add an int_value parameter and use it
1779 to initialize the new expand_operand field.
1780 (create_integer_operand): Replace with a declaration of a function
1781 that accepts poly_int64s. Move the implementation to...
1782 * optabs.c (create_integer_operand): ...here.
1783 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1784 the mode preserves the value of int_value, instead of calling
1785 const_int_operand on the rtx. Use gen_int_mode to generate
1788 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1789 Alan Hayward <alan.hayward@arm.com>
1790 David Sherwood <david.sherwood@arm.com>
1792 * dumpfile.h (dump_dec): Declare.
1793 * dumpfile.c (dump_dec): New function.
1794 * pretty-print.h (pp_wide_integer): Turn into a function and
1795 declare a poly_int version.
1796 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1798 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1799 Alan Hayward <alan.hayward@arm.com>
1800 David Sherwood <david.sherwood@arm.com>
1802 * doc/generic.texi (POLY_INT_CST): Document.
1803 * tree.def (POLY_INT_CST): New tree code.
1804 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1805 * tree-core.h (tree_poly_int_cst): New struct.
1806 (tree_node): Add a poly_int_cst field.
1807 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1808 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1809 instead of a wide_int_ref.
1810 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1812 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1813 instead of an unsigned HOST_WIDE_INT.
1814 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1815 (ptrdiff_tree_p): Declare.
1816 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1817 extern inline implementations if the target doesn't use POLY_INT_CST.
1818 (poly_int_tree_p): New function.
1819 (wi::unextended_tree): New class.
1820 (wi::int_traits <unextended_tree>): New override.
1821 (wi::extended_tree): Add a default constructor.
1822 (wi::extended_tree::get_tree): New function.
1823 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1824 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1825 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1826 (wi::tree_to_poly_wide_ref): New typedefs.
1827 (wi::ints_for): Provide overloads for extended_tree and
1829 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1830 (wi::to_wide): New functions.
1831 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1832 * tree.c (poly_int_cst_hasher): New struct.
1833 (poly_int_cst_hash_table): New variable.
1834 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1835 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1837 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1838 (init_ttree): Initialize poly_int_cst_hash_table.
1839 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1840 a poly_int64 instead of a HOST_WIDE_INT.
1841 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1842 instead of an unsigned HOST_WIDE_INT.
1843 (wide_int_to_tree): Rename to...
1844 (wide_int_to_tree_1): ...this.
1845 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1846 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1847 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1848 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1849 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1850 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1852 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1853 (streamer_read_tree_body): Likewise.
1854 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1855 (streamer_write_tree_body): Likewise.
1856 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1857 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1858 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1859 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1860 * gimple-expr.h (is_gimple_constant): Likewise.
1861 * gimplify.c (maybe_with_size_expr): Likewise.
1862 * print-tree.c (print_node): Likewise.
1863 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1864 * tree-pretty-print.c (dump_generic_node): Likewise.
1865 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1866 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1867 * tree-vrp.c (compare_values_warnv): Likewise.
1868 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1869 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1870 (force_expr_to_var_cost): Likewise.
1871 * tree-ssa-loop.c (for_each_index): Likewise.
1872 * fold-const.h (build_invariant_address, size_int_kind): Take a
1873 poly_int64 instead of a HOST_WIDE_INT.
1874 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1875 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1877 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1878 (int_const_binop_2): New function, split out from...
1879 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1880 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1881 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1882 * gimple-ssa-strength-reduction.c (slsr_process_add)
1883 (slsr_process_mul): Check for INTEGER_CSTs before using them
1885 * stor-layout.c (bits_from_bytes): New function.
1886 (bit_from_pos): Use it.
1887 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1888 by BITS_PER_UNIT to get the TYPE_SIZE.
1889 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1890 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1892 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1893 Alan Hayward <alan.hayward@arm.com>
1894 David Sherwood <david.sherwood@arm.com>
1896 * doc/rtl.texi (const_poly_int): Document. Also document the
1897 rtl sharing behavior.
1898 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1899 * rtl.h (const_poly_int_def): New struct.
1900 (rtx_def::u): Add a cpi field.
1901 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1902 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1903 (wi::rtx_to_poly_wide_ref): New typedef
1904 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1905 (poly_int_rtx_p): New functions.
1906 (trunc_int_for_mode): Declare a poly_int64 version.
1907 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1908 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1910 (strip_offset): Declare.
1911 (strip_offset_and_add): New function.
1912 * rtl.def (CONST_POLY_INT): New rtx code.
1913 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1914 (shared_const_p): Use poly_int_rtx_p.
1915 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1917 (gen_int_shift_amount): Likewise.
1918 * emit-rtl.c (const_poly_int_hasher): New class.
1919 (const_poly_int_htab): New variable.
1920 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1921 (const_poly_int_hasher::hash): New function.
1922 (const_poly_int_hasher::equal): Likewise.
1923 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1924 (immed_wide_int_const): Rename to...
1925 (immed_wide_int_const_1): ...this and make static.
1926 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1927 instead of a wide_int_ref.
1928 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1929 (gen_lowpart_common): Handle CONST_POLY_INT.
1930 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1931 * cselib.c (cselib_hash_rtx): Likewise.
1932 * dwarf2out.c (const_ok_for_output_1): Likewise.
1933 * expr.c (convert_modes): Likewise.
1934 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1935 * rtlhash.c (add_rtx): Likewise.
1936 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1937 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1938 Handle existing CONST_POLY_INT rtxes.
1939 * expmed.h (expand_shift): Take a poly_int64 instead of a
1941 * expmed.c (expand_shift): Likewise.
1942 * rtlanal.c (strip_offset): New function.
1943 (commutative_operand_precedence): Give CONST_POLY_INT the same
1944 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1946 * rtl-tests.c (const_poly_int_tests): New struct.
1947 (rtl_tests_c_tests): Use it.
1948 * simplify-rtx.c (simplify_const_unary_operation): Handle
1950 (simplify_const_binary_operation): Likewise.
1951 (simplify_binary_operation_1): Fold additions of symbolic constants
1952 and CONST_POLY_INTs.
1953 (simplify_subreg): Handle extensions and truncations of
1955 (simplify_const_poly_int_tests): New struct.
1956 (simplify_rtx_c_tests): Use it.
1957 * wide-int.h (storage_ref): Add default constructor.
1958 (wide_int_ref_storage): Likewise.
1959 (trailing_wide_ints): Use GTY((user)).
1960 (trailing_wide_ints::operator[]): Add a const version.
1961 (trailing_wide_ints::get_precision): New function.
1962 (trailing_wide_ints::extra_size): Likewise.
1964 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1965 Alan Hayward <alan.hayward@arm.com>
1966 David Sherwood <david.sherwood@arm.com>
1968 * emit-rtl.h (gen_int_shift_amount): Declare.
1969 * emit-rtl.c (gen_int_shift_amount): New function.
1970 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1972 * calls.c (shift_return_value): Likewise.
1973 * cse.c (fold_rtx): Likewise.
1974 * dse.c (find_shift_sequence): Likewise.
1975 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1976 (expand_shift, expand_smod_pow2): Likewise.
1977 * lower-subreg.c (shift_cost): Likewise.
1978 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1979 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1980 (expand_vec_perm_var): Likewise.
1981 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1982 (simplify_binary_operation_1): Likewise.
1983 * combine.c (try_combine, find_split_point, force_int_to_mode)
1984 (simplify_shift_const_1, simplify_shift_const): Likewise.
1985 (change_zero_ext): Likewise. Use simplify_gen_binary.
1987 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1989 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1991 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1993 * doc/invoke.texi (ARM Options): Document accepted extension options
1994 for -march=armv8.3-a.
1996 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1999 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2000 or --with-float=softfp, set the default CPU to arm10e.
2002 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2004 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2005 * config/visium/predicates.md (const_shift_operand): Likewise.
2006 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2007 (visium_legitimize_reload_address): Likewise.
2009 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2012 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2014 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2017 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2019 2017-12-20 Julia Koval <julia.koval@intel.com>
2021 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2022 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2023 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2024 * config.gcc: Include vpclmulqdqintrin.h.
2025 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2026 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2028 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2029 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2030 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2031 * config/i386/i386.c (isa2_opts): Add -mcx16.
2032 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2033 (ix86_option_override_internal): Move mcx16 to flags2.
2034 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2035 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2036 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2037 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2038 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2039 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2040 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2041 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2042 * doc/invoke.texi: Add -mvpclmulqdq.
2044 2017-12-20 Tom de Vries <tom@codesourcery.com>
2047 * config/i386/i386.c (ix86_static_chain): Move
2048 DECL_STATIC_CHAIN test ...
2049 * calls.c (rtx_for_static_chain): ... here. New function.
2050 * calls.h (rtx_for_static_chain): Declare.
2051 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2052 instead of targetm.calls.static_chain.
2053 * df-scan.c (df_get_entry_block_def_set): Same.
2055 2017-12-19 Tom de Vries <tom@codesourcery.com>
2057 PR tree-optimization/83493
2058 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2061 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2063 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2064 inform with hardcoded english plural handling.
2066 2017-12-18 Jeff Law <law@redhat.com>
2068 PR tree-optimization/83477
2069 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2070 a non-virtual PHI, always push a new range.
2072 2017-12-19 Martin Sebor <msebor@redhat.com>
2075 * builtins.c (compute_objsize): Handle non-constant offsets.
2077 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2079 PR tree-optimization/83444
2080 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2081 character load case, if get_stridx on MEM_REF's operand doesn't
2082 look usable, retry with get_addr_stridx.
2084 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2087 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2088 (variable_tracking_main_1): Keep markers even when VTA fails.
2091 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2092 even if there are markers before it.
2093 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2095 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2097 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2098 typical order conditions.
2099 * sel-sched.c (extract_new_fences_from): Likewise.
2100 * config/visium/constraints.md (J, K, L): Likewise.
2101 * config/visium/predicates.md (const_shift_operand): Likewise.
2102 * config/visium/visium.c (visium_legitimize_address,
2103 visium_legitimize_reload_address): Likewise.
2104 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2105 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2106 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2107 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2108 SET_DUMP_DETAIL): Likewise.
2109 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2110 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2111 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2112 avr_set_core_architecture, avr_set_current_function,
2113 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2114 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2115 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2116 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2117 avr_map_decompose, avr_fold_builtin): Likewise.
2118 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2119 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2120 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2121 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2122 * config/m32c/m32c.c (m32c_conditional_register_usage,
2123 m32c_address_cost): Likewise.
2124 * config/m32c/predicates.md (shiftcount_operand,
2125 longshiftcount_operand): Likewise.
2126 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2127 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2128 can_use_cdx_ldstw): Likewise.
2129 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2130 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2132 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2133 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2134 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2135 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2136 * config/vax/vax.c (adjacent_operands_p): Likewise.
2137 * config/ft32/constraints.md (L, b, KA): Likewise.
2138 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2140 * cfgexpand.c (expand_stack_alignment): Likewise.
2141 * gcse.c (insert_expr_in_table): Likewise.
2142 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2143 * cgraphunit.c (cgraph_node::expand): Likewise.
2144 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2145 * emit-rtl.c (add_insn): Likewise.
2146 * input.c (dump_location_info): Likewise.
2147 * passes.c (NEXT_PASS): Likewise.
2148 * read-rtl-function.c (parse_note_insn_name,
2149 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2151 * sched-rgn.c (sched_rgn_init): Likewise.
2152 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2153 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2154 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2156 * explow.c (eliminate_constant_term): Likewise.
2157 * final.c (leaf_renumber_regs_insn): Likewise.
2158 * cfgrtl.c (print_rtl_with_bb): Likewise.
2159 * genhooks.c (emit_init_macros): Likewise.
2160 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2161 * tree-data-ref.c (conflict_fn): Likewise.
2162 * selftest.c (assert_streq): Likewise.
2163 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2164 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2165 fold_binary_loc, multiple_of_p): Likewise.
2166 * reload.c (push_reload, find_equiv_reg): Likewise.
2167 * et-forest.c (et_nca, et_below): Likewise.
2168 * dbxout.c (dbxout_symbol_location): Likewise.
2169 * reorg.c (relax_delay_slots): Likewise.
2170 * dojump.c (do_compare_rtx_and_jump): Likewise.
2171 * gengtype-parse.c (type): Likewise.
2172 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2173 simplify_const_relational_operation): Likewise.
2174 * reload1.c (do_output_reload): Likewise.
2175 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2176 * gengtype.c (type_for_name): Likewise.
2177 * gimple-ssa-sprintf.c (format_directive): Likewise.
2179 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2182 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2183 accessing it. Adjust comment.
2185 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2188 * predict.c (zero_one_minusone): New function.
2189 (apply_return_prediction): Avoid return prediction for functions
2190 returning only -1, 0 and 1 values, unless they only return -1 and 0
2193 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2195 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2196 fall-through warning.
2197 (arc_compute_frame_size): Remove unused variables.
2198 (arc_print_operand): Fix fprintif format.
2199 (arc_can_follow_jump): Clean fall-through warning.
2201 2017-12-19 Marek Polacek <polacek@redhat.com>
2204 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2207 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2209 * config/arc/arc.c (overriderregs): New variable.
2210 (arc_override_options): Track fixed/call saved/call options.
2211 (arc_conditional_register_usage): Check against overrideregs
2212 variable whenever we change register properties.
2214 2017-12-19 Nathan Sidwell <nathan@acm.org>
2216 * opts.c (finish_options): Don't prefix dump_base_name if it
2217 already contains directories.
2219 2017-12-19 Martin Liska <mliska@suse.cz>
2221 PR rtl-optimization/82675
2222 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2223 more element in sbitmap.
2225 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2227 * gimplify.c (gimplify_expr): Use error_operand_p.
2229 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2232 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2235 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2237 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2238 the jump is to a label.
2240 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2242 PR tree-optimization/83444
2243 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2248 * ipa-inline.c (flatten_remove_node_hook): New function.
2249 (ipa_inline): Keep only nodes with flatten attribute at the end of
2250 the array in the order from ipa_reverse_postorder, only walk that
2251 portion of array for flattening, if there is more than one such
2252 node, temporarily register a removal hook and ignore removed nodes.
2254 PR tree-optimization/80631
2255 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2256 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2257 IFN_REDUC_MAX or IFN_REDUC_MIN.
2259 2017-12-18 Jeff Law <law@redhat.com>
2261 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2262 record symbolic equivalences from backedges in the CFG.
2265 2017-11-19 Jeff Law <law@redhat.com>
2267 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2268 of degenerates resulting from ignoring an edge.
2270 2017-12-18 Martin Sebor <msebor@redhat.com>
2273 PR tree-optimization/78450
2274 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2275 (handle_builtin_strlen): Call it.
2277 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2279 PR rtl-optimization/83424
2280 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2282 2017-12-18 Marek Polacek <polacek@redhat.com>
2285 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2286 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2289 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2291 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2292 (umaddsidi4, umaddsidi_split): Likewise.
2294 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2296 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2299 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2302 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2304 2017-12-18 Richard Biener <rguenther@suse.de>
2306 PR tree-optimization/81877
2307 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2308 (outermost_indep_loop): Adjust.
2309 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2310 (can_sm_ref_p): Adjust.
2312 2017-12-18 Richard Biener <rguenther@suse.de>
2315 * tree.c (array_at_struct_end_p): Return true if the underlying
2316 object has space for at least one element in excess of what
2317 the array domain specifies.
2319 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2321 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2322 example to avoid overfull hbox.
2323 * doc/invoke.texi (Option Summary): Add missing @gol.
2324 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2327 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2328 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2330 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2331 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2332 (AArch64 Options): Document -mverbose-cost-dump.
2333 (ARM Options): Likewise, plus -mflip-thumb.
2335 2017-12-17 Martin Sebor <msebor@redhat.com>
2338 * gimple-ssa-warn-restrict.c
2339 (builtin_memref::offset_out_of_bounds): Correct the handling of
2342 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2344 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2346 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2348 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2350 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2352 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2355 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2357 * doc/invoke.texi: Fix some typos.
2359 2017-12-16 Martin Sebor <msebor@redhat.com>
2361 PR tree-optimization/78918
2362 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2363 * builtins.c (check_sizes): Rename...
2364 (check_access): ...to this. Rename function arguments for clarity.
2365 (check_memop_sizes): Adjust names.
2366 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2367 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2368 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2369 (check_strncat_sizes, expand_builtin_strncat): Same.
2370 (expand_builtin_strncpy, expand_builtin_memset): Same.
2371 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2372 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2373 (maybe_emit_sprintf_chk_warning): Same.
2374 (expand_builtin_strcpy): Adjust.
2375 (expand_builtin_stpcpy): Same.
2376 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2377 in pointer-checking forms of memcpy, memmove, and mempcpy.
2378 (gcall_to_tree_minimal, max_object_size): Define new functions.
2379 * builtins.h (max_object_size): Declare.
2380 * calls.c (alloc_max_size): Call max_object_size instead of
2381 hardcoding ssizetype limit.
2382 (get_size_range): Handle new argument.
2383 * calls.h (get_size_range): Add a new argument.
2384 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2385 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2386 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2388 (gimple_fold_builtin_memory_chk): Same.
2389 (gimple_fold_builtin_stxcpy_chk): New function.
2390 * gimple-ssa-warn-restrict.c: New source.
2391 * gimple-ssa-warn-restrict.h: New header.
2392 * gimple.c (gimple_build_call_from_tree): Propagate location.
2393 * passes.def (pass_warn_restrict): Add new pass.
2394 * tree-pass.h (make_pass_warn_restrict): Declare.
2395 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2397 (handle_builtin_strcat): Same.
2398 (strlen_optimize_stmt): Rename...
2399 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2400 stpncpy, strncpy, and their checking forms.
2402 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2404 PR rtl-optimization/82849
2405 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2406 and get_max_loop_iterations_int.
2408 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2409 Alan Hayward <alan.hayward@arm.com>
2410 David Sherwood <david.sherwood@arm.com>
2412 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2413 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2414 size as a poly_uint64.
2415 (mode_for_vector, mode_for_int_vector): Take the number of vector
2416 elements as a poly_uint64.
2417 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2418 size as a poly_uint64.
2419 (mode_for_vector, mode_for_int_vector): Take the number of vector
2420 elements as a poly_uint64.
2422 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2423 Alan Hayward <alan.hayward@arm.com>
2424 David Sherwood <david.sherwood@arm.com>
2426 * machmode.h (MACRO_MODE): New macro.
2427 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2428 * targhooks.c (default_libcall_value, default_secondary_reload)
2429 (default_memory_move_cost, default_register_move_cost)
2430 (default_class_max_nregs): Likewise.
2432 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2433 Alan Hayward <alan.hayward@arm.com>
2434 David Sherwood <david.sherwood@arm.com>
2436 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2437 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2439 * genautomata.c (main): Likewise.
2440 * genconditions.c (write_header): Likewise.
2441 * genemit.c (main): Likewise.
2442 * genextract.c (print_header): Likewise.
2443 * genopinit.c (main): Likewise.
2444 * genoutput.c (output_prologue): Likewise.
2445 * genpeep.c (main): Likewise.
2446 * genpreds.c (write_insn_preds_c): Likewise.
2447 * genrecog.c (writer_header): Likewise.
2448 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2449 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2450 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2451 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2452 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2453 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2454 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2455 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2456 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2457 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2458 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2459 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2460 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2461 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2462 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2463 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2464 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2465 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2466 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2467 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2468 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2469 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2470 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2471 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2472 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2473 * config/darwin.c (IN_TARGET_CODE): Likewise.
2474 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2475 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2476 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2477 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2478 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2479 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2480 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2481 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2482 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2483 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2484 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2485 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2486 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2487 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2488 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2489 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2490 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2491 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2492 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2493 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2494 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2495 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2496 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2497 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2498 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2499 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2500 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2501 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2502 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2503 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2504 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2505 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2506 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2507 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2508 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2509 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2510 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2511 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2512 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2513 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2514 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2515 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2516 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2517 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2518 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2519 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2520 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2521 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2522 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2523 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2524 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2525 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2526 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2527 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2528 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2529 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2530 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2531 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2532 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2533 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2534 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2535 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2536 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2537 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2538 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2539 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2540 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2541 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2542 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2543 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2544 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2545 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2546 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2547 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2548 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2549 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2550 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2551 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2552 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2553 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2554 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2555 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2556 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2557 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2558 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2559 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2560 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2561 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2562 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2563 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2564 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2565 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2566 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2567 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2568 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2569 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2570 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2571 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2572 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2573 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2574 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2575 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2576 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2577 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2578 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2579 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2581 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2582 Alan Hayward <alan.hayward@arm.com>
2583 David Sherwood <david.sherwood@arm.com>
2585 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2588 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2589 Alan Hayward <alan.hayward@arm.com>
2590 David Sherwood <david.sherwood@arm.com>
2592 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2593 * doc/md.texi (vec_series@var{m}): Document.
2594 * tree.def (VEC_SERIES_EXPR): New tree code.
2595 * tree.h (build_vec_series): Declare.
2596 * tree.c (build_vec_series): New function.
2597 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2598 * tree-pretty-print.c (dump_generic_node): Likewise.
2599 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2600 * tree-inline.c (estimate_operator_cost): Likewise.
2601 * expr.c (expand_expr_real_2): Likewise.
2602 * optabs-tree.c (optab_for_tree_code): Likewise.
2603 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2604 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2605 * expmed.c (make_tree): Handle VEC_SERIES.
2606 * optabs.def (vec_series_optab): New optab.
2607 * optabs.h (expand_vec_series_expr): Declare.
2608 * optabs.c (expand_vec_series_expr): New function.
2609 * tree-vect-generic.c (expand_vector_operations_1): Check that
2610 the operands also have vector type.
2612 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2613 Alan Hayward <alan.hawyard@arm.com>
2614 David Sherwood <david.sherwood@arm.com>
2616 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2617 (VEC_COND_EXPR): Add missing @tindex.
2618 * doc/md.texi (vec_duplicate@var{m}): Document.
2619 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2620 * tree.c (build_vector_from_val): Add stubbed-out handling of
2621 variable-length vectors, using VEC_DUPLICATE_EXPR.
2622 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2623 * cfgexpand.c (expand_debug_expr): Likewise.
2624 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2625 * tree-inline.c (estimate_operator_cost): Likewise.
2626 * tree-pretty-print.c (dump_generic_node): Likewise.
2627 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2628 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2629 (test_vec_duplicate_folding): New function.
2630 (fold_const_c_tests): Call it.
2631 * optabs.def (vec_duplicate_optab): New optab.
2632 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2633 * optabs.h (expand_vector_broadcast): Declare.
2634 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2635 vec_duplicate_optab.
2636 * expr.c (store_constructor): Try using vec_duplicate_optab for
2638 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2640 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2643 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2644 div/mod latencies a bit.
2646 2017-12-15 Jeff Law <law@redhat.com>
2648 PR tree-optimization/36550
2649 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2650 (mark_threaded_blocks): Rewrite code to avoid block copying when
2651 optimizing for size. Don't pessimize blocks which will be
2652 copied, but all the statements will be dead.
2654 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2656 PR tree-optimization/81165
2657 * tree-ssa-threadupdate.c (uses_in_bb): New.
2658 (estimate_threading_killed_stmts): New.
2659 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2660 * tree-ssa-threadedge.c
2661 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2664 2017-12-15 Jeff Law <law@redhat.com>
2666 PR tree-optimization/83410
2667 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2668 threads when parallelizing loops.
2670 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2672 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2674 * config/alpha/alpha.c (vms_attribute_table): Swap
2675 affects_type_identity and handler fields, adjust comments.
2676 * config/mips/mips.c (mips_attribute_table): Likewise.
2677 * config/visium/visium.c (visium_attribute_table): Likewise.
2678 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2679 * config/microblaze/microblaze.c (microblaze_attribute_table):
2681 * config/spu/spu.c (spu_attribute_table): Likewise.
2682 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2683 * config/arc/arc.c (arc_attribute_table): Likewise.
2684 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2685 * config/v850/v850.c (v850_handle_interrupt_attribute,
2686 v850_handle_data_area_attribute): Formatting fixes.
2687 (v850_attribute_table): Swap affects_type_identity and handler
2688 fields, adjust comments.
2689 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2690 * config/arm/arm.c (arm_attribute_table): Likewise.
2691 * config/avr/avr.c (avr_attribute_table): Likewise.
2692 * config/s390/s390.c (s390_attribute_table): Likewise.
2693 * config/sh/sh.c (sh_attribute_table): Likewise.
2694 * config/i386/i386.c (ix86_handle_cconv_attribute,
2695 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2696 (ix86_attribute_table): Swap affects_type_identity and handler
2697 fields, adjust comments.
2698 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2699 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2700 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2701 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2702 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2703 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2704 * config/rx/rx.c (rx_attribute_table): Likewise.
2705 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2706 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2707 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2708 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2709 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2710 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2711 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2712 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2713 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2714 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2715 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2717 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2719 (register_scoped_attributes, decl_attributes): Formatting fixes.
2721 PR tree-optimization/83269
2722 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2723 subtraction in arg0's type if type is signed and arg0 is unsigned.
2727 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2728 convert? on @0 instead of convert. Check type of @1, not @0.
2729 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2730 convert? on @0 instead of convert. Check type of @1, not @0.
2731 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2732 only optimize if either both @1 and @2 types are narrower
2733 precision, or both are wider or equal precision, and in the former
2734 case only if both have undefined overflow.
2736 2017-12-15 Richard Biener <rguenther@suse.de>
2739 * internal-fn.def (IFN_NOP): Add.
2740 * internal-fn.c (expand_NOP): Do nothing.
2741 * lto-streamer-in.c (input_function): Instead of removing
2742 sanitizer calls replace them with IFN_NOP calls.
2744 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2745 Alan Hayward <alan.hayward@arm.com>
2746 David Sherwood <david.sherwood@arm.com>
2748 * dse.c (store_info, read_info_type): Replace begin and end with
2750 (print_range): New function.
2751 (set_all_positions_unneeded, any_positions_needed_p)
2752 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2754 (record_store): Likewise. Optimize the case in which all positions
2756 (get_stored_val): Replace read_begin and read_end with read_offset
2758 (replace_read): Update call accordingly.
2760 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2762 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2763 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2764 (loop_cand::loop_cand): Initialize above members.
2765 (loop_cand::supported_operations): Delete.
2766 (loop_cand::can_interchange_p): Inline above function.
2767 (loop_cand::classify_simple_reduction): Record number of constant
2768 initialized simple reductions.
2769 (should_interchange_loops): New parameters. Check stmt cost of loops
2771 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2772 Update call to should_interchange_loops.
2773 (should_interchange_loop_nest): Update call to
2774 should_interchange_loops.
2776 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2779 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2780 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2781 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2783 2017-12-15 Julia Koval <julia.koval@intel.com>
2785 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2786 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2788 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2789 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2790 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2792 2017-12-15 Julia Koval <julia.koval@intel.com>
2794 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2795 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2796 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2797 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2798 _mm_aesenc_epi128): New intrinsics.
2800 2017-12-15 Julia Koval <julia.koval@intel.com>
2802 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2803 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2805 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2806 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2807 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2809 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2811 * gimple-ssa-strength-reduction.c (analyze_increments):
2812 Distinguish replacement costs for constant strides from those for
2815 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2817 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2819 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2821 * doc/invoke.texi: Document -Wcast-function-type.
2822 * recog.h (stored_funcptr): Change signature.
2823 * tree-dump.c (dump_node): Avoid warning.
2824 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2826 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2829 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2832 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2834 PR tree-optimization/83312
2835 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2836 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2837 "val" param, and to cope with arbitrary basic blocks.
2838 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2839 handle NULL_TREE for "val", dropping "bb" param.
2840 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2841 drop "bb" param. Handle NULL_TREE for "val".
2842 (find_case_label_for_value): Make "switch_stmt" param const.
2843 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2845 (vrp_prop::check_all_array_refs): Reimplement as...
2846 (check_array_bounds_dom_walker::before_dom_children): ...this new
2847 vfunc. Replace linear search through BB block list, excluding
2848 those with non-executable in-edges via dominator walk.
2850 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2852 * config/arm/arm.opt (mverbose-cost-dump): New option.
2853 * config/arm/arm.c (arm_rtx_costs): Use it.
2855 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2858 * reload1.c (emit_input_reload_insns): Skip debug markers.
2860 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2862 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2866 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2867 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2868 * config/sh/sh.c: Skip debug insns besides notes.
2869 * config/sh/sh.md: Likewise.
2870 * config/sh/sh_treg_combine.cc: Likewise.
2871 * config/sh/sync.md: Likewise.
2873 2017-12-14 Tom de Vries <tom@codesourcery.com>
2875 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2876 Add item for weak_undefined.
2878 2017-12-14 Richard Biener <rguenther@suse.de>
2880 PR tree-optimization/67842
2881 * sese.h (bb_in_region): Remove #if 0'ed code.
2883 2017-12-14 Richard Biener <rguenther@suse.de>
2885 PR tree-optimization/83326
2886 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2887 may_be_zero parameter and handle it by not marking the first
2888 peeled copy as not exiting the loop.
2889 (try_peel_loop): Likewise.
2890 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2891 to handle the case of constant or zero iterations and perform
2892 loop header copying on-the-fly.
2894 2017-12-14 Richard Biener <rguenther@suse.de>
2896 PR tree-optimization/83418
2897 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2898 Instead of asserting we don't get unfolded comparisons deal with them.
2900 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2903 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2904 basic blocks. Assert debug bind insns don't appear outside of bbs,
2905 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2908 PR tree-optimization/83198
2909 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2910 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2911 value if arg is a REAL_CST with incompatible type.
2913 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2914 Bin Cheng <bin.cheng@arm.com>
2917 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2919 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2922 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2924 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2925 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2927 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2928 Alan Hayward <alan.hayward@arm.com>
2929 David Sherwood <david.sherwood@arm.com>
2931 * poly-int.h: New file.
2932 * poly-int-types.h: Likewise.
2933 * coretypes.h: Include them.
2934 (POLY_INT_CONVERSION): Define.
2935 * target.def (estimated_poly_value): New hook.
2936 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2937 * doc/tm.texi: Regenerate.
2938 * doc/poly-int.texi: New file.
2939 * doc/gccint.texi: Include it.
2940 * doc/rtl.texi: Describe restrictions on subreg modes.
2941 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2942 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2943 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2944 * targhooks.h (default_estimated_poly_value): Declare.
2945 * targhooks.c (default_estimated_poly_value): New function.
2946 * target.h (estimated_poly_value): Likewise.
2947 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2948 (wi::unary_traits): Delete.
2949 (wi::binary_traits::signed_shift_result_type): Define for
2950 offset_int << HOST_WIDE_INT, etc.
2951 (generic_wide_int::operator <<=): Define for all types and use
2952 wi::lshift instead of <<.
2953 (wi::hwi_with_prec): Add a default constructor.
2954 (wi::ints_for): New class.
2955 (operator <<): Define for all wide-int types.
2956 (operator /): New function.
2957 (operator %): Likewise.
2958 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2959 (ASSERT_MAYBE_NE_AT): New macros.
2961 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2962 Dominik Vogt <vogt@linux.vnet.ibm.com>
2965 * emit-rtl.c (init_emit): Remove ??? comment.
2966 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2967 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2968 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2969 alignment of 3 virtual registers to BITS_PER_WORD.
2971 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2973 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2975 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2976 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2978 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2981 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2983 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2985 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2986 with the exception of debug begin stmt markers appear before
2990 * final.c (rest_of_handle_final): Call variable_tracking_main only
2991 if !flag_var_tracking.
2993 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2994 Jakub Jelinek <jakub@redhat.com>
2998 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2999 labels debug stmts that can only appear after labels.
3001 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3003 PR rtl-optimization/82398
3004 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3005 EXPR_USEFULNESS in priority comparison.
3007 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3009 PR rtl-optimization/83393
3010 * combine.c (move_deaths): If reg_stat points to a too new insn in
3011 last_death, do not use it: find the proper insn instead.
3013 2017-12-12 Jeff Law <law@redhat.com>
3015 PR tree-optimization/83298
3016 PR tree-optimization/83362
3017 PR tree-optimization/83383
3018 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3019 push_value_range a public interface. Add new argument to
3020 record_ranges_from_stmt.
3021 * gimple-ssa-evrp-analyze.c
3022 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3023 Update comments. Handle recording temporary equivalences.
3024 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3025 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3026 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3027 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3028 gimple-ssa-evrp-analyze.h.
3029 (record_temporary_equivalences_from_phis): Add new argument. When
3030 the PHI arg is an SSA_NAME, set the result's range to the range
3032 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3033 from statements too.
3034 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3035 Pass it down to children as needed.
3036 (thread_outgoing_edges): Likewise.
3037 (thread_across_edge): Likewise. Push/pop range state as needed.
3038 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3040 2017-12-12 Julia Koval <julia.koval@intel.com>
3042 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3043 (PTA_CANNONLAKE): Remove PTA_CLWB.
3045 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3048 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3049 (vcondv2div2df): Likewise.
3050 (vconduv2dfv2di): Likewise.
3052 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3054 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3055 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3056 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3057 ATTR_MATHFN_FPROUNDING_ERRNO.
3059 2017-12-12 Richard Biener <rguenther@suse.de>
3061 PR tree-optimization/83385
3062 * graphite-scop-detection.c (get_order, order): Remove.
3063 (bb_to_rpo): New global.
3065 (build_scops): Sort pbbs in RPO order.
3067 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3069 * combine.c (simplify_set): Do not transform subregs to zero_extends
3070 if the destination mode is a vector mode.
3072 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3074 PR tree-optimization/83359
3075 * tree-cfg.h (fold_loop_internal_call): Declare.
3076 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3077 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3078 (find_loop_dist_alias): New function.
3079 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3080 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3083 PR tree-optimization/80631
3084 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3085 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3086 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3087 hardcoding zero as the value if COND_EXPR is never true. For
3088 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3089 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3090 hardcoding MAX_EXPR as the reduction operation.
3091 (is_nonwrapping_integer_induction): Allow negative step.
3092 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3093 vect_create_epilog_for_reduction, if no value is suitable, don't
3094 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3096 2017-12-12 Richard Biener <rguenther@suse.de>
3098 PR tree-optimization/81889
3099 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3100 range info from the non-wrapping IV instead of just the range
3103 2017-12-12 Julia Koval <julia.koval@intel.com>
3105 * config.gcc: Add vaesintrin.h.
3106 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3108 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3109 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3111 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3112 * config/i386/immintrin.h: Include vaesintrin.h.
3113 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3114 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3115 _mm_aesdec_epi128): New intrinsics.
3117 2017-12-12 Julia Koval <julia.koval@intel.com>
3119 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3120 OPTION_MASK_ISA_VAES_UNSET): New.
3121 (ix86_handle_option): Handle -mvaes.
3122 * config/i386/cpuid.h: Define bit_VAES.
3123 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3124 * config/i386/i386-c.c (__VAES__): New.
3125 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3126 (ix86_valid_target_attribute_inner_p): Ditto.
3127 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3128 * config/i386/i386.opt: Add -mvaes.
3129 * doc/invoke.texi: Ditto.
3131 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3133 * debug.h (gcc_debug_hooks): Add inline_entry.
3134 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3135 * debug.c (do_nothing_debug_hooks): Likewise.
3136 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3137 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3138 (dwarf2_lineno_debug_hooks): Likewise.
3140 * common.opt (gstatement-frontiers): New, setting
3141 debug_nonbind_markers_p.
3142 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3143 * toplev.c (process_options): Autodetect value for debug statement
3145 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3146 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3148 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3149 markers. Integrate source bind into debug stmt expand loop.
3150 (pass_expand::execute): Check debug marker limit. Avoid deep
3151 TER and expand debug locations for debug bind insns only.
3152 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3154 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3155 * final.c (reemit_insn_block_notes): Take current block from
3156 nonbind markers. Declare note where it's first set.
3157 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3158 begin stmt markers if enabled.
3159 (notice_source_line): Handle nonbind markers. Fail if their
3160 location is unknown or that of builtins.
3161 (rest_of_handle_final): Convert begin stmt markers to notes if
3162 var-tracking didn't run.
3163 (rest_of_clean_state): Skip begin stmt markers.
3164 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3166 * function.c (allocate_struct_function): Set begin_stmt_markers.
3167 * function.h (struct function): Add debug_marker_count counter
3168 and debug_nonbind_markers flag.
3169 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3170 * gimple-low.c (lower_function_body): Adjust
3171 debug_nonbind_markers.
3172 (lower_stmt): Drop or skip gimple debug stmts.
3173 (lower_try_catch): Skip debug stmts.
3174 * gimple.c (gimple_build_debug_begin_stmt): New.
3175 (gimple_copy): Increment debug_marker_count if copying one.
3176 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3177 * gimplify.c (rexpr_location): New.
3178 (rexpr_has_location): New.
3179 (warn_switch_unreachable_r): Handle gimple debug stmts.
3180 (shortcut_cond_r): Call expr_location.
3182 (find_goto_label): New.
3183 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3185 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3187 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3188 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3189 (LANG_HOOKS_INITIALIZER): ... this.
3190 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3191 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3193 (update_ebb_live_info): Skip debug insn markers.
3194 * lra.c (debug_insn_static_data): Rename to...
3195 (debug_bind_static_data): ... this.
3196 (debug_marker_static_data): New.
3197 (lra_set_insn_recog_data): Select one of the above depending
3199 (lra_update_isn_regno_info): Don't assume debug insns have
3201 (push_insns): Skip debug insns.
3202 * lto-streamer-in.c (input_function): Drop debug stmts
3203 depending on active options. Adjust debug_nonbind_markers.
3204 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3205 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3206 begin stmt marker notes.
3207 (print_insn): Likewise.
3208 * recog.c (extract_insn): Recognize rtl for debug markers.
3209 * rtl.def (DEBUG_MARKER): New.
3210 * tree-inline.c: Include params.h.
3211 (remap_gimple_stmt): Handle nonbind markers.
3212 (maybe_move_debug_stmts_to_successors): Likewise.
3213 (copy_debug_stmt): Likewise.
3214 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3215 markers regardless of no side effects.
3216 (tsi_link_before): Don't update container's side effects when adding
3217 a begin stmt marker.
3218 (tsi_link_after): Likewise.
3219 (expr_first): Skip begin stmt markers.
3220 (expr_last): Likewise.
3221 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3222 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3223 Disregard nonbind markers.
3224 * tree.c (make_node_stat): Don't set side effects for begin stmt
3226 (build1_stat): Likewise.
3227 * tree.def (DEBUG_BEGIN_STMT): New.
3228 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3229 * var-tracking.c (delete_debug_insns): Renamed to...
3230 (delete_vta_debug_insns): ... this.
3231 (reemit_marker_as_note): New.
3232 (vt_initialize): Reemit markers.
3233 (delete_vta_debug_insns): Likewise.
3234 (vt_debug_insns_local): Reemit or delete markers.
3235 (variable_tracking_main_1): Likewise.
3236 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3237 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3238 (gimple_debug_nonbind_marker_p): New.
3239 (gimple_build_debug_bind): Adjust.
3240 (gimple_build_debug_begin_stmt): New.
3241 * doc/invoke.texi (max-debug-marker-count): New param.
3242 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3243 (debug_parameter_ref, debug_marker): New.
3244 (NOTE_INSN_BEGIN_STMT): New.
3245 (DEBUG_INSN): Describe begin stmt markers.
3247 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3248 without debug insns, we wouldn't, but clean up debug insns
3249 after a control flow insn nevertheless.
3251 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3252 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3253 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3254 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3255 (rtl_tidy_fallthru_edge): Likewise.
3256 (rtl_verify_fallthru): Likewise.
3257 (rtl_verify_bb_layout): Likewise.
3258 (skip_insns_after_block): Likewise.
3259 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3260 * dwarf2out.c: Include print-rtl.h.
3261 (dwarf2out_next_real_insn): New.
3262 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3263 Dump debug binds in asm comments.
3264 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3265 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3266 callers to use gsi_start_nondebug_bb instead.
3267 (gsi_after_labels): Skip gimple debug stmts.
3268 (gsi_start_nondebug): New.
3269 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3270 (proper_loop_form_for_interchange): Adjust.
3271 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3272 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3273 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3274 (collect_fallthrough_labels): Likewise.
3275 (should_warn_for_implicit_fallthrough): Likewise.
3276 (warn_implicit_fallthrough_r): Likewise.
3277 (expand_FALLTHROUGH_r): Likewise.
3278 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3279 (graphite_copy_stmts_from_block): Skip nonbind markers.
3280 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3281 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3282 * jump.c (clean_barriers): Skip debug insns.
3283 * omp-expand.c (expand_parallel_call): Skip debug insns.
3284 (expand_task_call): Likewise.
3285 (remove_exit_barrier): Likewise.
3286 (expand_omp_taskreg): Likewise.
3287 (expand_omp_for_init_counts): Likewise.
3288 (expand_omp_for_generic): Likewise.
3289 (expand_omp_for_static_nochunk): Likewise.
3290 (expand_omp_for_static_chunk): Likewise.
3291 (expand_omp_simd): Likewise.
3292 (expand_omp_taskloop_for_outer): Likewise.
3293 (expand_omp_taskloop_for_inner): Likewise.
3294 (expand_oacc_for): Likewise.
3295 (expand_omp_sections): Likewise.
3296 (expand_omp_single): Likewise.
3297 (expand_omp_synch): Likewise.
3298 (expand_omp_atomic_load): Likewise.
3299 (expand_omp_atomic_store): Likewise.
3300 (expand_omp_atomic_fetch_op): Likewise.
3301 (expand_omp_atomic_pipeline): Likewise.
3302 (expand_omp_atomic_mutex): Likewise.
3303 (expand_omp_target): Likewise.
3304 (grid_expand_omp_for_loop): Likewise.
3305 (grid_expand_target_grid_body): Likewise.
3306 (build_omp_regions_1): Likewise.
3307 * omp-low.c (check_combined_parallel): Skip debug stmts.
3308 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3309 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3310 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3312 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3313 (make_edges): Likewise.
3314 (cleanup_dead_labels): Likewise.
3315 (gimple_can_merge_blocks_p): Likewise.
3316 (stmt_starts_bb_p): Likewise.
3317 (gimple_block_label): Likewise.
3318 (gimple_redirect_edge_and_branch): Likewise.
3319 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3321 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3323 * tree-pretty-print (print_declaration): Omit initializer in slim
3325 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3327 (eliminate_unnecessary_stmts): Stabilize block removal order.
3328 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3329 * var-tracking.c (get_first_insn): New.
3330 (vt_emit_notes): Call it.
3331 (vt_initialize): Walk any insns before the first BB.
3332 (delete_debug_insns): Likewise.
3334 * gimple.h (enum gimple_debug_subcode): Add
3335 GIMPLE_DEBUG_BEGIN_STMT.
3336 (gimple_debug_begin_stmt_p): New.
3337 (gimple_debug_nonbind_marker_p): New.
3338 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3339 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3340 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3341 * insn-notes.def (BEGIN_STMT): New.
3342 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3343 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3344 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3345 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3346 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3347 (INSN_DEBUG_MARKER_KIND): New.
3348 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3349 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3350 (INSN_VAR_LOCATION_PTR): New.
3351 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3353 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3354 deps only if debug bind insns are enabled.
3355 (pass_expand::execute): Avoid deep TER and expand
3356 debug locations for debug bind insns only.
3357 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3358 debug stmts special handling down to debug bind stmts.
3359 * combine.c (try_combine): Narrow debug insns special handling
3360 down to debug bind insns.
3361 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3362 Narrow debug insns preexisting special handling down to debug
3364 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3365 handling down to debug bind insns.
3366 * function.c (instantiate_virtual_regs): Skip debug markers,
3367 adjust handling of debug binds.
3368 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3369 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3370 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3371 insns to debug bind insns.
3372 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3373 special handling of debug stmts to debug bind stmts.
3374 * ipa-split.c (split_function): Likewise.
3375 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3376 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3378 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3379 * regrename.c (build_def_use): Likewise.
3380 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3381 (pass_cprop_hardreg): Narrow special casing of debug insns to
3383 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3384 * reload1.c (reload): Likewise.
3385 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3386 casing of debug stmts to debug bind stmts.
3387 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3388 * ssa-iterators.h (num_imm_uses): Likewise.
3389 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3390 debug stmts to debug bind stmts.
3391 * tree-inline.c (tree_function_versioning): Narrow special casing
3392 of debug stmts to debug bind stmts.
3393 * tree-loop-distribution.c (generate_loops_for_partition):
3394 Narrow special casing of debug stmts to debug bind stmts.
3395 * tree-sra.c (analyze_access_subtree): Narrow special casing
3396 of debug stmts to debug bind stmts.
3397 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3398 stmts to debug bind stmts.
3399 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3400 casing of debug stmts to debug bind stmts.
3401 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3402 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3403 casing of debug stmts to debug bind stmts.
3404 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3406 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3407 debug stmts to debug bind stmts.
3408 (gimple_replace_ssa_lhs): Likewise.
3409 (insert_debug_temp_for_var_def): Likewise.
3410 (insert_debug_temps_for_defs): Likewise.
3411 (reset_debug_uses): Likewise.
3412 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3413 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3414 (adjust_debug_stmts): Likewise.
3415 (adjust_phi_and_debug_stmts): Likewise.
3416 (vect_do_peeling): Likewise.
3417 * tree-vect-loop.c (vect_transform_loop): Likewise.
3418 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3419 * var-tracking.c (adjust_mems): Narrow special casing of debug
3420 insns to debug bind insns.
3421 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3422 (compute_bb_dataflow, vt_find_locations): Likewise.
3423 (vt_expand_loc, emit_notes_for_changes): Likewise.
3424 (vt_init_cfa_base): Likewise.
3425 (vt_emit_notes): Likewise.
3426 (vt_initialize): Likewise.
3427 (vt_finalize): Likewise.
3429 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3430 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3431 (next_nonnote_nondebug_insn_bb): New.
3432 (prev_nonnote_nondebug_insn_bb): New.
3433 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3434 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3435 (prev_nonnote_nondebug_insn_bb): Declare.
3436 (next_nonnote_nondebug_insn_bb): Declare.
3437 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3438 * cfgrtl.c (get_last_bb_insn): Likewise.
3439 * lra.c (push_insns): Likewise.
3441 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3444 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3446 * selftest.c (selftest::test_runner::test_runner): New ctor.
3447 (selftest::test_runner::~test_runner): New dtor.
3448 * selftest.h (class selftest::test_runner): New class.
3450 2017-12-11 Carl Love <cel@us.ibm.com>
3452 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3453 vec_extract_fp32_from_shortl]): Add #defines.
3454 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3455 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3456 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3457 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3458 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3459 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3460 * doc/extend.texi: Add documentation for the added builtins.
3462 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3464 PR rtl-optimization/80693
3465 PR rtl-optimization/81019
3466 PR rtl-optimization/81020
3467 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3468 are not mentioned in i3. Place the REG_UNUSED note on i2,
3469 possibly modified to REG_DEAD, if it did not originate in i3.
3471 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3473 * recog.c (store_data_bypass_p_1): New function.
3474 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3475 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3477 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3479 PR rtl-optimization/83361
3480 * ifcvt.c (if_convert): Call fixup_partitions.
3482 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3484 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3485 early folding of splat_u{8,16,32}.
3487 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3489 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3490 output_operand_lossage first argument with capital letter.
3491 (aarch64_override_options): Don't start error and sorry first argument
3492 with capital letter.
3494 2017-12-11 Andi Kleen <ak@linux.intel.com>
3496 PR gcov-profile/83355
3497 * auto-profile.c (string_table::get_index_by_decl): Don't
3498 recurse when abstract origin points to itself.
3500 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3502 PR tree-optimization/83320
3503 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3504 (prune_datarefs_not_in_loop): Ditto.
3506 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3508 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3510 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3512 PR tree-optimization/83337
3513 * gimple-loop-interchange.cc (compute_access_stride): Handle
3514 bitfield DRs properly.
3516 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3518 PR tree-optimization/83338
3519 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3520 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3521 vector of unsigned integers to vector of signed integers.
3523 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3525 PR rtl-optimization/83317
3526 * lra-constraints.c (process_address_1): Add insn code check.
3528 2017-12-08 Michael Matz <matz@suse.de>
3530 Fix PR tree-optimization/83323
3531 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3532 head-controlled loops and loop BBs.
3533 * common.opt (funroll-and-jam): Remove, instead ...
3534 (floop-unroll-and-jam): ... reuse this option.
3535 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3536 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3537 (-floop-unroll-and-jam): ... this option.
3539 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3541 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3542 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3543 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3545 2017-12-08 Richard Biener <rguenther@suse.de>
3548 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3549 handle accesses outside of zero-sized vars.
3551 2017-12-08 Martin Jambor <mjambor@suse.cz>
3553 PR tree-optimization/83141
3554 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3555 test for MEM_REFs implicitely changing types with padding. Remove
3557 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3559 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3561 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3563 * config/avr/avr.c (avr_attribute_table): Likewise.
3564 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3565 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3566 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3567 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3568 of struct attribute_spec.
3569 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3571 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3574 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3576 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3579 * profile-count.c (profile_count::from_gcov_type): Move from
3580 profile-count.h; handle overflow.
3581 * profile-count.h (profile_count::from_gcov_type): Move offline.
3583 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3585 PR rtl-optimization/83304
3586 * combine.c (move_deaths): If we do not know where a register died,
3589 2017-12-08 Richard Biener <rguenther@suse.de>
3591 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3592 Provide -fopt-info-loop feedback when we interchange in a nest.
3594 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3596 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3597 for armv6 ARM CPU IDs.
3599 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3601 * common/config/arm/arm-common.c: Include <algorithm>.
3602 (INCLUDE_VECTOR): Define.
3603 (compare_opt_names): New function.
3604 (arm_rewrite_selected_arch): Only strip out extensions that can be
3605 expressed through -mfpu. Sort the remaining extensions
3608 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3610 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3611 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3612 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3613 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3615 2017-06-08 Tristan Gingold <gindold@adacore.com>
3618 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3619 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3621 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3623 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3624 of struct attribute_spec.
3626 2017-12-08 Julia Koval <julia.koval@intel.com>
3628 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3629 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3630 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3631 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3632 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3633 _mm_maskz_dpwssds_epi32): Ditto.
3635 2017-12-08 Richard Biener <rguenther@suse.de>
3637 PR tree-optimization/81303
3638 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3639 conditions try to create a comparison vector type matching
3640 the data vector type.
3641 (vectorizable_condition): Adjust.
3642 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3643 Leave invariant conditions alone in case we can vectorize those.
3645 2017-12-08 Julia Koval <julia.koval@intel.com>
3647 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3648 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3649 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3650 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3651 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3653 2017-12-08 Julia Koval <julia.koval@intel.com>
3655 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3656 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3657 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3658 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3659 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3660 _mm_maskz_dpbusds_epi32): New intrinsics.
3662 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3664 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3666 (ld<bh>io_signed): Likewise.
3667 (st<bhw>io): Likewise.
3668 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3669 SMALL_INT12 constant integer operand.
3671 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3673 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3675 (nios2_large_constant_allowed): ...this. Adjust uses.
3676 (nios2_plus_symbolic_constant_p): Rename to...
3677 (nios2_plus_large_constant_p): ...this. Adjust uses.
3678 (nios2_legitimate_address_p): Correct CONST_INT handling.
3679 (nios2_symbolic_memory_operand_p): Rename to...
3680 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3681 (nios2_large_constant_p): Check for large constant integers too.
3682 (nios2_split_large_constant): Handle constant integers.
3683 (nios2_split_symbolic_memory_operand): Rename to...
3684 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3685 (nios2_legitimize_constant_address): Handle constant integers.
3686 (r0rel_constant_p): Handle small constant integers.
3687 (nios2_print_operand_address): Handle r0-relative integer addresses.
3688 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3689 * config/nios2/nios2.md: Adjust for renamed functions.
3691 2017-12-07 Andrew Waterman <andrew@sifive.com>
3693 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3694 (TARGET_HAVE_SRODATA_SECTION): New define.
3695 (riscv_select_section): New function.
3697 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3698 Alexander Monakov <amonakov@ispras.ru>
3699 Jakub Jelinek <jakub@redhat.com>
3702 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3704 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3706 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3707 a bool success value. Don't call output_operand_lossage here.
3708 (aarch64_print_ldpstp_address): Return a bool success value.
3709 (aarch64_print_operand_address): Call output_addr_const if
3710 aarch64_print_address_internal fails.
3711 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3712 'y'; call output_operand_lossage instead. Call output_operand_lossage
3713 if aarch64_print_ldpstp_address fails.
3715 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3717 * tree-vector-builder.h
3718 (tree_vector_builder::binary_encoded_nelts): Declare.
3719 * tree-vector-builder.c
3720 (tree_vector_builder::binary_encoded_nelts): New function.
3721 * fold-const.c (negate_expr_p): Likewise.
3722 (operand_equal_p, fold_checksum_tree): Likewise.
3723 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3724 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3725 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3726 (uniform_vector_p): Likewise.
3727 * varasm.c (const_hash_1, compare_constant): Likewise.
3728 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3729 (valid_lattice_transition): Operate directly on the VECTOR_CST
3731 * ipa-icf.c: Include tree-vector-builder.h.
3732 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3733 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3735 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3737 * tree.c (build_vector): Delete.
3738 * tree.h (build_vector): Make static and move into the self-testing
3741 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3743 * vector-builder.h (vector_builder::derived): New const overload.
3744 (vector_builder::elt): New function.
3745 * tree-vector-builder.h (tree_vector_builder::type): New function.
3746 (tree_vector_builder::apply_step): Declare.
3747 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3749 * gimple-fold.h (tree_vector_builder): Declare.
3750 (gimple_build_vector): Take a tree_vector_builder instead of a
3751 type and vector of elements.
3752 * gimple-fold.c (gimple_build_vector): Likewise.
3753 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3755 (get_initial_defs_for_reduction): Likewise.
3756 (vectorizable_induction): Likewise.
3758 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3760 * tree-vector-builder.h
3761 (tree_vector_builder::new_binary_operation): Declare.
3762 * tree-vector-builder.c
3763 (tree_vector_builder::new_binary_operation): New function.
3764 * fold-const.c (fold_relational_const): Use it.
3765 (const_binop): Likewise. Check that both input vectors have
3766 the same number of elements, thus excluding things like WIDEN_SUM.
3767 Check whether it is possible to operate directly on the encodings
3770 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3772 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3773 new_unary_operation, operating only on the encoded elements.
3774 (const_unop): Likewise.
3775 (exact_inverse): Likewise.
3776 (distributes_over_addition_p): New function.
3777 (const_binop): Use tree_vector_builder and new_unary_operation
3778 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3779 on the encoded elements unless the encoding is strided and the
3780 operation does not distribute over addition.
3781 (fold_convert_const): Use tree_vector_builder and
3782 new_unary_operation. Operate only on the encoded elements
3783 for truncating integer conversions, or for non-stepped encodings.
3785 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3787 * config/sparc/sparc.c: Include tree-vector-builder.h.
3788 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3789 * expmed.c: Include tree-vector-builder.h.
3790 (make_tree): Use tree_vector_builder instead of build_vector.
3791 * fold-const.c: Include tree-vector-builder.h.
3792 (const_binop): Use tree_vector_builder instead of build_vector.
3793 (const_unop): Likewise.
3794 (native_interpret_vector): Likewise.
3795 (fold_vec_perm): Likewise.
3796 (fold_ternary_loc): Likewise.
3797 * gimple-fold.c: Include tree-vector-builder.h.
3798 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3800 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3801 (simplify_vector_constructor): Use tree_vector_builder instead
3803 * tree-vect-generic.c: Include tree-vector-builder.h.
3804 (add_rshift): Use tree_vector_builder instead of build_vector.
3805 (expand_vector_divmod): Likewise.
3806 (optimize_vector_constructor): Likewise.
3807 * tree-vect-loop.c: Include tree-vector-builder.h.
3808 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3809 of build_vector. Explicitly use a stepped encoding for
3811 * tree-vect-slp.c: Include tree-vector-builder.h.
3812 (vect_get_constant_vectors): Use tree_vector_builder instead
3814 (vect_transform_slp_perm_load): Likewise.
3815 (vect_schedule_slp_instance): Likewise.
3816 * tree-vect-stmts.c: Include tree-vector-builder.h.
3817 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3818 (vect_gen_perm_mask_any): Likewise.
3819 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3820 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3822 (build_vector_from_val): Likewise. Explicitly use a duplicate
3825 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3827 * doc/generic.texi (VECTOR_CST): Describe new representation of
3829 * vector-builder.h: New file.
3830 * tree-vector-builder.h: Likewise.
3831 * tree-vector-builder.c: Likewise.
3832 * Makefile.in (OBJS): Add tree-vector-builder.o.
3833 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3834 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3835 (tree_vector): Change the number of elements to
3836 vector_cst_encoded_nelts.
3837 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3838 (VECTOR_CST_ELTS): Delete.
3839 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3840 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3841 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3842 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3843 (VECTOR_CST_ENCODED_ELT): Likewise.
3844 (vector_cst_encoded_nelts): New function.
3845 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3846 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3847 (vector_cst_int_elt, vector_cst_elt): Declare.
3848 * tree.c: Include tree-vector-builder.h.
3849 (tree_code_size): Abort if passed VECTOR_CST.
3850 (tree_size): Update for new VECTOR_CST layout.
3851 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3852 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3853 (build_vector): Use tree_vector_builder.
3854 (vector_cst_int_elt, vector_cst_elt): New functions.
3855 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3856 encoded elements and then create the vector in the canonical form.
3857 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3858 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3859 (tree_c_tests): Call test_vector_cst_patterns.
3860 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3862 (hash_tree): Likewise.
3863 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3864 (streamer_write_tree_header): Likewise.
3865 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3866 (streamer_alloc_tree): Likewise. Update call to make_vector.
3867 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3869 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3871 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3872 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3873 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3875 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3876 Richard Biener <rguenther@suse.de>
3878 PR tree-optimization/81303
3879 * Makefile.in (gimple-loop-interchange.o): New object file.
3880 * common.opt (floop-interchange): Reuse the option from graphite.
3881 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3882 -floop-interchange and mention it for -O3.
3883 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3884 * gimple-loop-interchange.cc: New file.
3885 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3886 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3887 * passes.def (pass_linterchange): New pass.
3888 * timevar.def (TV_LINTERCHANGE): New time var.
3889 * tree-pass.h (make_pass_linterchange): New declaration.
3890 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3891 interchange. Record IV before/after increment in new parameters.
3892 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3893 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3895 (check_reduction_path): ...New function here.
3896 * tree-vectorizer.h (check_reduction_path): New declaration.
3898 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3901 PR rtl-optimization/80818
3902 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3903 always early clobbered.
3904 * lra-lives.c (process_bb_lives): Check input hard regs for early
3905 clobbered non-operand hard reg.
3907 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3910 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3911 types_compatible_p, just that TYPE_MODE is the same.
3913 2017-12-07 Martin Sebor <msebor@redhat.com>
3916 * attribs.c (empty_attribute_table): Initialize new member of
3917 struct attribute_spec.
3918 (decl_attributes): Add argument. Handle mutually exclusive
3919 combinations of attributes.
3920 (selftests::test_attribute_exclusions): New function.
3921 (selftests::attribute_c_tests): Ditto.
3922 * attribs.h (decl_attributes): Add default argument.
3923 * selftest.h (attribute_c_tests): Declare.
3924 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3925 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3927 * doc/extend.texi (Common Function Attributes): Update const and pure.
3928 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3929 of struct attribute_spec.
3930 * config/arc/arc.c (arc_attribute_table): Same.
3931 * config/arm/arm.c (arm_attribute_table): Same.
3932 * config/avr/avr.c ( avr_attribute_table): Same.
3933 * config/bfin/bfin.c (bfin_attribute_table): Same.
3934 * config/cr16/cr16.c (cr16_attribute_table): Same.
3935 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3936 * config/h8300/h8300.c (h8300_attribute_table): Same.
3937 * config/i386/i386.c (ix86_attribute_table): Same.
3938 * config/ia64/ia64.c (ia64_attribute_table): Same.
3939 * config/m32c/m32c.c (m32c_attribute_table): Same.
3940 * config/m32r/m32r.c (m32r_attribute_table): Same.
3941 * config/m68k/m68k.c (m68k_attribute_table): Same.
3942 * config/mcore/mcore.c (mcore_attribute_table): Same.
3943 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3944 * config/mips/mips.c (mips_attribute_table): Same.
3945 * config/msp430/msp430.c (msp430_attribute_table): Same.
3946 * config/nds32/nds32.c (nds32_attribute_table): Same.
3947 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3948 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3949 * config/rl78/rl78.c (rl78__attribute_table): Same.
3950 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3951 * onfig/rx/rx.c (rx_attribute_table): Same.
3952 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3953 * config/sh/sh.c (sh_attribute_table): Same.
3954 * config/sparc/sparc.c (sparc_attribute_table): Same.
3955 * config/spu/spu.c (spu_attribute_table): Same.
3956 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3957 * config/v850/v850.c (v850_attribute_table): Same.
3958 * config/visium/visium.c (visium_attribute_table): Same.
3960 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3963 * config/arm/arm.c (INCLUDE_STRING): Define.
3964 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3965 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3968 2017-12-07 Michael Matz <matz@suse.de>
3970 Add unroll and jam pass
3972 * gimple-loop-jam.c: New file.
3973 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3974 * common.opt (funroll-and-jam): New option.
3975 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3976 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3977 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3978 * passes.def: Add pass_loop_jam.
3979 * timevar.def (TV_LOOP_JAM): Add.
3980 * tree-pass.h (make_pass_loop_jam): Declare.
3981 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3982 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3983 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3984 to flow_loop_tree_node_add.
3985 (duplicate_subloops, copy_loops_to): Append to sibling list.
3986 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3987 * doc/invoke.texi (-funroll-and-jam): Document new option.
3988 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3990 2017-12-07 Richard Biener <rguenther@suse.de>
3992 PR tree-optimization/83296
3993 PR tree-optimization/67769
3994 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3995 flow sensitive info in an unrelated BB.
3996 (value_replacement): Use reset_flow_sensitive_info.
3997 (minmax_replacement): Reset flow sensitive info on the def
3998 we move. Do not reset flow sensitive info in the whole BB
3999 we move the stmt to.
4000 (abs_replacement): Likewise.
4002 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4005 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4006 rs6000_cpu to the given -mcpu=, or to the default processor.
4008 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4010 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4011 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4012 instead of rs6000_cpu_attr.
4013 (rs6000_adjust_cost): Ditto.
4014 (is_microcoded_insn): Ditto.
4015 (rs6000_adjust_priority): Ditto.
4016 (rs6000_issue_rate): Ditto.
4017 (rs6000_use_sched_lookahead): Ditto.
4018 (rs6000_use_sched_lookahead_guard): Ditto.
4019 (rs6000_sched_reorder): Ditto.
4020 (force_new_group): Ditto.
4021 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4022 (group_ending_nop): Ditto.
4024 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4026 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4027 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4028 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4029 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4031 (rs6000_reassociation_width): Ditto.
4032 (rs6000_emit_epilogue): Ditto.
4033 (rs6000_adjust_cost): Ditto.
4034 (is_microcoded_insn): Ditto.
4035 (is_cracked_insn): Ditto.
4036 (rs6000_adjust_priority): Ditto.
4037 (rs6000_sched_reorder): Ditto.
4038 (rs6000_sched_reorder2): Ditto.
4039 (insn_must_be_first_in_group): Ditto.
4040 (insn_must_be_last_in_group): Ditto.
4041 (rs6000_register_move_cost): Ditto.
4042 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4045 2017-12-07 Julia Koval <julia.koval@intel.com>
4047 * config.gcc: Add vaesintrin.h.
4048 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4050 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4051 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4053 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4054 * config/i386/immintrin.h: Include vaesintrin.h.
4055 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4056 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4057 _mm_aesdec_epi128): New intrinsics.
4059 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4061 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4062 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4063 c-family/c-spellcheck.cc.
4064 (best_macro_match::best_macro_match): Likewise.
4066 (struct edit_distance_traits<cpp_hashnode *>): Move to
4067 c-family/c-spellcheck.h.
4068 (class best_macro_match): Likewise.
4070 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4072 PR tree-optimization/83293
4073 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4074 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4075 might insert into empty bb.
4078 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4079 simplify for plus with :c added, and pointer_plus without that.
4080 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4081 with undefined overflow and the conversion is not widening,
4082 perform negation in utype and only convert to type afterwards.
4083 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4084 simplify for plus with :c added, and pointer_plus without that.
4085 If type is integral with undefined overflow and the conversion is
4086 not widening, perform minus in utype and only convert to type
4087 afterwards. Move the last pointer_diff_expr simplify into the
4090 2017-12-06 Martin Sebor <msebor@redhat.com>
4092 PR tree-optimization/82646
4093 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4094 strncpy, not maxlen.
4096 2017-12-06 Martin Sebor <msebor@redhat.com>
4098 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4101 PR tree-optimization/83075
4102 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4103 strncat/strncpy don't change length of source string.
4105 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4108 2017-11-29 Martin Aberg <maberg@gaisler.com>
4110 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4111 to prevent b2bst errata sequence.
4112 (sqrtdf2_fix): Likewise.
4114 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4116 PR tree-optimization/81945
4117 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4118 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4119 to dest_cfun has orig_loop_num set, either remap it to the new
4120 loop number if the loop got moved too, or clear it.
4122 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4124 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4126 (thunderx2t99_nothing): New insn reservation.
4127 (thunderx2t99_mrs): New insn reservation.
4128 (thunderx2t99_multiple): New insn reservation.
4129 (thunderx2t99_alu_basi): Add bfx to reservation.
4130 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4132 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4135 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4137 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4139 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4140 tree-ssa-pre.c::remove_dead_inserted_code.
4141 * tree-ssa-dce.h: New file.
4142 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4143 (remove_dead_inserted_code): Move and rename to function
4144 tree-ssa-dce.c::simple_dce_from_worklist.
4145 (pass_pre::execute): Update use.
4147 2017-12-05 Richard Biener <rguenther@suse.de>
4149 PR tree-optimization/83277
4150 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4151 to code-gen liveout vars.
4153 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4155 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4156 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4157 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4160 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4162 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4164 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4165 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4168 2017-12-05 Richard Biener <rguenther@suse.de>
4170 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4171 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4172 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4173 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4175 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4177 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4179 2017-12-05 Martin Liska <mliska@suse.cz>
4180 Jakub Jelinek <jakub@redhat.com>
4182 * doc/invoke.texi: Document the options.
4183 * flag-types.h (enum sanitize_code): Add
4184 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4185 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4186 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4187 * opts.c: Define new sanitizer options.
4188 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4189 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4191 2017-12-05 Julia Koval <julia.koval@intel.com>
4193 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4194 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4195 (ix86_handle_option): Handle -mavx512vnni.
4196 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4197 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4198 * config/i386/i386-c (__AVX512VNNI__): New.
4199 * config/i386/i386.c (ix86_target_string): Handle new option.
4200 (ix86_valid_target_attribute_inner_p): Handle new option.
4201 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4202 * config/i386/i386.opt (mavx512vnni): New option.
4204 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4207 * athlon.md: Disable for generic.
4208 * haswell.md: Enable for generic.
4209 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4210 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4212 (ix86_adjust_cost): Move generic to haswell path.
4214 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4216 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4217 instead of MEM_P in a couple more places. Fix formatting issues.
4219 2017-12-04 Jim Wilson <jimw@sifive.com>
4221 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4222 instead of GP_REG_LAST-1.
4223 (riscv_adjust_libcall_cfi_prologue): Likewise.
4224 (riscv_adjust_libcall_cri_epilogue): Likewise.
4225 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4228 2017-12-04 Luis Machado <luis.machado@linaro.org>
4230 * ipa-pure-const.c (check_decl): Add missing newline.
4231 (state_from_flags): Likewise.
4233 2017-12-04 Jeff Law <law@redhat.com>
4235 PR tree-optimizatin/78496
4236 * gimple-ssa-evrp-analyze.h
4237 (evrp_range_analyzer::get_vr_values): Simplify.
4238 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4239 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4240 and gimple-ssa-evrp-analyze.h.
4241 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4242 (simplify_stmt_for_jump_threading): Copy a blob of code from
4243 tree-vrp.c to use ranges to simplify statements.
4244 (dom_opt_dom_walker::before_dom_children): Call
4245 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4246 (dom_opt_dom_walker::after_dom_children): Similarly for
4247 evrp_range_analyzer::leave.
4248 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4251 * gimple-ssa-evrp-analyze.c
4252 (evrp_range_analyzer::extract_range_from_stmt): Always use
4253 vr_values::update_value_range so preexisting range info is
4254 medged with new range info, even if the new range is VR_VARYING.
4256 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4258 * combine.c: Adjust comment.
4259 (use_crosses_set_p): Delete.
4260 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4261 (try_combine): Ditto.
4263 2017-12-04 Richard Biener <rguenther@suse.de>
4265 PR tree-optimization/83255
4266 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4267 Re-add zero-iteration check.
4269 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4271 PR rtl-optimization/83245
4272 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4273 hard registers as earlyclobber, also if not in an asm.
4275 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4279 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4282 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4283 rs6000_cpu based on cpu_index, not tune_index.
4285 2017-12-04 Richard Biener <rguenther@suse.de>
4287 PR tree-optimization/83238
4288 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4289 code match comment, rejecting invalid SESE regions.
4291 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4293 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4294 require base operand is a REG_POINTER prior to reload on targets
4295 with non-equivalent space registers.
4297 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4299 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4300 (update_specialized_profile): Fix updating of counts.
4301 (perhaps_add_new_callers): Likewise.
4303 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4306 * x86-tune.def: Remove obsolette FIXMEs.
4307 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4308 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4309 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4311 (X86_TUNE_PAD_RETURNS): Disable for generic.
4313 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4315 PR tree-optimization/83170
4316 PR tree-optimization/83241
4317 * gimple-ssa-store-merging.c
4318 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4319 gimple_vuse (ins_stmt) in case it has changed.
4320 (imm_store_chain_info::output_merged_store): Likewise.
4322 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4326 * tree-cfg.c (pass_warn_function_return::execute): Handle
4327 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4328 with BUILTINS_LOCATION.
4332 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4333 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4334 instead of DECL_MODE.
4336 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4337 last argument to rtx pointer.
4338 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4339 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4340 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4341 depending on the chosen ISAs.
4342 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4343 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4344 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4346 * config/i386/sse.md (mov<mode>_internal): Likewise.
4347 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4349 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4351 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4353 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4355 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4356 parameters from prototype.
4357 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4358 parameters. Don't print a comment.
4359 (emit_fusion_gpr_load): Adjust.
4360 (emit_fusion_load_store): Adjust.
4361 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4362 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4363 comment on the second line.
4365 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4368 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4369 rs6000_cpu based on cpu_index, not tune_index.
4371 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4373 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4374 which_alternative instead of which_alternative + 1.
4375 (output_asm_insn): Print an extra tab if the template is short.
4377 2017-12-01 Jim Wilson <jimw@sifive.com>
4379 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4381 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4382 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4384 * doc/tm.texi: Regenerate.
4386 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4389 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4390 whether we can allocate pseudos before trying to fix an address.
4391 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4392 memory address is indexed or indirect.
4393 (floatuns_<mode>si2_hw2): Likewise.
4395 2017-12-01 Jason Merrill <jason@redhat.com>
4397 * Makefile.in (TAGS): Add c-family/*.cc.
4399 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4401 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4402 (call_value_insn): Likewise.
4403 (sibcall_insn): Likewise.
4404 (sibcall_value_insn): Likewise.
4405 (movsi_aarch64): Likewise.
4406 (movdi_aarch64): Likewise.
4407 (add_losym_): Likewise.
4408 (ldr_got_small_): Likewise.
4409 (ldr_got_small_sidi): Likewise.
4410 (ldr_got_small_28k_): Likewise.
4411 (ldr_got_small_28k_sidi): Likewise.
4412 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4413 Move output_addr_const to symbolic case. Add error check.
4415 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4417 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4418 * config/s390/s390.md: Disable MVC merging peephole if it would
4419 disable operand forwarding.
4420 (new peephole2): Split MVCs if it would turn them into up to 2
4423 2017-12-01 Richard Biener <rguenther@suse.de>
4425 PR tree-optimization/83232
4426 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4427 detection of same access. Instead of breaking the group here
4428 do not consider the duplicate. Add comment explaining real fix.
4430 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4432 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4434 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4436 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4438 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4440 * function.h (struct function): Remove cilk_frame_decl,
4441 is_cilk_function and calls_cilk_spawn fields.
4442 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4444 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4445 cilk_elemental field.
4446 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4447 * target.def: Adjust comment.
4448 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4449 Don't test cilk_elemental.
4451 PR tree-optimization/83233
4452 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4453 bswap_stat name for the struct.
4456 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4457 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4458 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4459 SWITCH_BREAK_LABEL_P set on the label.
4460 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4461 added for default case if it was missing and not all cases covered.
4462 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4463 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4465 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4466 to avoid -Wimplicit-fallthrough warning.
4467 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4468 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4470 PR tree-optimization/83221
4471 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4473 (init_reassoc): Formatting fix.
4476 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4477 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4481 * tree-cfg.c: Include asan.h.
4482 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4484 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4486 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4488 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4490 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4491 vpcmpeqd instruction.
4493 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4495 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4496 registers type for 128bit mode.
4498 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4500 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4502 * gengtype-state.c (read_state_pair): Likewise.
4503 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4504 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4505 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4506 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4507 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4508 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4509 * builtins.c (fold_builtin_3): Likewise.
4510 * graphite-scop-detection.c
4511 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4512 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4515 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4518 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4519 parameter. Return true if ignore_decl_rtl_set_p is true and other
4520 conditions are satisfied.
4521 * asan.h (asan_protect_global): Add new parameter.
4522 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4523 to asan_protect_global calls.
4525 2017-11-30 Jim Wilson <jimw@sifive.com>
4527 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4528 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4529 -mstrict-align, add info on default value. Delete redundant lines for
4530 -mabi. Add missing -mexplicit-relocs docs.
4532 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4534 * config/arc/arc.md (trap): New pattern.
4536 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4538 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4539 instruction to end into a delay slot.
4540 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4541 can be placed into a delay slot against reg_note.
4543 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4545 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4546 labels number of usages.
4548 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4550 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4552 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4554 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4557 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4558 multiplication by power of 2 constant into two shifts + comparison.
4560 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4563 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4565 2017-11-30 Richard Biener <rguenther@suse.de>
4567 PR tree-optimization/83202
4568 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4569 allow_peel argument and guard peeling.
4570 (canonicalize_loop_induction_variables): Likewise.
4571 (canonicalize_induction_variables): Pass false.
4572 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4573 peeling from cunrolli.
4575 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4577 * combine.c (try_combine): Print a message to dump file whenever
4578 I0, I1, or I2 cannot be combined into I3.
4580 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4582 PR rtl-optimization/83156
4583 PR rtl-optimization/82621
4584 * combine.c (try_combine): Don't split an I2 if one of the dests is
4585 set again before I3. Allow unused dests.
4587 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4589 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4591 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4593 PR rtl-optimization/80818
4594 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4595 recursively. Use insn code for clobber.
4596 (lra_set_insn_recog_data): Pass the new arg to
4597 collect_non_operand_hard_regs.
4598 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4600 (lra_update_insn_regno_info): Pass insn to
4601 add_regs_to_insn_regno_info.
4603 2017-11-29 Jim Wilson <jimw@sifive.com>
4604 Andrew Waterman <andrew@sifive.com>
4606 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4607 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4608 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4609 use SINGLE_SHIFT_COST.
4610 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4611 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4613 2017-11-29 Julia Koval <julia.koval@intel.com>
4615 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4616 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4617 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4618 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4619 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4620 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4621 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4622 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4623 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4624 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4625 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4626 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4627 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4628 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4629 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4630 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4631 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4632 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4633 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4634 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4635 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4636 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4637 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4638 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4639 __builtin_ia32_vpshldv_v2di_mask,
4640 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4641 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4642 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4644 2017-11-29 Julia Koval <julia.koval@intel.com>
4646 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4647 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4648 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4649 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4650 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4651 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4652 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4653 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4654 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4655 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4656 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4657 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4658 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4659 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4660 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4661 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4662 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4663 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4664 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4665 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4666 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4668 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4670 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4671 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4674 2017-11-29 Martin Aberg <maberg@gaisler.com>
4676 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4677 to prevent b2bst errata sequence.
4678 (sqrtdf2_fix): Likewise.
4680 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4682 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4683 (div_sqrt_insn_p): New function.
4684 (sparc_do_work_around_errata): Insert NOP instructions to
4685 prevent sequences that could trigger the TN-0013 errata for
4686 certain LEON3 processors.
4687 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4688 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4689 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4690 (in_branch_delay): Prevent div and sqrt in delay slot if
4692 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4694 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4696 * config/sparc/sparc.c (atomic_insn_p): New function.
4697 (sparc_do_work_around_errata): Insert NOP instructions to
4698 prevent sequences that could trigger the TN-0010 errata for
4700 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4701 instruction referable in atomic_insns_p.
4703 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4705 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4706 (atomic_compare_and_swap_leon3_1): Likewise.
4709 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4711 * config/sparc/sparc.c (fpop_insn_p): New function.
4712 (sparc_do_work_around_errata): Insert NOP instructions to
4713 prevent sequences that could trigger the TN-0012 errata for
4715 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4716 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4717 (in_branch_annul_delay): Prevent floating-point instructions
4718 in delay slot of annulled integer branch.
4720 2017-11-29 Richard Biener <rguenther@suse.de>
4722 PR tree-optimization/83202
4723 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4725 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4726 nodes vectorized to the same stmts multiple times.
4727 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4728 (vect_analyze_slp_instance): Adjust.
4729 (scalar_stmts_to_slp_tree_map_t): New typedef.
4730 (vect_schedule_slp_instance): Add a map recording the SLP node
4731 representing the vectorized stmts for a set of scalar stmts.
4732 Avoid code-generating redundancies.
4733 (vect_schedule_slp): Allocate map and pass it down.
4735 2017-11-29 Nathan Sidwell <nathan@acm.org>
4738 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4741 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4743 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4744 cases for printing LDP/STP memory addresses.
4745 (aarch64_print_address_internal): Renamed from
4746 aarch64_print_operand_address, added parameter, add Pmode check.
4747 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4748 (aarch64_print_operand_address): Indirect to
4749 aarch64_print_address_internal.
4750 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4751 'y' operand output specifier.
4753 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4756 * tree.c (build_simple_mem_ref_loc): Handle
4757 get_addr_base_and_unit_offset returning a MEM_REF.
4760 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4763 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4766 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4768 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4770 (layout::move_to_column): Likewise.
4772 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4775 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4776 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4778 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4779 Martin Jambor <mjambor@suse.cz>
4782 * tree.h (expr_type_first_operand_type_p): Declare
4783 * tree.c (expr_type_first_operand_type_p): New function.
4784 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4785 (ipa_value_from_jfunc): Adjust declaration.
4786 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4787 Use it as result type for arithmetics, unless it is NULL in which case
4788 be more conservative.
4789 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4790 ipa_get_jf_pass_through_result.
4791 (propagate_vals_across_pass_through): Likewise.
4792 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4793 is to propagate_vals_across_pass_through.
4794 (propagate_constants_across_call): Pass PARM_TYPE to
4795 propagate_scalar_across_jump_function.
4796 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4797 ipa_value_from_jfunc.
4798 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4799 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4800 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4801 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4802 target_type, pass it to ipa_value_from_jfunc.
4803 (update_indirect_edges_after_inlining): Pass parameter type to
4804 try_make_edge_direct_simple_call.
4806 2017-11-28 Jeff Law <law@redhat.com>
4808 * gimple-ssa-evrp-analyze.c
4809 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4810 refine ranges if scev_initialized_p returns true.
4811 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4813 2017-11-28 Julia Koval <julia.koval@intel.com>
4815 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4816 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4817 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4818 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4819 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4820 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4821 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4822 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4823 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4824 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4825 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4826 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4827 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4828 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4829 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4830 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4831 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4832 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4833 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4834 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4835 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4836 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4838 2017-11-28 Julia Koval <julia.koval@intel.com>
4840 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4841 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4842 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4843 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4844 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4845 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4846 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4847 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4848 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4849 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4850 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4851 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4852 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4853 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4854 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4855 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4856 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4857 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4858 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4859 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4860 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4861 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4862 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4863 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4864 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4865 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4866 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4867 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4868 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4869 __builtin_ia32_vpshld_v2di_mask): New builtins.
4870 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4871 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4873 2017-11-28 Richard Biener <rguenther@suse.de>
4875 PR tree-optimization/80776
4876 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4878 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4880 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4881 If the incoming edge is an effective fallthru because the other
4882 edge only reaches a __builtin_unreachable () then record ranges
4883 derived from the controlling condition in SSA info.
4884 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4885 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4887 2017-11-28 Olivier Hainque <hainque@adacore.com>
4889 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4890 on mingw build hosts.
4892 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4895 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4897 2017-11-28 Richard Biener <rguenther@suse.de>
4900 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4901 copies generated from memcpy use a character array as reference
4904 2017-11-28 Julia Koval <julia.koval@intel.com>
4905 Sebastian Peryt <sebastian.peryt@intel.com>
4907 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4908 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4909 cilk-common.o, cilk.h, cilk-common.c): Remove.
4911 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4912 * builtins.c (is_builtin_name): Remove cilkplus condition.
4913 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4914 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4915 cilk-builtins.def, cilkplus.def): Remove.
4916 * cif-code.def (CILK_SPAWN): Remove.
4917 * cilk-builtins.def: Delete.
4918 * cilk-common.c: Ditto.
4920 * cilkplus.def: Ditto.
4921 * config/darwin.h (fcilkplus): Delete.
4922 * cppbuiltin.c: Ditto.
4923 * doc/extend.texi: Remove cilkplus doc.
4924 * doc/generic.texi: Ditto.
4925 * doc/invoke.texi: Ditto.
4926 * doc/passes.texi: Ditto.
4927 * gcc.c (fcilkplus): Remove.
4928 * gengtype.c (cilk.h): Remove.
4929 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4931 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4933 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4935 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4936 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4937 cilkplus conditions.
4938 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4939 inline_read_section): Ditto.
4940 * ipa-inline-analysis.c (cilk.h): Remove.
4941 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4942 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4943 append_linker_options): Remove condition for fcilkplus.
4944 * lto/lto-lang.c (cilk.h): Remove.
4945 (lto_init): Remove condition for fcilkplus.
4946 * omp-expand.c (expand_cilk_for_call): Delete.
4947 (expand_omp_taskreg, expand_omp_for_static_chunk,
4948 expand_omp_for): Remove cilkplus
4950 (expand_cilk_for): Delete.
4951 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4952 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4953 execute_lower_omp, diagnose_sb_0): Ditto.
4954 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4955 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4956 * tree-nested.c: Ditto.
4957 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4958 (dump_generic_node): Ditto.
4959 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4960 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4962 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4964 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4966 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4967 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4968 for vector mode and !TARGET_SIMD.
4970 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4972 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4974 * tree.h (SWITCH_LABELS): Remove.
4975 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4976 assert SWITCH_BODY is non-NULL.
4977 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4979 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4981 PR tree-optimization/80788
4982 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4983 has TREE_OVERFLOW set, call drop_tree_overflow.
4985 2017-11-28 Richard Biener <rguenther@suse.de>
4987 PR tree-optimization/83158
4988 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4990 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4993 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4994 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4996 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4998 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4999 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5001 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5004 * varasm.c (compare_constant): Take the mode of the constants into
5005 account when comparing floating point constants.
5007 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5009 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5010 from explicit instantiation of debug_helper.
5011 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5013 2017-11-27 Richard Biener <rguenther@suse.de>
5015 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5018 2017-11-27 Richard Biener <rguenther@suse.de>
5020 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5021 (type_hash_canon): Also clear int_cst_hash_table entry for
5023 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5025 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5027 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5028 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5029 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5031 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5033 * hash-map.h (gt_cleare_cache): Avoid UB.
5035 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5037 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5038 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5039 (decide_unroll_runtime_iterations): Likewise.
5040 (decide_unroll_stupid): Likewise.
5042 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5045 * config/i386/i386.md: Add a loop with incssp.
5047 2017-11-27 Martin Jambor <mjambor@suse.cz>
5049 PR tree-optimization/81248
5050 * tree-sra.c (splice_param_accesses): Remove size check.
5051 (decide_one_param_reduction): Fix size check.
5052 * gimple-pretty-print.c (dump_profile): Silence warning.
5053 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5055 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5058 * dbxout.c (lastlineno): New variable.
5059 (dbx_debug_hooks): Use dbxout_switch_text_section as
5060 switch_text_section debug hook.
5061 (dbxout_function_end): Switch to current_function_section
5062 rather than function_section. If crtl->has_bb_partition,
5063 output just one N_FUN, depending on in_cold_section_p.
5064 (dbxout_source_line): Remember last lineno in lastlineno.
5065 (dbxout_switch_text_section): New function.
5066 (dbxout_function_decl): Adjust dbxout_block caller.
5067 (dbx_block_with_cold_children): New function.
5068 (dbxout_block): Return true if any LBRAC/RBRAC have been
5069 emitted. Use dbx_block_with_cold_children at depth == 0
5070 in second partition. Add PARENT_BLOCKNUM argument, pass
5071 it optionally adjusted to children. Output LBRAC/RBRAC
5072 around recursive call only if the block is in the current
5073 partition, if not and anything was output, emit empty
5075 * final.c (final_scan_insn): Compute cold_function_name
5076 before calling switch_text_section debug hook. Call
5077 that hook even if dwarf2out_do_frame if not emitting
5081 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5082 TREE_READONLY decls.
5084 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5086 PR rtl-optimization/82488
5087 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5089 2017-11-26 Julia Koval <julia.koval@intel.com>
5091 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5093 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5096 2017-11-26 Julia Koval <julia.koval@intel.com>
5098 * config/i386/driver-i386.c (host_detect_local_cpu):
5099 Detect skylake-avx512.
5101 2017-11-26 Julia Koval <julia.koval@intel.com>
5103 * config.gcc: Add -march=cannonlake.
5104 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5105 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5106 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5107 (PTA_CANNONLAKE): New.
5108 (processor_target_table): Add cannonlake.
5109 (ix86_option_override_internal): Ditto.
5110 (fold_builtin_cpu): Ditto.
5111 (get_builtin_code_for_version): Handle cannonlake.
5112 (M_INTEL_COREI7_CANNONLAKE): New.
5113 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5114 * doc/invoke.texi: Add -march=cannonlake.
5116 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5118 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5119 (try_init_one_plugin): Alternative implementation for MinGW.
5120 * Makefile.in (plugin_implib): New.
5121 (gengtype-lex.c): Fix broken AIX workaround.
5122 * configure: Regenerate.
5123 * doc/plugins.texi: Document support for MinGW.
5125 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5127 PR rtl-optimization/81553
5128 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5129 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5130 is a shift where C1 has different mode than the whole shift, use C1's
5131 mode for MULT rather than the shift's mode.
5134 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5135 builtins not enabled in the currently selected ISA.
5137 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5139 PR tree-optimization/71026
5140 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5141 (insert_reciprocals): Change to insert reciprocals before a division
5142 by a square and to insert the square of a reciprocal.
5143 (execute_cse_reciprocals_1): Change to consider division by a square.
5144 (register_division_in): Add importance parameter.
5146 2017-11-24 Richard Biener <rguenther@suse.de>
5148 PR tree-optimization/82402
5149 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5150 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5152 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5154 * match.pd (0-ptr): New transformation.
5156 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5159 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5160 when skipping checking.
5162 2017-11-24 Richard Biener <rguenther@suse.de>
5164 PR tree-optimization/83128
5165 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5166 (vn_reference_lookup_3): Likewise.
5168 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5171 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5172 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5175 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5176 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5177 check for the latter with gimple_call_builtin_p. Do not handle
5178 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5180 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5182 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5185 2017-11-23 Julia Koval <julia.koval@intel.com>
5187 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5188 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5189 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5190 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5191 _mm512_maskz_expandloadu_epi16): New intrinsics.
5192 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5193 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5194 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5195 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5196 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5197 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5198 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5199 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5200 _mm256_maskz_expandloadu_epi8): New intrinsics.
5201 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5202 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5203 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5204 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5205 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5206 * config/i386/sse.md (VI248_VLBW): New iterator.
5207 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5209 2017-11-23 Julia Koval <julia.koval@intel.com>
5211 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5212 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5213 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5214 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5215 _mm512_mask_compressstoreu_epi16): New.
5216 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5217 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5218 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5219 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5220 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5221 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5222 _mm256_mask_compressstoreu_epi8): New.
5223 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5224 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5225 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5226 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5227 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5228 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5229 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5230 __builtin_ia32_compresshi128_mask,
5231 __builtin_ia32_compressstoreuqi512_mask,
5232 __builtin_ia32_compressstoreuhi512_mask,
5233 __builtin_ia32_compressstoreuqi256_mask,
5234 __builtin_ia32_compressstoreuqi128_mask,
5235 __builtin_ia32_compressstoreuhi256_mask,
5236 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5237 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5239 (ix86_expand_special_args_builtin): Handle new types.
5240 (s4fma_expand): Handle new builtin array.
5241 * config/i386/immintrin.h: Include new headers.
5242 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5243 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5245 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5247 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5248 fixes. Declare temp and g variables at the top in order to avoid
5249 {} in most of the cases.
5251 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5253 * match.pd (ptr-0): New transformation.
5255 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5257 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5258 (struct addr_mode_cost_table): New.
5259 (struct tune_params): Add field addr_mode_costs.
5260 * config/arm/arm.c (generic_addr_mode_costs): New.
5261 (arm_slowmul_tune): Initialise addr_mode_costs field.
5262 (arm_fastmul_tune): Likewise.
5263 (arm_strongarm_tune): Likewise.
5264 (arm_xscale_tune): Likewise.
5265 (arm_9e_tune): Likewise.
5266 (arm_marvell_pj4_tune): Likewise.
5267 (arm_v6t2_tune): Likewise.
5268 (arm_cortex_tune): Likewise.
5269 (arm_cortex_a8_tune): Likewise.
5270 (arm_cortex_a7_tune): Likewise.
5271 (arm_cortex_a15_tune): Likewise.
5272 (arm_cortex_a35_tune): Likewise.
5273 (arm_cortex_a53_tune): Likewise.
5274 (arm_cortex_a57_tune): Likewise.
5275 (arm_exynosm1_tune): Likewise.
5276 (arm_xgene1_tune): Likewise.
5277 (arm_cortex_a5_tune): Likewise.
5278 (arm_cortex_a9_tune): Likewise.
5279 (arm_cortex_a12_tune): Likewise.
5280 (arm_cortex_a73_tune): Likewise.
5281 (arm_v7m_tune): Likewise.
5282 (arm_cortex_m7_tune): Likewise.
5283 (arm_v6m_tune): Likewise.
5284 (arm_fa726te_tune): Likewise.
5285 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5287 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5289 * config/arm/arm.c (arm_mem_costs): New function.
5290 (arm_rtx_costs_internal): Use arm_mem_costs.
5292 2017-11-23 Mark Wielaard <mark@klomp.org>
5294 * dwarf2out.c (init_sections_and_labels): Use generation to create
5295 unique ranges_section_label and ranges_base_label. Return generation.
5296 (output_rnglists): Add generation argument. Use generation to create
5297 unique ranges labels.
5298 (dwarf2out_finish): Get generation from init_sections_and_labels
5299 and pass generation to output_rnglists.
5301 2017-11-23 Mike Stump <mikestump@comcast.net>
5302 Eric Botcazou <ebotcazou@adacore.com>
5304 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5305 * cfgloop.h (struct loop): Add unroll field.
5306 * function.h (struct function): Add has_unroll bitfield.
5307 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5308 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5309 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5310 (pass_rtl_unroll_loops::gate): Likewise.
5311 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5312 for which loop->unroll==1.
5313 (decide_unroll_constant_iterations): Use note for consistency and
5314 take loop->unroll into account. Return early if loop->unroll is set.
5315 Fix thinko in existing test.
5316 (decide_unroll_runtime_iterations): Use note for consistency and
5317 take loop->unroll into account.
5318 (decide_unroll_stupid): Likewise.
5319 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5320 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5321 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5323 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5324 (print_loop): Print loop->unroll if set.
5325 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5326 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5327 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5329 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5330 loop->unroll is set and smaller than the trip count. Otherwise bypass
5331 entirely the heuristics if loop->unroll is set. Remove dead note.
5332 Fix off-by-one bug in other note.
5333 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5334 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5336 (tree_unroll_loops_completely): Make static.
5337 (pass_complete_unroll::execute): Use correct type for variable.
5338 (pass_complete_unrolli::execute): Fix formatting.
5339 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5341 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5343 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5344 enable when TARGET_PREFER_AVX128 is set.
5346 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5348 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5350 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5351 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5353 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5355 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5356 frequencies. Fix estimation of aggregate parameters.
5358 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5360 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5363 2017-11-23 Tom de Vries <tom@codesourcery.com>
5365 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5367 2017-11-23 Tom de Vries <tom@codesourcery.com>
5369 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5372 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5375 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5376 sibcall_value_pcrel_fdpic): Use local variable instead of
5378 (calli_tbr_rel): Add missing operand 2.
5379 (call_valuei_tbr_rel): Add missing operand 3.
5381 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5384 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5385 bitpos/bitsize covering the whole destination, use store_expr only if
5386 the complex mode is the same. Otherwise, use expand_normal and if
5387 it returns CONCAT, subreg each part separately instead of trying to
5388 subreg the whole result.
5390 2017-11-23 Richard Biener <rguenther@suse.de>
5392 PR tree-optimization/23094
5393 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5394 come here from walking over backedges in the first iteration.
5395 (vn_reference_lookup_3): Skip clobbers that store the same value.
5397 2017-11-23 Richard Biener <rguenther@suse.de>
5399 PR tree-optimization/81403
5400 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5401 a block we need a leader relative to.
5402 (phi_translate_1): For nary processing require a leader from
5403 get_representative_for given we run expression simplification
5404 using match-and-simplify. Remove previous fix.
5406 2017-11-22 Jeff Law <law@redhat.com>
5408 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5409 Use new method allocate_value_range rather than accessing the
5410 vrp_value_range_pool data member directly.
5411 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5412 to use extract_range_from_stmt method to avoid need for
5413 extract_range_from_assignment method.
5414 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5415 method rather than setting values_propgated data member directly.
5416 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5417 and values propagated data members and extract_range_from_assignment
5418 method. Reorder private data members to conform to standards.
5419 Add new methods set_lattice_propagation_complete and
5420 allocate_value_range.
5422 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5424 PR rtl-optimization/83030
5425 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5426 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5427 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5428 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5429 CROSSING_JUMP_P flag is already set before setting it.
5430 * cfgrtl.c (fixup_partition_crossing): Likewise.
5431 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5434 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5436 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5437 Handle the case where both arguments are using gen_const_vec_series.
5439 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5442 * pretty-print.c (pp_format): Move quoting implementation to
5443 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5444 to pass address of "quote" local.
5445 (pp_begin_quote): New function.
5446 (pp_end_quote): New function.
5447 * pretty-print.h (printer_fn): Convert penultimate param from bool
5449 (pp_begin_quote): New decl.
5450 (pp_end_quote): New decl.
5451 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5452 param from bool to bool *.
5453 * tree-diagnostic.h (default_tree_printer): Likewise.
5455 2017-11-22 Jeff Law <law@redhat.com>
5457 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5458 Initialize vr_values.
5459 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5460 vr_values via vr_values class instance rather than delegators.
5461 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5462 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5463 (evrp_range_analyzer::push_value_range): Likewise.
5464 (evrp_range_analyzer::pop_value_range): Likewise.
5465 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5466 most delegators. Those remaining are exposed as public interfaces.
5467 Make vr_values a pointer and private.
5468 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5470 (evrp_range_analyzer::get_vr_value): New method.
5471 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5472 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5473 (class evrp_dom_walker): Attach evrp_folder class, initialize
5474 it in the ctor. Remove temporary delegators.
5475 (evrp_dom_walker::before_dom_children): Call methods in attached
5476 evrp_range_analyzer class via class instance pointer. Use
5477 free value_range_constant_singleton to remove need for
5478 op_with_constant_singleton_value delegator method. Do not
5479 create a vrp_prop class instance for every call! Narrow
5480 scope of a couple variables.
5481 (evrp_dom_walker::cleanup): Call methods in attached
5482 evrp_range_analyzer class via class instance pointer.
5483 * vr-values.h (class vr_values): Privatize many methods and
5486 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5488 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5489 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5490 * expr.c (expand_expr_real_2): Likewise.
5491 * fold-const.c (const_unop): Likewise.
5492 * optabs-tree.c (optab_for_tree_code): Likewise.
5493 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5494 * tree-inline.c (estimate_operator_cost): Likewise.
5495 * tree-pretty-print.c (dump_generic_node): Likewise.
5496 (op_code_prio): Likewise.
5497 (op_symbol_code): Likewise.
5498 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5499 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5500 * internal-fn.c (direct_internal_fn_optab): New function.
5501 (direct_internal_fn_array, direct_internal_fn_supported_p
5502 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5503 * fold-const-call.c (fold_const_reduction): New function.
5504 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5506 * tree-vect-loop.c: Include internal-fn.h.
5507 (reduction_code_for_scalar_code): Rename to...
5508 (reduction_fn_for_scalar_code): ...this and return an internal
5510 (vect_model_reduction_cost): Take an internal_fn rather than
5512 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5514 (vectorizable_reduction): Use internal functions rather than tree
5515 codes for the reduction operation. Update calls to the functions
5517 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5518 Use calls to internal functions rather than REDUC tree codes.
5519 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5521 2017-11-22 Olivier Hainque <hainque@adacore.com>
5523 * config/vxworks.c (vxworks_override_options): Pick default
5524 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5525 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5526 DWARF_GNAT_ENCODINGS_DEFAULT.
5527 * config/vxworksae.h: Likewise.
5529 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5531 PR tree-optimization/83104
5532 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5535 2017-11-22 Marek Polacek <polacek@redhat.com>
5536 H.J. Lu <hongjiu.lu@intel.com>
5537 Jason Merrill <jason@redhat.com>
5542 * calls.c (initialize_argument_information): Call
5543 warn_parameter_passing_abi target hook.
5544 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5545 argument onto stack.
5546 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5547 * common.opt: Update -fabi-version description.
5548 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5549 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5551 (ix86_is_empty_record): New function.
5552 (ix86_warn_parameter_passing_abi): New function.
5553 (TARGET_EMPTY_RECORD_P): Redefine.
5554 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5555 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5556 * doc/tm.texi: Regenerated.
5557 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5558 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5559 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5560 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5561 instead of int_size_in_bytes.
5562 * expr.c (copy_blkmode_to_reg): Likewise.
5563 * function.c (aggregate_value_p): Return 0 for empty types.
5564 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5566 (locate_and_pad_parm): Call arg size_in_bytes instead
5568 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5569 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5570 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5572 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5573 (std_gimplify_va_arg_expr): Skip empty records. Call
5574 arg_size_in_bytes instead size_in_bytes.
5575 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5576 * tree-core.h (tree_type_common): Add empty_flag.
5577 (tree_decl_common): Update comments.
5578 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5580 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5581 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5583 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5584 * tree.c (default_is_empty_type): New function.
5585 (default_is_empty_record): New function.
5586 (arg_int_size_in_bytes): New function.
5587 (arg_size_in_bytes): New function.
5588 (get_ultimate_context): New function.
5589 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5590 TRANSLATION_UNIT_WARN_EMPTY_P.
5591 (default_is_empty_record, arg_int_size_in_bytes,
5592 arg_size_in_bytes, get_ultimate_context): Declare.
5594 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5596 * config/arm/arm.c (cmse_clear_registers): New function.
5597 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5598 code by call to cmse_clear_registers.
5599 (cmse_nonsecure_entry_clear_before_return): Likewise.
5601 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5603 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5604 (vdot_s32, vdotq_s32): New.
5605 (vdot_lane_u32, vdotq_lane_u32): New.
5606 (vdot_lane_s32, vdotq_lane_s32): New.
5609 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5612 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5613 for unsigned values with fewer HWIs than the precision.
5614 (test_overflow): New function.
5615 (wide_int_cc_tests): Call it.
5617 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5619 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5620 has the same mode class as Pmode.
5622 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5624 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5625 padding_bits_to_clear_ptr.
5626 (cmse_nonsecure_entry_clear_before_return): Likewise.
5628 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5630 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5631 auto_sbitap instead of integer bitfield to control register needing
5634 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5636 PR tree-optimization/83044
5637 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5638 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5639 that need the upper bound. Subtract offset from
5640 get_addr_base_and_unit_offset only if positive and subtract it
5641 before division by eltsize rather than after it.
5644 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5645 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5646 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5650 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5651 expand_binop with *mul_widen_optab, make sure at least one of the
5652 operands doesn't have VOIDmode.
5655 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5657 PR rtl-optimization/82044
5658 PR tree-optimization/82042
5659 * dse.c (record_store): Check for overflow.
5660 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5661 add_wild_read instead of clear_rhs_from_active_local_stores on
5662 overflow and log it into dump_file.
5664 2017-11-22 Richard Biener <rguenther@suse.de>
5666 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5667 fake edges to exit when looking for a place to insert.
5668 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5670 (insert_into_preds_of_block): Commit edge insertion immediately,
5671 assert that doesn't require new BBs.
5672 (fini_pre): Release expressions.
5673 (pass_pre::execute): Shuffle things around a bit, if the fn
5674 is too large do not compute AVAIL either as this is really the
5677 2017-11-22 Richard Biener <rguenther@suse.de>
5679 PR tree-optimization/83089
5680 * tree-if-conv.c (pass_if_conversion::execute): If anything
5681 changed reset SCEV and free the number of iteration estimates.
5683 2017-11-21 Martin Sebor <msebor@redhat.com>
5685 PR tree-optimization/82945
5686 * calls.h (warn_nonstring_bound): Remove unused function.
5688 2017-11-21 Martin Sebor <msebor@redhat.com>
5690 PR tree-optimization/82945
5691 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5692 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5693 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5695 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5696 * calls.h (get_attr_nonstring_decl): Declare new function.
5697 * doc/extend.texi (attribute nonstring): Update.
5698 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5699 get_attr_nonstring_decl and handle it.
5700 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5701 detection of nul-termination.
5702 (strlen_to_stridx): Change to a pointer.
5703 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5704 (pass_strlen::execute): Same.
5706 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5708 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5709 for the new option -mprefer-vector-width=[none|128|256|512].
5710 * config/i386/i386.c (ix86_target_string): remove old style options
5711 -mprefer-avx256 and make -mprefer-avx128 as alias.
5712 (ix86_option_override_internal): Apply defaults for the
5713 -mprefer-vector-width=[128|256] option.
5714 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5715 Implement macros to work with -mprefer-vector-width=.
5716 * config/i386/i386.opt: Implemented option
5717 -mprefer-vector-width=[none|128|256|512].
5718 * doc/invoke.texi: Documentation for
5719 -mprefer-vector-width=[none|128|256|512].
5721 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5723 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5725 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5727 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5729 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5731 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5733 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5734 * expr.c (expand_expr_real_2): Likewise.
5735 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5736 fold_binary_loc): Likewise.
5737 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5738 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5739 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5740 MINUS_EXPR transformations.
5741 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5742 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5743 * tree-inline.c (estimate_operator_cost): Likewise.
5744 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5745 op_symbol_code): Likewise.
5746 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5747 * vr-values.c (extract_range_from_binary_expr): Likewise.
5748 * varasm.c (initializer_constant_valid_p_1): Likewise.
5749 * tree.def: New tree code POINTER_DIFF_EXPR.
5751 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5753 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5754 integer suffix to movbe mnemonic.
5755 (*bswaphi2_movbe): Ditto.
5756 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5758 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5761 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5762 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5763 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5765 2017-11-21 Martin Liska <mliska@suse.cz>
5767 * tree-inline.c (expand_call_inline): Remove not needed
5770 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5771 Jakub Jelinek <jakub@redhat.com>
5774 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5775 Remove static keyword from f variable.
5777 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5779 PR tree-optimization/83086
5780 * gimple-ssa-store-merging.c
5781 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5782 rather than n.base_addr.
5784 2017-11-21 Martin Liska <mliska@suse.cz>
5786 PR rtl-optimization/82044
5787 PR tree-optimization/82042
5788 * dse.c (check_mem_read_rtx): Check for overflow.
5790 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5792 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5795 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5797 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5798 and its corresponding call arg location note.
5800 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5801 Andrew Burgess <andrew.burgess@embecosm.com>
5803 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5805 (arc_return_slot_offset): Likewise.
5806 (arc_eh_return_address_location): New declaration.
5807 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5808 (MUST_SAVE_REGISTER): Add exception handler case.
5809 (MUST_SAVE_RETURN_ADDR): Likewise.
5810 (arc_frame_pointer_required): Likewise.
5811 (arc_frame_pointer_needed): New function.
5812 (arc_compute_frame_size): Changed.
5813 (arc_expand_prologue): Likewise.
5814 (arc_expand_epilogue): Likewise.
5815 (arc_initial_elimination_offset): Likewise.
5816 (arc_return_slot_offset): Delete.
5817 (arc_eh_return_address_location): New function.
5818 (arc_builtin_setjmp_frame_value): Likewise.
5819 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5820 (EH_RETURN_STACKADJ_RTX): Define.
5821 (EH_RETURN_HANDLER_RTX): Likewise.
5822 * config/arc/arc.md (eh_return): Delete.
5824 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5826 * print-tree.h (debug_vec_tree): Remove prototype.
5827 * gdbinit.in (pvt): Remove macro.
5829 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5831 PR tree-optimization/83047
5832 * gimple-ssa-store-merging.c
5833 (imm_store_chain_info::output_merged_store): If the loads with the
5834 same vuse are in different basic blocks, for load_gsi pick a load
5835 location that is dominated by the other loads.
5838 * config/i386/i386.c (ix86_memmodel_check): Start
5839 -Winvalid-memory-model diagnostics with lowercase letter.
5842 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5843 set in_cold_section_p to the partition containing loc_list->first.
5844 When seeing loc_list->last_before_switch node, update secname and
5845 perform range_across_switch second partition handling only after that.
5848 * run-rtl-passes.c: Include debug.h.
5849 (run_rtl_passes): Call debug_hooks->assembly_start.
5850 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5854 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5855 OPTAB_DIRECT in calls to expand_simple_binop.
5857 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5860 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5862 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5865 * spellcheck.h (best_match::blithely_get_best_candidate): New
5868 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5870 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5871 with lower case letter.
5873 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5875 * config/i386/i386.md (bswaphi2): New expander.
5876 (*bswaphi2_movbe): New insn pattern.
5877 (bswaphi -> rorhi pepehole2): New peephole pattern.
5879 2017-11-20 Jeff Law <law@redhat.com>
5881 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5882 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5883 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5884 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5885 gimple-ssa-evrp-analyze.h.
5887 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5888 set BB_VISITED here.
5889 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5891 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5893 * tree-predcom.c: Add general comment on Store-Store chains.
5894 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5896 (get_chain_last_ref_at): Rename into...
5897 (get_chain_last_write_at): ...this.
5898 (get_chain_last_write_before_load): New function.
5899 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5900 CT_STORE_STORE when write reference is added.
5901 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5902 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5903 (initialize_root_vars_store_elim_1): Ditto.
5904 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5905 definition is created.
5906 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5907 chain by replacing it with dominant stored value.
5909 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5911 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5913 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5915 * vr-values.c (extract_range_from_binary_expr): Use a full range
5918 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5920 * config/arm/arm.md (R4_REGNUM): Define constant.
5921 (nonsecure_call_internal): Remove r4 clobber.
5922 (nonsecure_call_value_internal): Likewise.
5923 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5924 clobber and resequence match_operands.
5925 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5926 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5927 (nonsecure_call_value_reg_thumb2): Likewise.
5929 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5931 PR tree-optimization/78821
5932 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5933 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5934 to the address of the base rather than the base itself.
5935 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5936 (find_bswap_or_nop_finalize): New function.
5937 (find_bswap_or_nop): Use it.
5938 (bswap_replace): Return a tree rather than bool, change first
5939 argument from gimple * to gimple_stmt_iterator, allow inserting
5940 into an empty sequence, allow ins_stmt to be NULL - then emit
5941 all stmts into gsi. Fix up MEM_REF address gimplification.
5942 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5943 (struct store_immediate_info): Add N and INS_STMT non-static
5945 (store_immediate_info::store_immediate_info): Initialize them
5946 from newly added ctor args.
5947 (merged_store_group::apply_stores): Formatting fixes. Sort by
5949 (stmts_may_clobber_ref_p): For stores call also
5950 refs_anti_dependent_p.
5951 (gather_bswap_load_refs): New function.
5952 (imm_store_chain_info::try_coalesce_bswap): New method.
5953 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5954 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5955 (imm_store_chain_info::output_merged_store): Fail if number of
5956 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5957 and NOP_EXPR rhs_code.
5958 (pass_store_merging::process_store): Compute n and ins_stmt, if
5959 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5960 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5962 (pass_store_merging::execute): Calculate dominators.
5964 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5965 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5966 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5967 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5968 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5969 class pass_optimize_bswap, bswap_replace,
5970 pass_optimize_bswap::execute): Moved to ...
5971 * gimple-ssa-store-merging.c: ... this file.
5972 Include optabs-tree.h.
5973 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5974 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5975 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5976 anonymous namespace, remove static keywords.
5977 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5978 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5980 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5983 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5985 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5987 * vec.h (debug_helper): New function.
5988 (DEFINE_DEBUG_VEC): New macro.
5989 * hash-set.h (debug_helper): New function.
5990 (DEFINE_DEBUG_HASH_SET): New macro.
5991 * cfg.c (debug_slim (edge)): New function.
5992 Call DEFINE_DEBUG_VEC for edges.
5993 Call DEFINE_DEBUG_HASH_SET for edges.
5994 * cfghooks.c (debug_slim (basic_block)): New function.
5995 Call DEFINE_DEBUG_VEC for basic blocks.
5996 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5997 * print-tree.c (debug_slim): New function to handle trees.
5998 Call DEFINE_DEBUG_VEC for trees.
5999 Call DEFINE_DEBUG_HASH_SET for trees.
6000 (debug (vec<tree, va_gc>) &): Remove.
6001 (debug (<vec<tree, va_gc>) *): Remove.
6002 * print-rtl.c (debug_slim): New function to handle const_rtx.
6003 Call DEFINE_DEBUG_VEC for rtx_def.
6004 Call DEFINE_DEBUG_VEC for rtx_insn.
6005 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6006 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6007 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6008 (debug (vec<rtx_insn *> *ptr): Remove.
6009 (debug_insn_vector): Remove.
6010 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6012 2017-11-20 Tom de Vries <tom@codesourcery.com>
6014 PR rtl-optimization/82020
6015 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6016 IF_THEN_ELSE condition.
6018 2017-11-19 Jeff Law <law@redhat.com>
6020 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6021 of degenerates resulting from ignoring an edge.
6023 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6026 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6028 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6031 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6032 for uninitialied counts.
6034 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6037 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6039 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6042 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6044 2017-11-19 Tom de Vries <tom@codesourcery.com>
6046 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6049 2017-11-19 Tom de Vries <tom@codesourcery.com>
6051 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6054 2017-11-19 Tom de Vries <tom@codesourcery.com>
6056 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6059 2017-11-19 Tom de Vries <tom@codesourcery.com>
6061 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6063 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6065 2017-11-19 Tom de Vries <tom@codesourcery.com>
6067 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6069 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6070 semicolon after MCORE_EXPORT_NAME call.
6072 2017-11-19 Tom de Vries <tom@codesourcery.com>
6075 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6076 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6078 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6080 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6081 (cgraph_edge::clone): Cleanup updating of profile.
6082 * ipa-cp.c (update_profiling_info): Likewise.
6083 * ipa-inline-transform.c (inline_transform): Likewise.
6084 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6085 * ipa-split.c (execute_split_functions): Do not split when function
6087 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6088 * profile-count.c (profile_count::dump): Add global0.
6089 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6091 (profile_count::to_sreal_scale): Likewise.
6092 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6093 (profile_count::combine_with_ipa_count): New function.
6094 * profile-count.h (profile_guessed_global0adjusted): New.
6095 (profile_count::adjusted_zero): New.
6096 (profile_count::global0adjusted): New.
6097 (profile_count::combine_with_ipa_count): New.
6098 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6099 correct profile of return block of split functions.
6100 (copy_cfg_body): Remove unused profile_count.
6101 (copy_body): Likewise.
6102 (expand_call_inline): Update.
6103 (tree_function_versioning): Update.
6105 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6107 * hash-set.h (hash_set::empty): New.
6108 * tree-ssa-threadbackward.h: Delete.
6109 * tree-ssa-threadbackward.c (class thread_jumps): New.
6110 Move max_threaded_paths into class.
6111 (fsm_find_thread_path): Remove arguments that are now in class.
6112 (profitable_jump_thread_path): Rename to...
6113 (thread_jumps::profitable_jump_thread_path): ...this.
6114 (convert_and_register_jump_thread_path): Rename to...
6115 (thread_jumps::convert_and_register_current_path): ...this.
6116 (check_subpath_and_update_thread_path): Rename to...
6117 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6118 (register_jump_thread_path_if_profitable): Rename to...
6119 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6120 (handle_phi): Rename to...
6121 (thread_jumps::handle_phi): ...this.
6122 (handle_assignment): Rename to...
6123 (thread_jumps::handle_assignment): ...this.
6124 (fsm_find_control_statement_thread_paths): Rename to...
6125 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6126 (find_jump_threads_backwards): Rename to...
6127 (thread_jumps::find_jump_threads_backwards): ...this.
6128 Initialize path local data.
6129 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6130 from within thread_jumps class.
6131 (pass_early_thread_jumps::execute): Same.
6133 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6135 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6137 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6139 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6140 to canonical location.
6142 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6145 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6147 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6149 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6151 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6152 rcrt1.o%s/grcrt1.o%s for -static-pie.
6154 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6156 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6157 ix86_shift_rotate_cost): Break out from ...
6158 (ix86_rtx_costs): ... here.
6159 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6162 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6164 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6165 when entry block was promoted unlikely.
6166 (estimate_bb_frequencies): Increase frequency scale.
6167 * profile-count.h (profile_count): Export precision info.
6169 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6171 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6172 disturbin profile of entry block.
6174 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6176 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6179 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6181 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6182 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6184 2017-11-17 Jeff Law <law@redhat.com>
6186 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6187 from evrp_dom_walker class. Various methods moved into new class.
6188 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6189 (evrp_range_analyzer::enter): New method.
6190 (evrp_range_analyzer::leave): New method.
6191 (evrp_dom_walker): Remove delegators no longer needed by this class.
6192 Replace vr_values data member with evrp_range_analyzer
6194 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6195 method extracted from evrp_dom_walker::before_dom_children.
6196 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6197 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6199 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6200 Add private copy constructor and move assignment operators.
6201 Privatize methods and class data where trivially possible.
6202 (evrp_dom_walker::cleanup): New function, extracted from
6203 execute_early_vrp. Simplify access to class data.
6205 * vr-values.h (get_output_for_vrp): Prototype.
6206 * vr-values.c (get_output_for_vrp): New function extracted from
6207 vrp_visit_assignment_or_call and extract_range_from_stmt.
6208 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6210 2017-11-17 Luis Machado <luis.machado@linaro.org>
6212 * config/aarch64/aarch64.c
6213 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6214 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6215 tune_params::AUTOPREFETCHER_WEAK.
6217 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6220 * config/arm/arm.c (arm_valid_target_attribute_rec):
6221 Parse "arch=" and "+<ext>".
6222 (arm_valid_target_attribute_tree): Re-init global options.
6223 (arm_option_override): Make non-static.
6224 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6225 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6226 (__ARM_FEATURE_CRC32): Support undef.
6227 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6228 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6230 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6232 * gdbinit.in (break-on-diagnostic): New command.
6234 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6236 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6237 return to indirect jump for EH return if control-flow
6238 protection is enabled. Change explicit 'false' argument in
6239 pro_epilogue_adjust_stack with a value of
6241 * config/i386/i386.md (simple_return_indirect_internal):
6242 Remove SImode restriction to support 64-bit.
6244 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6246 * combine.c (added_notes_insn): New.
6247 (try_combine): Handle added_notes_insn like added_links_insn.
6248 Rewrite return value code.
6249 (distribute_notes): Set added_notes_insn to the earliest insn we added
6252 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6254 PR rtl-optimization/82621
6255 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6256 dest of one of those SETs is unused.
6258 2017-11-17 Richard Biener <rguenther@suse.de>
6261 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6262 * tree-pretty-print.c (dump_generic_node): Handle
6263 annot_expr_parallel_kind.
6264 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6265 * gimplify.c (gimple_boolify): Likewise.
6267 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6269 * config.gcc (extra_headers): Add cet.h for x86 targets.
6270 * config/i386/cet.h: New file.
6271 * doc/install.texi: Add --enable-cet/--disable-cet.
6273 2017-11-17 Richard Biener <rguenther@suse.de>
6275 PR tree-optimization/83017
6276 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6277 (gen_parallel_loop): Properly count iterations.
6278 (parallelize_loops): Handle loop->can_be_parallel independent
6279 of flag_loop_parallelize_all. Make static profitability test match
6281 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6282 * invoke.texi (parloops-min-per-thread): Document.
6284 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6286 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6287 upstreaming review comments.
6289 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6291 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6292 with fast unaligned access.
6293 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6295 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6297 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6298 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6300 2017-11-17 Richard Biener <rguenther@suse.de>
6302 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6303 folding of references.
6305 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6308 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6309 of replacing call to strncmp with corresponding call to strcmp when
6312 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6314 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6315 option prefer-avx256 for skylake-avx512 configuration.
6316 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6317 (get_builtin_code_for_version): Ditto.
6319 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6320 Monk Chiang <sh.chiang04@gmail.com>
6322 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6323 (FIXED_REGISTERS): Reserve more register numbers.
6324 (CALL_USED_REGISTERS): Likewise.
6325 (REG_ALLOC_ORDER): Likewise.
6326 (REG_CLASS_CONTENTS): Likewise.
6327 (REGISTER_NAMES): Likewise.
6329 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6330 Kito Cheng <kito.cheng@gmail.com>
6332 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6334 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6336 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6338 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6339 parameter list for vec_splats.
6341 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6343 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6345 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6347 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6350 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6351 with the required version of automake 1.11.6.
6353 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6355 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6357 (DU_C2_3_power9): Correct reservation combinations.
6358 (FP_DIV_power9, VEC_DIV_power9): New.
6359 (power9-alu): Split out rotate/shift...
6360 (power9-rot): ...to here, correct dispatch resource.
6361 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6363 (power9-fp): Correct latency.
6364 (power9-sdiv): Add div/sqrt resource.
6365 (power9-ddiv): Correct latency, add div/sqrt resource.
6366 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6367 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6369 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6371 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6373 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6374 switch statement mapping KF built-ins to TF built-ins if we don't
6375 have the proper ISA 3.0 assembler support.
6377 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6379 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6380 (lower_emutls_function_body): Do not compute it.
6382 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6384 * ipa-split.c (split_bb_info): Turn time to sreal.
6385 (split_point): Likewise.
6386 (dump_split_point): Likewise.
6387 (fine_split_points): Likewise.
6388 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6390 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6392 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6395 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6397 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6399 (expensive_function_p): Remove useless assert.
6400 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6402 2017-11-16 Martin Sebor <msebor@redhat.com>
6404 PR tree-optimization/82588
6405 PR tree-optimization/82583
6406 * tree-vrp.c (check_array_ref): Handle flexible array members,
6407 string literals, and inner indices.
6408 (search_for_addr_array): Add detail to diagnostics.
6410 2017-11-16 Nathan Sidwell <nathan@acm.org>
6414 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6415 (SET_DECL_ASSEMBLER_NAME): Forward to
6416 overwrite_decl_assembler_name.
6417 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6418 (overwrite_decl_assembler_name): Declare.
6419 * tree.c (overwrite_decl_assembler_name): New.
6420 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6421 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6422 (LANG_HOOKS_INITIALIZER): Add it.
6423 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6424 * langhooks.c (lhd_set_decl_assembler_name): Use
6425 SET_DECL_ASSEMBLER_NAME.
6426 (lhd_overwrite_decl_assembler_name): Default implementation.
6428 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6429 Jackson Woodruff <jackson.woodruff@arm.com>
6431 PR tree-optimization/71026
6432 * match.pd: Canonicalize constant multiplies in division.
6434 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6436 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6437 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6438 Remove OPT_fomit_frame_pointer entry.
6439 * common/config/arc/arc-common.c: Likewise.
6440 * common/config/arm/arm-common.c: Likewise.
6441 * common/config/avr/avr-common.c: Likewise.
6442 * common/config/c6x/c6x-common.c: Likewise.
6443 * common/config/cr16/cr16-common.c: Likewise.
6444 * common/config/cris/cris-common.c: Likewise.
6445 * common/config/epiphany/epiphany-common.c: Likewise.
6446 * common/config/fr30/fr30-common.c: Likewise.
6447 * common/config/frv/frv-common.c: Likewise.
6448 * common/config/ia64/ia64-common.c: Likewise.
6449 * common/config/iq2000/iq2000-common.c: Likewise.
6450 * common/config/lm32/lm32-common.c: Likewise.
6451 * common/config/m32r/m32r-common.c: Likewise.
6452 * common/config/mcore/mcore-common.c: Likewise.
6453 * common/config/microblaze/microblaze-common.c: Likewise.
6454 * common/config/mips/mips-common.c: Likewise.
6455 * common/config/mmix/mmix-common.c: Likewise.
6456 * common/config/mn10300/mn10300-common.c: Likewise.
6457 * common/config/nios2/nios2-common.c: Likewise.
6458 * common/config/pa/pa-common.c: Likewise.
6459 * common/config/pdp11/pdp11-common.c: Likewise.
6460 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6461 * common/config/riscv/riscv-common.c: Likewise.
6462 * common/config/rs6000/rs6000-common.c: Likewise.
6463 * common/config/rx/rx-common.c: Likewise.
6464 * common/config/s390/s390-common.c: Likewise.
6465 * common/config/sh/sh-common.c: Likewise.
6466 * common/config/sparc/sparc-common.c: Likewise.
6467 * common/config/tilegx/tilegx-common.c: Likewise.
6468 * common/config/tilepro/tilepro-common.c: Likewise.
6469 * common/config/v850/v850-common.c: Likewise.
6470 * common/config/visium/visium-common.c: Likewise.
6471 * common/config/xstormy16/xstormy16-common.c: Likewise.
6472 * common/config/xtensa/xtensa-common.c: Likewise.
6473 * invoke.texi (-fomit-frame-pointer): Update documentation.
6475 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6477 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6479 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6481 * cfg.c (scale_bbs_frequencies_int,
6482 cale_bbs_frequencies_gcov_type): Remove.
6483 * cfg.h (scale_bbs_frequencies_int,
6484 cale_bbs_frequencies_gcov_type): Remove.
6486 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6488 * tree-ssa-loop-manip.c
6489 (scale_dominated_blocks_in_loop): Update to profile counts.
6490 (tree_transform_and_unroll_loop): Likewise.
6492 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6494 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6495 scale_bbs_frequencies_int.
6497 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6499 * final.c (compute_alignments): Use counts rather than frequencies.
6501 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6503 * cfgloopanal.c: Include sreal.h
6504 (average_num_loop_insns): Use counts and sreal for accounting.
6506 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6508 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6511 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6512 Kito Cheng <kito.cheng@gmail.com>
6514 * config/nds32/constraints.md: Provide more constraints.
6515 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6516 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6517 support constraints usage.
6519 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6521 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6523 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6524 Kito Cheng <kito.cheng@gmail.com>
6526 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6527 * config/nds32/nds32.opt: Refine the layout.
6528 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6529 TARGET_EXT_STRING): Support new options.
6530 * config/nds32/nds32.h: Likewise.
6531 * config/nds32/nds32.md: Likewise.
6532 * config/nds32/nds32-predicates.c: Likewise.
6533 * config/nds32/constraints.md: Likewise.
6534 * common/config/nds32/nds32-common.c: Likewise.
6536 2017-11-16 Julia Koval <julia.koval@intel.com>
6539 * config/i386/gfniintrin.h: Add sse check.
6540 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6542 2017-11-16 Julia Koval <julia.koval@intel.com>
6544 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6545 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6546 (ix86_handle_option): Handle -mavx512vbmi2.
6547 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6548 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6549 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6550 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6551 (ix86_valid_target_attribute_inner_p): Ditto.
6552 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6553 * config/i386/i386.opt (mavx512vbmi2): New option.
6554 * doc/invoke.texi: Add new option.
6556 2017-11-16 Julia Koval <julia.koval@intel.com>
6558 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6559 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6560 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6561 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6562 _mm512_gf2p8mul_epi8): New intrinsics.
6563 * config/i386/i386-builtin-types.def
6564 (V64QI_FTYPE_V64QI_V64QI): New type.
6565 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6566 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6567 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6568 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6569 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6570 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6572 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6574 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6575 explicitly as a stream of bytes.
6577 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6579 * config/rs6000/altivec.h (vec_xst_be): New #define.
6580 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6581 and externalize from *altivec_vperm_<mode>_internal.
6582 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6584 (XL_BE_V8HI): Likewise.
6585 (XL_BE_V4SI): Likewise.
6586 (XL_BE_V4SI): Likewise.
6587 (XL_BE_V2DI): Likewise.
6588 (XL_BE_V4SF): Likewise.
6589 (XL_BE_V2DF): Likewise.
6590 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6591 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6592 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6593 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6594 VSX_BUILTIN_VEC_XST_BE.
6595 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6596 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6598 (altivec_init_builtins): Replace conditional calls to def_builtin
6599 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6600 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6601 with unconditional calls. Remove calls to def_builtin for
6602 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6603 __builtin_vec_xst_be.
6604 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6605 to define_expand, and add alternate RTL generation for P8.
6606 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6607 vsx_ld_elemrev_v8hi.
6608 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6609 add alternate RTL generation for P8.
6610 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6611 vsx_ld_elemrev_v16qi.
6612 (vsx_st_elemrev_v8hi): Convert define_insn
6613 to define_expand, and add alternate RTL generation for P8.
6614 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6615 vsx_st_elemrev_v8hi.
6616 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6617 add alternate RTL generation for P8.
6618 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6619 vsx_st_elemrev_v16qi.
6621 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6624 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6625 TARGET_AVX512ER check.
6626 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6627 neither -mzeroupper nor -mno-zeroupper is used and
6628 TARGET_EMIT_VZEROUPPER is set.
6629 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6630 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6632 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6634 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6635 folding of vector compares.
6636 (fold_build_vec_cmp): New helper function.
6637 (fold_compare_helper): New helper function.
6638 (builtin_function_type): Add compare builtins to the list of functions
6639 having unsigned arguments. Cosmetic updates to comment indentation.
6640 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6641 the not+eq combination.
6643 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6645 PR tree-optimization/82726
6646 PR tree-optimization/70754
6647 * tree-predcom.c (order_drefs_by_pos): New function.
6648 (combine_chains): Move code setting has_max_use_after to...
6649 (try_combine_chains): ...here. New parameter. Sort combined chains
6650 according to position information.
6651 (tree_predictive_commoning_loop): Update call to above function.
6652 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6654 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6656 PR tree-optimization/82726
6658 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6660 PR tree-optimization/70754
6661 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6662 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6663 combined stmt before it if not NULL.
6664 (combine_chains): Process refs reversely and compute dominance point
6668 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6670 PR tree-optimization/79663
6671 * tree-predcom.c (combine_chains): Process refs in reverse order
6672 only for ZERO length chains, and add explaining comment.
6674 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6676 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6677 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6678 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6680 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6682 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6684 2017-11-15 Martin Liska <mliska@suse.cz>
6686 * tree-cfg.c (pass_warn_function_return::execute):
6687 Compare warn_return_type for greater than zero.
6689 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6693 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6694 to return true on Xeon and not on Xeon Phi.
6695 (ix86_check_avx256_register): Changed to ...
6696 (ix86_check_avx_upper_register): ... this. Add extra check for
6697 VALID_AVX512F_REG_OR_XI_MODE.
6698 (ix86_avx_u128_mode_needed): Changed
6699 ix86_check_avx256_register to ix86_check_avx_upper_register.
6700 (ix86_check_avx256_stores): Changed to ...
6701 (ix86_check_avx_upper_stores): ... this. Changed
6702 ix86_check_avx256_register to ix86_check_avx_upper_register.
6703 (ix86_avx_u128_mode_after): Changed
6704 avx_reg256_found to avx_upper_reg_found. Changed
6705 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6706 (ix86_avx_u128_mode_entry): Changed
6707 ix86_check_avx256_register to ix86_check_avx_upper_register.
6708 (ix86_avx_u128_mode_exit): Ditto.
6709 * config/i386/i386.h: (host_detect_local_cpu): New define.
6711 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6713 * config/arm/xgene1.md (xgene1): Split into automatons
6714 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6715 (xgene1_f_load): Adjust reservations and/or types.
6716 (xgene1_f_store): Likewise.
6717 (xgene1_load_pair): Likewise.
6718 (xgene1_store_pair): Likewise.
6719 (xgene1_fp_load1): Likewise.
6720 (xgene1_load1): Likewise.
6721 (xgene1_store1): Likewise.
6722 (xgene1_move): Likewise.
6723 (xgene1_alu): Likewise.
6724 (xgene1_simd): Likewise.
6725 (xgene1_bfm): Likewise.
6726 (xgene1_neon_load1): Likewise.
6727 (xgene1_neon_store1): Likewise.
6728 (xgene1_neon_logic): Likewise.
6729 (xgene1_neon_st1): Likewise.
6730 (xgene1_neon_ld1r): Likewise.
6731 (xgene1_alu_cond): Added.
6732 (xgene1_shift_reg): Likwise.
6733 (xgene1_bfx): Likewise.
6734 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6736 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6739 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6741 (can_widen_mult_without_libcall): New function.
6742 (expand_mul_overflow): If only checking unsigned mul overflow,
6743 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6744 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6745 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6746 (expand_DIVMOD): Formatting fix.
6747 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6748 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6749 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6751 PR tree-optimization/82977
6752 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6753 constructed temporary to strlen_to_stridx.put.
6755 2017-11-15 Martin Liska <mliska@suse.cz>
6757 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6758 * configure: Regenerate.
6760 2017-11-15 Martin Liska <mliska@suse.cz>
6763 * config/sh/sh-mem.cc: Use proper probability for
6766 2017-11-14 Jeff Law <law@redhat.com>
6768 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6769 the red zone for stack_clash_protection_final_dynamic_probe targets
6770 when the total dynamic stack size is zero bytes.
6772 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6773 blocks is post order.
6775 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6777 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6778 * final.c (rest_of_clean_state): Set it for the
6779 -fcompare-debug dump.
6780 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6781 class when TDF_COMPARE_DEBUG is set.
6783 * dwarf2out.c (gen_producer_string): Discard
6786 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6789 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6790 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6791 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6792 (__TGMATH_REAL_2_3): Remove macros.
6793 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6794 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6796 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6798 (remquo): Define using __TGMATH_REAL_3.
6800 2017-11-14 Jeff Law <law@redhat.com>
6802 * vr-values.c: New file with contents extracted from tree-vrp.c.
6803 * Makefile.in (OBJS): Add vr-values.o
6804 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6805 (set_value_range, set_and_canonicalize_value_range): Likewise.
6806 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6807 (value_range_constant_singleton, symbolic_range_p): Likewise.
6808 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6809 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6810 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6811 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6812 (find_case_label_range, find_case_label_index): Likewise.
6813 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6814 (range_int_cst_singleton_p, value_inside_range): Likewise.
6815 (get_single_symbol): Likewise.
6816 (switch_update): Move structure definition here.
6817 (to_remove_edges, to_update_switch_stmts): Provide externs.
6818 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6819 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6820 (vrp_val_is_min, set_value_range): Likewise.
6821 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6822 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6823 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6824 (range_is_nonnull, range_int_cst_p): Likewwise.
6825 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6826 (get_single_symbol, operand_less_p): Likewise
6827 (compare_values_warnv, compare_values): Likewise.
6828 (value_inside_range, value_range_constant_singleton): Likewise.
6829 (zero_nonzero_bitgs_from_vr): Likewise.
6830 (extract_range_from_binary_expr_1): Likewise.
6831 (overflow_comparison_p): Likewise.
6832 (to_remove_edges, to_update_switch_stmts): Likewise.
6833 (find_case_label-index, find_case_label_range): Likewise.
6834 (switch_update, set_value_range_to_nonnegative): Remove.
6835 (set_value_range_to_truthvalue): Likewise.
6836 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6837 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6838 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6839 (find_case_label_ranges, test_for_singularity): Likewise.
6840 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6841 (x_vr_values): Move to its remaining use site.
6843 2017-11-10 Jeff Law <law@redhat.com>
6845 * vr-values.h (VR_INITIALIZER): Move #define here.
6846 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6847 * Makefile.in (OBJS): Add tree-evrp.o
6848 * tree-vrp.h (assert_info): Move structure definition here.
6849 (set_value_range_to_varying): Prototype.
6850 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6851 (infer_value_range, register_edge_assert_for): Likewise.
6852 (stmt_interesting_for_vrp): Likewise.
6853 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6854 (set_value_range_to_varying): No longer static.
6855 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6856 (infer_value_range, register_edge_assert_for): Likewise.
6858 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6860 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6861 generate the XXBRD instruction.
6863 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6864 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6865 as being compatible if -mabi=ieeelongdouble.
6866 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6867 to setup float128 built-ins with hardware support.
6868 (BU_FLOAT128_HW_2): Likewise.
6869 (BU_FLOAT128_HW_3): Likewise.
6870 (BU_FLOAT128_HW_VSX_1): Likewise.
6871 (BU_FLOAT128_HW_VSX_2): Likewise.
6872 (scalar_extract_expq): Change float128 built-in functions to
6873 accommodate having both KFmode and TFmode functions. Use the
6874 KFmode variant as the default.
6875 (scalar_extract_sigq): Likewise.
6876 (scalar_test_neg_qp): Likewise.
6877 (scalar_insert_exp_q): Likewise.
6878 (scalar_insert_exp_qp): Likewise.
6879 (scalar_test_data_class_qp): Likewise.
6880 (sqrtf128_round_to_odd): Delete processing the round to odd
6881 built-in functions as special built-in functions, and define them
6882 as float128 built-ins. Use the KFmode variant as the default.
6883 (truncf128_round_to_odd): Likewise.
6884 (addf128_round_to_odd): Likewise.
6885 (subf128_round_to_odd): Likewise.
6886 (mulf128_round_to_odd): Likewise.
6887 (divf128_round_to_odd): Likewise.
6888 (fmaf128_round_to_odd): Likewise.
6889 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6890 support for KFmode and TFmode xststdcqp calls.
6891 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6892 point, switch the built-in handlers for the get/set float128
6893 exponent, get float128 mantissa, float128 test built-ins, and the
6894 float128 round to odd built-in functions. Eliminate creating the
6895 float128 round to odd built-in functions as special built-ins.
6896 (rs6000_init_builtins): Eliminate special creation of the float128
6897 round to odd built-in functions.
6898 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6899 function insns to support both TFmode and KFmode variants.
6900 (xsxsigqp_<mode>): Likewise.
6901 (xsiexpqpf_<mode>): Likewise.
6902 (xsiexpqp_<mode>): Likewise.
6903 (xststdcqp_<mode>): Likewise.
6904 (xststdcnegqp_<mode>): Likewise.
6905 (xststdcqp_<mode>): Likewise.
6907 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6909 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6910 unused path_in_freq_ptr parameter.
6911 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6913 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6915 * ipa-inline.c (edge_badness): Dump sreal frequency.
6916 (compute_inlined_call_time): Match natural implementaiton ...
6917 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6918 forgotten division by CGRAPH_FREQ_BASE.
6920 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6922 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6923 Solaris 11. Update comment.
6924 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6926 * config/sol2.h (STARTFILE_SPEC): Likewise.
6927 * configure: Regenerate.
6929 2017-11-14 Carl Love <cel@us.ibm.com>
6931 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6932 le_ and be_ prefixes to swap* variables. Remove
6933 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6936 2017-11-14 Jason Merrill <jason@redhat.com>
6938 Support GTY((cache)) on hash_map.
6939 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6940 (ggc_cache_remove): Override it instead of ggc_mx.
6941 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6942 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6943 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6944 (simple_cache_map_traits): Override maybe_mx.
6945 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6946 (hash_map): Friend gt_cleare_cache.
6947 (gt_cleare_cache): New.
6948 * tree.h (tree_cache_traits): New hash_map traits class.
6949 (tree_cache_map): New typedef.
6951 2017-11-14 Richard Biener <rguenther@suse.de>
6953 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6954 paramter and handling.
6955 (cleanup_control_flow_bb): Likewise.
6956 (cleanup_control_flow_pre): New helper performing a DFS walk
6957 to call cleanup_control_flow_bb in PRE order.
6958 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6959 via cleanup_control_flow_pre.
6961 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6963 * config/aarch64/aarch64-simd.md
6964 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6965 (*aarch64_simd_bsl<mode>_alt): Likewise.
6966 (aarch64_simd_bsldi_internal): New.
6967 (aarch64_simd_bsldi_alt): Likewise.
6969 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6971 * tracer.c (better_p): Do not compare frequencies.
6972 * reg-stack.c (better_edge): Likewise.
6973 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6976 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6978 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6979 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6980 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6981 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6982 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6983 * predict.c (maybe_hot_frequency_p): Inline to ...
6984 (maybe_hot_count_p): ... here; rewrite to counts.
6985 (counts_to_freqs): Rename to ...
6986 (update_max_bb_count): ... this one.
6987 (expensive_function_p): Use counts.
6988 (estimate_bb_frequencies): Update.
6989 (rebuild_frequencies): Update.
6990 * predict.h (counts_to_freqs): Rename to ...
6991 (update_max_bb_count): ... this one.
6992 * profile.c (compute_branch_probabilities): Add debug info
6993 * tree-inline.c (expand_call_inline): Update debug info.
6994 (optimize_inline_calls): Use update_max_bb_count..
6995 (tree_function_versioning): Use update_max_bb_count..
6996 * value-prof.c (gimple_value_profile_transformations):
6997 Do not use update_max_bb_count.
6999 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7001 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7002 always use frequencies.
7004 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7006 * bb-reorder.c: Remove frequencies from comments.
7007 (better_edge_p): Use profile counts.
7008 (find_traces): Dump profile counts.
7009 (rotate_loop): Use profile counts.
7010 (find_traces_1_round): Likewise.
7011 (connect_better_edge_p): Use counts instead of probabilities for
7013 (copy_bb_p): Drop early check for non-0 frequency.
7014 (sanitize_hot_paths): Update comments.
7016 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7018 * ipa-split.c (struct split_point): Add count.
7019 (consider_split): Do not compute incoming frequency; compute incoming
7020 count and store it to split_point.
7021 (split_function): Set count of the call to split part correctly.
7023 2017-11-13 Carl Love <cel@us.ibm.com>
7025 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7027 2017-11-13 Tom Tromey <tom@tromey.com>
7029 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7031 2017-11-13 Carl Love <cel@us.ibm.com>
7033 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7034 Add support for builtins:
7035 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7036 vector {un,}signed {char,int,short},
7037 vector {un,}signed {char,int,short}) arguments.
7038 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7039 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7040 Add BU_P9V_AV_2 expansions for the builtins.
7041 * config/rs6000/altivec.h (vec_first_match_index,
7042 vec_first_mismatch_index, vec_first_match_or_eos_index,
7043 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7044 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7045 new extern declaration.
7046 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7047 * config/rs6000/vsx.md (first_match_index_<mode>,
7048 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7049 first_mismatch_or_eos_index_<mode>): Add define expand.
7050 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7051 * doc/extend.texi: Update the built-in documenation file for the new
7054 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7056 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7057 into the min/max operations for _Float<N> and _Float<N>X types.
7059 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7062 * dwarf2out.c (do_eh_frame): New static variable.
7063 (dwarf2out_begin_prologue): Set it.
7064 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7066 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7068 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7070 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7072 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7075 * debug.h (dwarf2out_do_eh_frame): Declare.
7076 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7077 (dwarf2out_do_frame): Use it.
7078 (dwarf2out_do_cfi_asm): Likewise.
7079 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7080 (dwarf2out_assembly_start): Likewise.
7081 (dwarf2out_begin_prologue): Fix comment.
7082 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7083 if the target needs either debug or unwind DWARF2 info.
7084 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7085 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7086 -ftrapping-math, -ftrapv and -fwrapv.
7088 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7090 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7091 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7092 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7093 (estimate_edge_size_and_time): Likewise.
7094 (ipa_merge_fn_summary_after_inlining): Likewise.
7095 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7096 (compute_uninlined_call_time): Use sreal_frequency.
7097 (compute_inlined_call_time): Likewise.
7098 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7099 * profile-count.c: Include sreal.h.
7100 (profile_count::to_sreal_scale): New.
7101 * profile-count.h: Forward declare sreal.
7102 (profile_count::to_sreal_scale): Declare.
7104 2017-11-13 Nathan Sidwell <nathan@acm.org>
7106 * diagnostic.c (maybe_line_and_column): New.
7107 (diagnostic_get_location_text): Use it.
7108 (diagnostic_report_current_module): Likewise.
7109 (test_diagnostic_get_location_text): Add tests.
7111 2017-11-13 Luis Machado <luis.machado@linaro.org>
7113 * doc/md.texi (Specifying processor pipeline description): Fix
7114 incorrect latency for the div instruction example.
7116 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7118 PR tree-optimization/78821
7119 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7120 that bit_not_p is the same.
7121 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7122 (split_group): Count precisely bit_not_p bits in each statement.
7123 (invert_op): New function.
7124 (imm_store_chain_info::output_merged_store): Use invert_op to
7125 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7126 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7128 2017-11-13 Martin Liska <mliska@suse.cz>
7130 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7131 (struct source_info): Likewise.
7132 (add_branch_counts): Likewise.
7133 (add_line_counts): Likewise.
7134 (function_summary): Likewise.
7135 (output_intermediate_line): Likewise.
7136 (generate_results): Likewise.
7138 2017-11-13 Martin Liska <mliska@suse.cz>
7140 * gcov.c (struct block_info): Remove typedef for block_t.
7141 (struct line_info): Likewise.
7142 (line_info::has_block): Likewise.
7143 (EXIT_BLOCK): Likewise.
7144 (unblock): Likewise.
7145 (circuit): Likewise.
7146 (get_cycles_count): Likewise.
7147 (process_file): Likewise.
7148 (read_graph_file): Likewise.
7149 (solve_flow_graph): Likewise.
7150 (find_exception_blocks): Likewise.
7151 (add_line_counts): Likewise.
7152 (accumulate_line_info): Likewise.
7153 (output_line_details): Likewise.
7155 2017-11-13 Martin Liska <mliska@suse.cz>
7157 * gcov.c (struct arc_info): Remove typedef for arc_t.
7158 (struct line_info): Likewise.
7159 (add_branch_counts): Likewise.
7160 (output_branch_count): Likewise.
7161 (function_info::~function_info): Likewise.
7162 (circuit): Likewise.
7163 (output_intermediate_line): Likewise.
7164 (read_graph_file): Likewise.
7165 (solve_flow_graph): Likewise.
7166 (find_exception_blocks): Likewise.
7167 (add_line_counts): Likewise.
7168 (accumulate_line_info): Likewise.
7169 (output_line_details): Likewise.
7170 (output_function_details): Likewise.
7172 2017-11-13 Martin Liska <mliska@suse.cz>
7174 * gcov.c (struct function_info): Remove typedef for function_t.
7175 (struct source_info): Likewise.
7176 (source_info::get_functions_at_location): Likewise.
7177 (solve_flow_graph): Likewise.
7178 (find_exception_blocks): Likewise.
7179 (add_line_counts): Likewise.
7180 (output_intermediate_file): Likewise.
7181 (process_file): Likewise.
7182 (generate_results): Likewise.
7183 (release_structures): Likewise.
7184 (read_graph_file): Likewise.
7185 (read_count_file): Likewise.
7186 (accumulate_line_counts): Likewise.
7187 (output_lines): Likewise.
7189 2017-11-13 Martin Liska <mliska@suse.cz>
7191 * gcov.c (function_info::function_info): Remove num_counts
7192 and add vector<gcov_type>.
7193 (function_info::~function_info): Use the vector.
7194 (process_file): Likewise.
7195 (read_graph_file): Likewise.
7196 (read_count_file): Likewise.
7197 (solve_flow_graph): Likewise.
7199 2017-11-13 Martin Liska <mliska@suse.cz>
7201 * gcov.c (function_info::is_artificial): New function.
7202 (process_file): Erase all artificial early.
7203 (generate_results): Skip as all artificial are already
7206 2017-11-13 Martin Liska <mliska@suse.cz>
7208 * gcov.c (read_graph_file): Store to global vector of functions.
7209 (read_count_file): Iterate the vector.
7210 (process_file): Likewise.
7211 (generate_results): Likewise.
7212 (release_structures): Likewise.
7214 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7216 PR tree-optimization/82954
7217 * gimple-ssa-store-merging.c
7218 (imm_store_chain_info::coalesce_immediate_stores): If
7219 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7221 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7223 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7224 Upddate call to ENDIAN_LANE_N.
7225 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7226 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7227 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7228 and use aarch64_endian_lane_rtx.
7229 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7231 2017-11-12 Tom de Vries <tom@codesourcery.com>
7233 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7235 2017-11-12 Tom de Vries <tom@codesourcery.com>
7237 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7239 2017-11-12 Tom de Vries <tom@codesourcery.com>
7241 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7243 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7244 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7245 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7247 2017-11-11 Martin Sebor <msebor@redhat.com>
7250 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7253 * prefic.c (translate_name): Replace strncpy with memcpy to
7254 avoid -Wstringop-truncation.
7256 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7258 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7260 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7262 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7264 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7266 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7269 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7271 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7272 enable generating XXBRH if the value is in a vector register.
7273 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7274 value is in a vector register.
7275 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7276 register to register bswap64's instead of doing the GPR sequence
7277 used on previous machines.
7278 (bswapdi2_xxbrd): New insn.
7279 (bswapdi2_reg): Disallow on ISA 3.0.
7280 (register to register bswap64 splitter): Do not split the insn on
7281 ISA 3.0 systems that use XXBRD.
7283 2017-11-10 Martin Sebor <msebor@redhat.com>
7286 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7288 (find_subframework_file): Same.
7290 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7292 * auto-profile.c (afdo_indirect_call): Drop frequency.
7293 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7294 (cgraph_node::create_edge): Drop frequency argument.
7295 (cgraph_node::create_indirect_edge): Drop frequency argument.
7296 (cgraph_edge::make_speculative): Drop frequency arguments.
7297 (cgraph_edge::resolve_speculation): Do not update frequencies
7298 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7299 (cgraph_node::dump): Check consistency in IPA mode.
7300 (cgraph_edge::maybe_hot_p): Use IPA counter.
7301 (cgraph_edge::verify_count_and_frequency): Rename to ...
7302 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7303 (cgraph_node::verify_node): Update.
7304 * cgraph.h (struct cgraph_edge): Drop frequency.
7305 (cgraph_edge::frequency): New function.
7306 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7308 (cgraph_edge::rebuild_edges): Likewise.
7309 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7310 (duplicate_thunk_for_node): Do not pass frequency.
7311 (cgraph_node::create_clone): Scale only counts.
7312 (cgraph_node::create_virtual_clone): Do not pass frequency.
7313 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7314 (cgraph_node::create_version_clone): Do not pass frequency.
7315 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7316 (cgraph_node::expand_thunk): Do not pass frequency.
7317 (cgraph_node::create_wrapper): Do not pass frequency.
7318 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7320 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7321 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7322 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7323 (gather_caller_stats): Use frequency function.
7324 (ipcp_cloning_candidate_p): Use frequency function.
7325 (ipcp_propagate_stage): Use frequency function.
7326 (get_info_about_necessary_edges): Use frequency function.
7327 (update_profiling_info): Update only IPA profile.
7328 (update_specialized_profile): Use frequency functoin.
7329 (perhaps_add_new_callers): Update only IPA profile.
7330 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7331 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7332 (dump_ipa_call_summary): Use frequency function.
7333 (estimate_edge_size_and_time): Use frequency function.
7334 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7335 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7336 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7337 (update_noncloned_counts): ... ths one; scale counts only.
7338 (clone_inlined_nodes): Do not scale frequency.
7339 (inline_call): Do not pass frequency.
7340 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7341 (compute_inlined_call_time): Use IPA profile.
7342 (want_inline_small_function_p): Use IPA profile.
7343 (want_inline_self_recursive_call_p): Use IPA profile.
7344 (edge_badness): Use IPA profile.
7345 (lookup_recursive_calls): Use IPA profile.
7346 (recursive_inlining): Do not pass frequency.
7347 (resolve_noninline_speculation): Do not update frequency.
7348 (inline_small_functions): Collect max of IPA profile.
7349 (dump_overall_stats): Dump IPA porfile.
7350 (dump_inline_stats): Dump IPA porfile.
7351 (ipa_inline): Collect IPA stats.
7352 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7353 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7354 (ipa_propagate_frequency): Use frequency function.
7355 (ipa_profile): Cleanup.
7356 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7357 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7358 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7359 (input_node): Do not stream frequency.
7360 (input_edge): Do not stream frequency.
7361 (merge_profile_summaries): Scale only IPA profiles.
7362 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7363 * predict.c (drop_profile): Do not recompute frequency.
7364 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7365 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7366 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7367 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7368 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7369 * tree-inline.c (copy_bb): Do not scale frequency.
7370 (expand_call_inline): Do not scale frequency.
7371 (tree_function_versioning): Do not scale frequency.
7372 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7374 2017-11-10 Julia Koval <julia.koval@intel.com>
7376 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7377 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7378 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7379 (_mm256_maskz_gf2p8affine_epi64_epi8)
7380 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7381 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7382 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7383 (__builtin_ia32_vgf2p8affineqb_v32qi)
7384 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7385 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7387 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7391 (arm_option_override): Refactor.
7392 (arm_option_reconfigure_globals): New.
7393 (arm_options_perform_arch_sanity_checks): New.
7394 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7396 (arm_options_perform_arch_sanity_checks): Likewise
7398 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7400 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7401 (power9-qpmul): New.
7402 * rs6000/rs6000.md ("type" attr): Add qmul.
7403 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7404 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7405 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7407 2017-11-10 Martin Sebor <msebor@redhat.com>
7410 * builtins.c (compute_objsize): Handle arrays that
7411 compute_builtin_object_size likes to fail for. Make extern.
7412 * builtins.h (compute_objsize): Declare.
7413 (check_strncpy_sizes): New function.
7414 (expand_builtin_strncpy): Call check_strncpy_sizes.
7415 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7416 -Wstringop-truncation.
7417 (gimple_fold_builtin_strncat): Same.
7418 * gimple.c (gimple_build_call_from_tree): Set call location.
7419 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7420 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7421 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7422 (handle_builtin_strlen): Use strlen_to_stridx.
7423 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7425 Use strlen_to_stridx.
7426 (pass_strlen::execute): Release strlen_to_stridx.
7427 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7428 (-Wstringop-truncation): Document new option.
7430 2017-11-10 Martin Liska <mliska@suse.cz>
7432 PR gcov-profile/82702
7433 * gcov.c (main): Handle intermediate files in a different
7435 (get_gcov_intermediate_filename): New function.
7436 (output_gcov_file): Remove support of intermediate files.
7437 (generate_results): Allocate intermediate file.
7438 (release_structures): Clean-up properly fn_end.
7439 (output_intermediate_file): Start iterating with line 1.
7441 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7443 PR tree-optimization/82929
7444 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7445 ops_swapped_p non-static data member.
7446 (store_immediate_info::store_immediate_info): Clear it.
7447 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7448 ops set ops_swapped_p.
7449 (count_multiple_uses): Handle ops_swapped_p.
7451 2017-11-10 Martin Liska <mliska@suse.cz>
7453 * coverage.c (coverage_init): Stream information about
7454 support of has_unexecuted_blocks.
7455 * doc/gcov.texi: Document that.
7456 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7457 * gcov.c (read_graph_file): Likewise.
7458 (output_line_beginning): Fix a small issue with
7461 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7463 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7464 reference of trivial component.
7466 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7469 * gimple-ssa-store-merging.c
7470 (pass_store_merging::terminate_all_aliasing_chains): For
7471 gimple_store_p stmts also call refs_output_dependent_p.
7473 PR rtl-optimization/82913
7474 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7477 2017-11-09 Jeff Law <law@redhat.com>
7479 * vr-values.h: New file with vr_values class.
7480 * tree-vrp.c: Include vr-values.h
7481 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7482 data objects into the vr_values class.
7483 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7484 (get_value_range): Make it a member function within vr_values class.
7485 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7486 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7487 (op_with_constant_singleton_value_range): Likewise.
7488 (extract_range_for_var_from_comparison_expr): Likewise.
7489 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7490 (extract_range_from_binary_expr): Likewise.
7491 (extract_range_from_unary_expr): Likewise.
7492 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7493 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7494 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7495 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7496 (compare_name_with_value, compare_names): Likewise.
7497 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7498 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7499 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7500 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7501 (extract_range_from_phi_node): Likewise.
7502 (simplify_truth_ops_using_ranges): Likewise.
7503 (simplify_div_or_mod_using_ranges): Likewise.
7504 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7505 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7506 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7507 (simplify_float_conversion_using_ranges): Likewise.
7508 (simplify_internal_call_using_ranges): Likewise.
7509 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7510 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7511 poitner via x_vr_values for calls into gimple folder.
7512 (vrp_initialize_lattice): Make this the vr_values ctor.
7513 (vrp_free_lattice): Make this the vr_values dtor.
7514 (set_vr_value): New function.
7515 (class vrp_prop): Add vr_values data member. Add various member
7516 functions as well as member functions that delegate to vr_values.
7517 (check_array_ref): Make a member function within vrp_prop class.
7518 (search_for_addr_array, vrp_initialize): Likewise.
7519 (vrp_finalize): Likewise. Revamp to avoid direct access to
7520 vr_value, values_propagated, etc.
7521 (check_array_bounds): Extract vrp_prop class instance pointer from
7522 walk info structure. Use it to call member functions.
7523 (check_all_array_refs): Make a member function within vrp_prop class.
7524 Smuggle class instance pointer via walk info structure.
7525 (x_vr_values): New local static.
7526 (vrp_valueize): Use x_vr_values to get class instance.
7527 (vr_valueize_1): Likewise.
7528 (class vrp_folder): Add vr_values data member. Add various member
7529 functions as well as member functions that delegate to vr_values.
7530 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7531 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7532 class instance from vr_values. Use it to call member functions.
7533 (vrp_dom_walker): Add vr_values data member.
7534 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7535 instance via x_vr_values.
7536 (identify_jump_threads): Accept vr_values as argument. Store
7537 it into the walker structure.
7538 (evrp_dom_walker): Add vr_values class data member. Add various
7540 (evrp_dom_walker::try_find_new_range): Use vr_values data
7541 member to access the memory allocator.
7542 (evrp_dom_walker::before_dom_children): Store vr_values class
7543 instance into the vrp_folder class.
7544 (evrp_dom_walker::push_value_range): Rework to avoid direct
7545 access to num_vr_values and vr_value.
7546 (evrp_dom_walker::pop_value_range): Likewise.
7547 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7548 Use vr_values to get to dump_all_value_ranges member function.
7549 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7550 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7551 Pass vr_values class instance down to identify_jump_threads.
7552 Remove call to vrp_free_lattice.
7553 (debug_all_value_ranges): Remove.
7555 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7556 (vrp_folder): Likewise.
7558 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7559 Get it from the existing bitmap instead.
7560 (vrp_intersect_ranges_1): Likewise.
7562 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7564 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7566 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7567 set bit_not_p to it.
7568 (imm_store_chain_info::coalesce_immediate_stores): Break group
7569 if bit_not_p is different.
7570 (count_multiple_uses, split_group,
7571 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7572 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7573 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7574 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7575 to store_immediate_info ctor.
7577 2017-11-09 Jim Wilson <jimw@sifive.com>
7579 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7580 (scan_prog_file): Likewise.
7582 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7584 * bb-reorder.c (max_entry_frequency): Remove.
7585 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7586 connect_traces, push_to_next_round_p): Remove prototypes.
7587 (find_traces_1_round): Use counts only.
7588 (push_to_next_round_p): Likewise.
7589 (find_traces): Likewise.
7590 (rotate_loop): Likewise.
7591 (find_traces_1_round): Likewise.
7592 (connect_traces): Likewise.
7593 (edge_order): Likewise.
7595 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7597 * config/arm/arm.c (output_return_instruction): Add comments to
7598 indicate requirement for cmse_nonsecure_entry return to account
7599 for the size of clearing instruction output here.
7600 (thumb_exit): Likewise.
7601 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7602 return in hardfloat mode.
7604 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7606 * config/rs6000/rs6000.c (machine_function): Add a bool,
7607 "toc_is_wrapped_separately".
7608 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7609 if it wasn't explicitly set or unset, we are optimizing for speed, and
7610 doing separate shrink-wrapping.
7611 (rs6000_get_separate_components): Enable the TOC component if
7612 saving the TOC register in the prologue.
7613 (rs6000_components_for_bb): Handle the TOC component.
7614 (rs6000_emit_prologue_components): Store the TOC register where needed.
7615 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7616 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7618 2017-11-09 Martin Jambor <mjambor@suse.cz>
7620 * ipa-param-manipulation.c: New file.
7621 * ipa-param-manipulation.h: Likewise.
7622 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7623 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7624 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7625 (ipa_parm_adjustment): Likewise.
7626 (ipa_parm_adjustment_vec): Likewise.
7627 (ipa_get_vector_of_formal_parms): Moved declaration to
7628 ipa-param-manipulation.h.
7629 (ipa_get_vector_of_formal_parm_types): Likewise.
7630 (ipa_modify_formal_parameters): Likewise.
7631 (ipa_modify_call_arguments): Likewise.
7632 (ipa_combine_adjustments): Likewise.
7633 (ipa_dump_param_adjustments): Likewise.
7634 (ipa_modify_expr): Likewise.
7635 (ipa_get_adjustment_candidate): Likewise.
7636 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7637 ipa-param-manipulation.c.
7638 (ipa_get_vector_of_formal_parm_types): Likewise.
7639 (ipa_modify_formal_parameters): Likewise.
7640 (ipa_modify_call_arguments): Likewise.
7641 (ipa_modify_expr): Likewise.
7642 (get_ssa_base_param): Likewise.
7643 (ipa_get_adjustment_candidate): Likewise.
7644 (index_in_adjustments_multiple_times_p): Likewise.
7645 (ipa_combine_adjustments): Likewise.
7646 (ipa_dump_param_adjustments): Likewise.
7647 * tree-sra.c: Also include ipa-param-manipulation.h
7648 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7651 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7652 Alan Hayward <alan.hayward@arm.com>
7653 David Sherwood <david.sherwood@arm.com>
7655 * doc/sourcebuild.texi (vect_masked_store): Document.
7657 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7658 Alan Hayward <alan.hayward@arm.com>
7659 David Sherwood <david.sherwood@arm.com>
7661 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7663 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7664 Alan Hayward <alan.hayward@arm.com>
7665 David Sherwood <david.sherwood@arm.com>
7667 * doc/sourcebuild.texi (vect_variable_length): Document.
7669 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7670 Alan Hayward <alan.hayward@arm.com>
7671 David Sherwood <david.sherwood@arm.com>
7673 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7675 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7676 Alan Hayward <alan.hayward@arm.com>
7677 David Sherwood <david.sherwood@arm.com>
7679 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7681 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7682 Alan Hayward <alan.hayward@arm.com>
7683 David Sherwood <david.sherwood@arm.com>
7685 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7686 previously undocumented selectors.
7687 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7689 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7691 * doc/rtl.texi (const_vector): Say that elements can be
7692 const_wide_ints too.
7693 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7694 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7695 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7696 * optabs.c (expand_vector_broadcast): Likewise.
7698 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7699 Alan Hayward <alan.hayward@arm.com>
7700 David Sherwood <david.sherwood@arm.com>
7702 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7703 scaled index even if the unscaled address was invalid.
7704 Don't increase the complexity of using a scale in that case.
7706 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7707 Alan Hayward <alan.hayward@arm.com>
7708 David Sherwood <david.sherwood@arm.com>
7710 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7711 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7712 UNITS_PER_WORD bytes.
7713 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7714 based on REGMODE_NATURAL_SIZE of the inner mode.
7715 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7716 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7717 * expmed.c (lowpart_bit_field_p): Divide the value up into
7718 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7719 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7720 whether something is likely to occupy more than one register.
7722 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7725 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7726 profile_count::adjust_for_ipa_scaling.
7727 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7728 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7730 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7732 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7734 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7735 (split_group): Add total_orig and total_new arguments, estimate the
7736 number of statements related to the store group without store merging
7737 and with store merging.
7738 (imm_store_chain_info::output_merged_store): Adjust split_group
7739 callers, punt if estimated number of statements with store merging
7740 is not smaller than estimated number of statements without it.
7742 (handled_load): Remove has_single_use checks.
7743 (pass_store_merging::process_store): Likewise.
7745 2017-11-09 Richard Biener <rguenther@suse.de>
7747 PR tree-optimization/82902
7748 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7750 2017-11-09 Martin Liska <mliska@suse.cz>
7753 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7756 2017-11-09 Martin Liska <mliska@suse.cz>
7758 PR tree-optimization/82669
7759 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7761 2017-11-09 Martin Liska <mliska@suse.cz>
7763 PR gcov-profile/48463
7764 * coverage.c (coverage_begin_function): Output also end locus
7765 of a function and information whether the function is
7767 * gcov-dump.c (tag_function): Parse and print the information.
7768 * gcov.c (INCLUDE_MAP): Add include.
7769 (INCLUDE_SET): Likewise.
7770 (struct line_info): Move earlier in the source file because
7771 of vector<line_info> in function_info structure.
7772 (line_info::line_info): Likewise.
7773 (line_info::has_block): Likewise.
7774 (struct source_info): Add new member index.
7775 (source_info::get_functions_at_location): New function.
7776 (function_info::group_line_p): New function.
7777 (output_intermediate_line): New function.
7778 (output_intermediate_file): Use the mentioned function.
7779 (struct function_start): New.
7780 (struct function_start_pair_hash): Likewise.
7781 (process_file): Add code that identifies group functions.
7782 Assign lines either to global or function scope.
7783 (generate_results): Skip artificial functions.
7784 (find_source): Assign index for each source file.
7785 (read_graph_file): Read new flag artificial and end_line.
7786 (add_line_counts): Assign it either to global of function scope.
7787 (accumulate_line_counts): Isolate core of the function to
7788 accumulate_line_info and call it for both function and global
7790 (accumulate_line_info): New function.
7791 (output_line_beginning): Fix GNU coding style.
7792 (print_source_line): New function.
7793 (output_line_details): Likewise.
7794 (output_function_details): Likewise.
7795 (output_lines): Iterate both source (global) scope and function
7797 (struct function_line_start_cmp): New class.
7798 * doc/gcov.texi: Reflect changes in documentation.
7800 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7803 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7804 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7805 and similarly for not instead of neg.
7807 2017-11-08 Andi Kleen <ak@linux.intel.com>
7809 * config/i386/i386.opt: Add -mforce-indirect-call.
7810 * config/i386/predicates.md: Check for flag_force_indirect_call.
7811 * doc/invoke.texi: Document -mforce-indirect-call
7813 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7815 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7817 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7818 config/riscv/riscv.c (predict.h): New include.
7819 (riscv_slow_unaligned_access_p): No longer static.
7820 (riscv_block_move_straight): Add require.
7821 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7823 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7826 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7827 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7828 nonimmediate_operand predicate for operand 1 instead of
7831 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7833 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7835 * config/aarch64/constraints.md (Uml): New constraint.
7836 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7839 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7841 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7842 of two vec_duplicates into a vec_concat.
7844 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7846 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7847 Simplify vec_merge of vec_duplicate and vec_concat.
7848 * config/aarch64/constraints.md (Utq): New constraint.
7849 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7852 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7854 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7855 Simplify vec_merge of vec_duplicate and const_vector.
7856 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7858 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7859 mode iterator. Update predicate on operand 1 to
7860 handle non-const_vec constants. Delete constraints.
7861 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7863 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7865 PR tree-optimization/78821
7866 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7868 (store_operand_info::store_operand_info): Initialize it to false.
7869 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7870 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7871 store in the group, and if chain_info is non-NULL, to ignore altogether
7873 (compatible_load_p): Fail if bit_not_p does not match.
7874 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7875 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7876 (pass_store_merging::process_store): Adjust
7877 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7878 call terminate_all_aliasing_chains newly when adding a store into
7879 a chain with non-NULL chain_info.
7881 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7883 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7885 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7887 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7889 (aarch64_layout_frame): Initialise emit_frame_chain.
7890 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7891 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7893 2017-11-08 Martin Liska <mliska@suse.cz>
7895 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7896 of gimple_call_internal_p.
7898 2017-11-07 Tom de Vries <tom@codesourcery.com>
7900 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7902 2017-11-07 Tom de Vries <tom@codesourcery.com>
7904 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7907 2017-11-08 Martin Liska <mliska@suse.cz>
7910 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7912 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7914 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7916 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7919 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7920 (*cmp<mode>_ccz_1): New insn with $k alternative.
7923 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7924 RTX_COMM_COMPARE as commutative as well.
7925 (ix86_binary_operator_ok): Formatting fix.
7926 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7927 *<code><mode>3<mask_name><round_saeonly_name>,
7928 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7929 *<s>mul<mode>3_highpart<mask_name>,
7930 *vec_widen_umult_even_v16si<mask_name>,
7931 *vec_widen_umult_even_v8si<mask_name>,
7932 *vec_widen_umult_even_v4si<mask_name>,
7933 *vec_widen_smult_even_v16si<mask_name>,
7934 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7935 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7936 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7937 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7938 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7939 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7940 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7941 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7942 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7943 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7944 ix86_binary_operator_ok. Formatting fixes.
7945 (*<plusminus_insn><mode>3<mask_name><round_name>,
7946 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7949 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7951 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7952 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7953 eq and ne if TARGET_ISEL.
7954 (cmp): New code_iterator.
7955 (UNS, UNSU_, UNSIK): New code_attrs.
7956 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7957 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7959 ("eq<mode>3"): ... this.
7960 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7962 ("ne<mode>3"): ... this.
7964 2017-11-07 Julia Koval <julia.koval@intel.com>
7967 * common/config/i386/i386-common.c
7968 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7969 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7970 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7971 * config/i386/i386.opt: Ditto.
7972 * config/i386/i386.c (ix86_target_string): Ditto.
7973 (ix86_option_override_internal): Ditto.
7974 (ix86_init_mpx_builtins): Move MPX to args2.
7975 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7976 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7977 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7978 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7979 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7980 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7981 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7983 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7986 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7987 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7988 (zero-extendsidi peephole2): Remove peephole.
7990 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7993 * stor-layout.c: Include attribs.h.
7994 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7995 explicit lookup of "aligned" attribute.
7997 2017-11-07 Andrew Waterman <andrew@sifive.com>
7999 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8000 (riscv_expand_block_move): Likewise.
8001 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8003 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8004 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8005 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8006 (riscv_adjust_block_mem): Likewise.
8007 (riscv_block_move_loop): Likewise.
8008 (riscv_expand_block_move): Likewise.
8009 * config/riscv/riscv.md (movmemsi): New pattern.
8011 2017-11-07 Michael Clark <michaeljclark@mac.com>
8013 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8014 (MUSL_DYNAMIC_LINKER): Likewise.
8016 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8018 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8019 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8022 2017-11-07 Richard Biener <rguenther@suse.de>
8024 * match.pd: Fix build.
8026 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8027 Jackson Woodruff <jackson.woodruff@arm.com>
8029 PR tree-optimization/71026
8030 * match.pd: Canonicalize negate in division.
8032 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8035 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8037 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8039 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8040 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8042 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8044 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8045 non-scalar integral types.
8046 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8047 (-(A-B), -(~A)): New transformations.
8049 2017-11-07 Tom de Vries <tom@codesourcery.com>
8051 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8052 semicolon after "do {} while (0)".
8053 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8054 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8055 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8056 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8057 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8059 2017-11-07 Tom de Vries <tom@codesourcery.com>
8061 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8062 after "do {} while (0)".
8063 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8064 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8065 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8066 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8067 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8069 2017-11-07 Tom de Vries <tom@codesourcery.com>
8072 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8074 (arm_rtx_costs_internal): Add missing semicolon after
8075 HANDLE_NARROW_SHIFT_ARITH call.
8077 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8079 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8080 disable isel if it was not set explicitly.
8082 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8084 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8085 (add_type_attribute) likewise.
8087 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8089 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8090 of struct ix86_frame.
8091 (ix86_initial_elimination_offset): Likewise.
8092 (ix86_expand_split_stack_prologue): Likewise.
8094 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8096 * tree-vrp.h (enum value_range_type): Update stale comment.
8098 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8099 Alan Hayward <alan.hayward@arm.com>
8100 David Sherwood <david.sherwood@arm.com>
8102 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8103 (aarch64_expand_vec_perm_const): Take the number of units too.
8104 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8105 (aarch64_expand_vec_perm_const): Likewise.
8106 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8107 (vec_perm<mode>): Update accordingly.
8109 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8110 Alan Hayward <alan.hayward@arm.com>
8111 David Sherwood <david.sherwood@arm.com>
8113 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8114 Take the number of units too.
8115 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8116 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8117 but check for a vector mode before rather than after the call.
8118 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8119 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8120 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8121 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8122 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8123 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8124 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8125 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8126 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8127 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8128 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8129 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8130 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8131 (aarch64_sqdmull2_n<mode>): Update accordingly.
8133 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8134 Alan Hayward <alan.hayward@arm.com>
8135 David Sherwood <david.sherwood@arm.com>
8137 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8138 the number of units too.
8139 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8140 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8141 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8142 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8143 (vec_store_lanesxi<mode>): Update accordingly.
8145 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8146 Alan Hayward <alan.hayward@arm.com>
8147 David Sherwood <david.sherwood@arm.com>
8149 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8150 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8151 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8152 of units rather than the mode.
8153 * config/aarch64/iterators.md (nunits): New mode attribute.
8154 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8155 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8156 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8157 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8158 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8159 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8160 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8161 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8162 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8163 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8164 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8165 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8166 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8167 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8168 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8169 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8170 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8171 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8172 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8173 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8174 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8175 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8176 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8177 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8178 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8179 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8180 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8181 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8182 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8183 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8184 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8185 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8186 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8187 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8188 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8189 (aarch64_simd_vec_setv2di): Likewise.
8191 2017-11-06 Carl Love <cel@us.ibm.com>
8193 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8195 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8196 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8198 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8200 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8201 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8202 Add power 8 macro expansions.
8203 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8204 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8205 power 8 instructions. (VSX_XXBR): Add iterator.
8207 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8209 * config/arm/arm.md (predicable_short_it): Change default to "no",
8210 improve documentation, remove uses that are identical to the default.
8211 (enabled_for_depr_it): Rename to enabled_for_short_it.
8212 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8213 Remove default uses.
8214 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8215 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8216 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8217 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8219 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8222 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8223 float128 helper macros, which are no longer used after deleting
8224 the old 'q' built-in functions, and moving the round to odd
8225 built-in functions to being special built-in functions.
8226 (BU_FLOAT128_2): Likewise.
8227 (BU_FLOAT128_1_HW): Likewise.
8228 (BU_FLOAT128_2_HW): Likewise.
8229 (BU_FLOAT128_3_HW): Likewise.
8230 (FABSQ): Delete old 'q' built-in functions.
8231 (COPYSIGNQ): Likewise.
8232 (SQRTF128_ODD): Move round to odd built-in functions to be
8233 special built-in functions, so that we can handle
8234 -mabi=ieeelongdouble.
8235 (TRUNCF128_ODD): Likewise.
8236 (ADDF128_ODD): Likewise.
8237 (SUBF128_ODD): Likewise.
8238 (MULF128_ODD): Likewise.
8239 (DIVF128_ODD): Likewise.
8240 (FMAF128_ODD): Likewise.
8241 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8242 built-in names to 'f128'.
8243 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8244 old 'q' built-in functions, as the machine independent code for
8245 'f128' built-in functions handles this.
8246 (rs6000_expand_builtin): Add expansion for float128 round to odd
8247 functions, keying off on -mabi=ieeelongdouble of whether to use
8248 the KFmode or TFmode variant.
8249 (rs6000_init_builtins): Initialize the _Float128 round to odd
8251 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8252 _Float128 'q' built-in functions are now mapped into the new
8253 'f128' built-in functions.
8255 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8257 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8258 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8260 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8263 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8264 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8265 * toplev.c: Include "ipa-fnsummary.h".
8266 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8268 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8270 PR tree-optimization/82838
8271 * gimple-ssa-store-merging.c
8272 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8273 on a separate gimple_seq which is then appended to seq.
8275 2017-11-06 Jeff Law <law@redhat.com>
8278 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8279 (get_probe_interval): New functions.
8280 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8281 (ix86_adjust_stack_and_probe): Likewise.
8282 (output_adjust_stack_and_probe): Likewise.
8283 (ix86_emit_probe_stack_range): Likewise.
8284 (ix86_expand_prologue): Likewise.
8286 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8288 PR tree-optimization/82816
8289 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8290 if the modes of the two types are the same.
8291 (convert_plusminus_to_widen): Likewise.
8293 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8295 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8297 (usadv16qi): New define_expand.
8298 (usadv8hi): New define_expand.
8300 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8303 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8305 (inline_transform): Likewise.
8306 * predict.c (counts_to_freqs): Remove useless conditional.
8307 * profile-count.h (profile_count::apply_scale): Move sanity check.
8308 * tree-inline.c (copy_bb): Always scale.
8309 (copy_cfg_body): Likewise.
8311 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8314 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8316 (*cmp_ite0): Add enabled_for_depr_it attribute.
8317 (*cmp_ite1): Likewise.
8319 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8321 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8324 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8326 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8327 return the value by pointer.
8328 (extract_range_from_multiplicative_op_1): Update accordingly.
8329 Return as soon as an operation fails.
8331 2017-11-05 Tom de Vries <tom@codesourcery.com>
8334 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8335 (DEF_SANITIZER_BUILTIN): ... here.
8336 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8337 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8339 2017-11-05 Tom de Vries <tom@codesourcery.com>
8342 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8344 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8345 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8346 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8348 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8349 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8350 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8352 2017-11-05 Tom de Vries <tom@codesourcery.com>
8355 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8358 2017-11-04 Michael Clark <michaeljclark@mac.com>
8360 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8361 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8363 (*addsi3_extended): Likewise.
8364 (*addsi3_extended2): Likewise.
8365 (<optab>si3): Likewise.
8366 (<optab>di3): Likewise.
8367 (<optab><mode>3): Likewise.
8368 (<*optabe>si3_internal): Likewise.
8369 (zero_extendqi<SUPERQI:mode>2): Likewise.
8370 (*add<mode>hi3): Likewise.
8371 (*xor<mode>hi3): Likewise.
8372 (<optab>di3): Likewise.
8373 (*<optab>si3_extend): Likewise.
8374 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8375 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8376 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8378 2017-11-04 Andrew Waterman <andrew@sifive.com>
8380 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8381 TARGET_STRICT_ALIGN based upon -mtune argument.
8383 2017-11-04 Andrew Waterman <andrew@sifive.com>
8385 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8387 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8389 * config/i386/i386.c (choose_basereg): Use optional scratch
8390 register and add assertion.
8391 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8392 needed, and don't allocate stack.
8393 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8394 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8395 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8397 2017-11-03 Jeff Law <law@redhat.com>
8399 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8400 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8401 to probe at the start of a noreturn function.
8403 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8405 PR tree-optimization/78821
8406 * gimple-ssa-store-merging.c: Update the file comment.
8407 (MAX_STORE_ALIAS_CHECKS): Define.
8408 (struct store_operand_info): New type.
8409 (store_operand_info::store_operand_info): New constructor.
8410 (struct store_immediate_info): Add rhs_code and ops data members.
8411 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8412 and op1r arguments to the ctor, initialize corresponding data members.
8413 (struct merged_store_group): Add load_align_base and load_align
8415 (merged_store_group::merged_store_group): Initialize them.
8416 (merged_store_group::do_merge): Update them.
8417 (merged_store_group::apply_stores): Pick the constant for
8418 encode_tree_to_bitpos from one of the two operands, or skip
8419 encode_tree_to_bitpos if neither operand is a constant.
8420 (class pass_store_merging): Add process_store method decl. Remove
8421 bool argument from terminate_all_aliasing_chains method decl.
8422 (pass_store_merging::terminate_all_aliasing_chains): Remove
8423 var_offset_p argument and corresponding handling.
8424 (stmts_may_clobber_ref_p): New function.
8425 (compatible_load_p): New function.
8426 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8427 if there is overlap and rhs_code is not INTEGER_CST. For
8428 non-overlapping stores terminate group if rhs is not mergeable.
8429 (get_alias_type_for_stmts): Change first argument from
8430 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8431 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8432 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8434 (get_location_for_stmts): Change first argument from
8435 auto_vec<gimple *> & to vec<gimple *> &.
8436 (struct split_store): Remove orig_stmts data member, add orig_stores.
8437 (split_store::split_store): Create orig_stores rather than orig_stmts.
8438 (find_constituent_stmts): Renamed to ...
8439 (find_constituent_stores): ... this. Change second argument from
8440 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8441 to info structures rather than the statements.
8442 (split_group): Rename ALLOW_UNALIGNED argument to
8443 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8444 it. Adjust find_constituent_stores caller.
8445 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8446 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8447 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8448 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8449 (mem_valid_for_store_merging): New function.
8450 (handled_load): New function.
8451 (pass_store_merging::process_store): New method.
8452 (pass_store_merging::execute): Use process_store method. Adjust
8453 terminate_all_aliasing_chains caller.
8455 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8457 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8458 Return true for more constants, symbols and label references.
8459 (aarch64_valid_floating_const): Remove unused function.
8461 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8464 * config/aarch64/aarch64.c (aarch64_layout_frame):
8465 Undo forcing of LR at bottom of frame.
8467 2017-11-03 Jeff Law <law@redhat.com>
8470 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8471 for int_registers_saved.
8473 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8474 extracted from tree-ssa-dom.c.
8475 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8476 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8477 (record_equivalences_from_incoming_edge): Add additional argument
8478 to single_pred_edge_ignoring_loop_edges call.
8479 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8480 (uncprop_dom_walker::before_dom_children): Add additional argument
8481 to single_pred_edge_ignoring_loop_edges call.
8482 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8483 single_pred_edge_ignoring_loop_edges rather than open coding.
8484 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8486 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8488 * match.pd (-(-A)): Rewrite.
8490 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8492 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8493 (rs6000_emit_int_cmove): New declaration.
8494 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8495 (rs6000_emit_sISEL): Delete.
8496 (rs6000_emit_int_cmove): Make non-static.
8497 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8498 instead of rs6000_emit_sISEL.
8500 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8502 * asan.c (create_cond_insert_point): Maintain profile.
8503 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8505 * basic-block.h (struct basic_block_def): Remove frequency.
8506 (EDGE_FREQUENCY): Use to_frequency
8507 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8509 (find_traces): Update to use to_frequency.
8510 (find_traces_1_round): Likewise; use only IPA counts.
8511 (bb_to_key): Likewise.
8512 (connect_traces): Use IPA counts only.
8513 (copy_bb_p): Update to use to_frequency.
8514 (fix_up_crossing_landing_pad): Likewise.
8515 (sanitize_hot_paths): Likewise.
8516 * bt-load.c (basic_block_freq): Likewise.
8517 * cfg.c (init_flow): Set count_max to uninitialized.
8518 (check_bb_profile): Remove frequencies; check counts.
8519 (dump_bb_info): Do not dump frequencies.
8520 (update_bb_profile_for_threading): Update counts only.
8521 (scale_bbs_frequencies_int): Likewise.
8522 (MAX_SAFE_MULTIPLIER): Remove.
8523 (scale_bbs_frequencies_gcov_type): Update counts only.
8524 (scale_bbs_frequencies_profile_count): Update counts only.
8525 (scale_bbs_frequencies): Update counts only.
8526 * cfg.h (struct control_flow_graph): Add count-max.
8527 (update_bb_profile_for_threading): Update prototype.
8528 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8529 (find_many_sub_basic_blocks): Likewise.
8530 * cfgcleanup.c (try_forward_edges): Likewise.
8531 (try_crossjump_to_edge): Likewise.
8532 * cfgexpand.c (expand_gimple_cond): Likewise.
8533 (expand_gimple_tailcall): Likewise.
8534 (construct_init_block): Likewise.
8535 (construct_exit_block): Likewise.
8536 * cfghooks.c (verify_flow_info): Check consistency of counts.
8537 (dump_bb_for_graph): Do not dump frequencies.
8538 (split_block_1): Do not update frequencies.
8539 (split_edge): Do not update frequencies.
8540 (make_forwarder_block): Do not update frequencies.
8541 (duplicate_block): Do not update frequencies.
8542 (account_profile_record): Do not update frequencies.
8543 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8544 for global heuristics.
8545 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8546 (expected_loop_iterations_unbounded): Use counts only.
8547 * cfgloopmanip.c (scale_loop_profile): Simplify.
8548 (create_empty_loop_on_edge): Simplify
8550 (duplicate_loop_to_header_edge): Simplify
8551 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8552 (update_br_prob_note): Take care of removing note when profile
8554 (relink_block_chain): Do not dump frequency.
8555 (rtl_account_profile_record): Use to_frequency.
8556 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8557 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8558 (cgraph_update_edges_for_call_stmt_node): Likewise.
8559 (cgraph_edge::verify_count_and_frequency): Update.
8560 (cgraph_node::verify_node): Temporarily disable frequency verification.
8561 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8562 to_cgraph_frequency.
8563 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8564 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8566 (cgraph_node::expand_thunk): Update profile.
8567 * except.c (dw2_build_landing_pads): Do not update frequency.
8568 * final.c (compute_alignments): Use to_frequency.
8569 (dump_basic_block_info): Do not dump frequency.
8570 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8571 (dump_gimple_bb_header): Do not dump frequency.
8572 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8574 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8575 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8576 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8577 (init_before_recovery): Do not update frequency.
8578 (sched_create_recovery_edges): Do not update frequency.
8579 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8580 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8581 (ipa_cp_c_finalize): Set max_count to uninitialized.
8582 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8583 (param_change_prob): Use counts.
8584 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8586 * ipa-split.c (consider_split): Use to_frequency.
8587 (split_function): Use to_frequency.
8588 * ira-build.c (loop_compare_func): Likewise.
8589 (mark_loops_for_removal): Likewise.
8590 (mark_all_loops_for_removal): Likewise.
8591 * loop-doloop.c (doloop_modify): Do not update frequency.
8592 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8594 * lto-streamer-in.c (input_function): Update count_max.
8595 * omp-expand.c (expand_omp_taskreg): Update count_max.
8596 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8597 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8598 (maybe_hot_count_p): Use ipa counts only.
8599 (maybe_hot_bb_p): Simplify.
8600 (maybe_hot_edge_p): Simplify.
8601 (probably_never_executed): Do not take frequency argument.
8602 (probably_never_executed_bb_p): Do not pass frequency.
8603 (probably_never_executed_edge_p): Likewise.
8604 (combine_predictions_for_bb): Check that profile is nonzero.
8605 (propagate_freq): Do not set frequency.
8606 (drop_profile): Simplify.
8607 (counts_to_freqs): Simplify.
8608 (expensive_function_p): Use to_frequency.
8609 (propagate_unlikely_bbs_forward): Simplify.
8610 (determine_unlikely_bbs): Simplify.
8611 (estimate_bb_frequencies): Add hack to silence graphite issues.
8612 (compute_function_frequency): Use ipa counts.
8613 (pass_profile::execute): Update.
8614 (rebuild_frequencies): Use counts only.
8615 (force_edge_cold): Use counts only.
8616 * profile-count.c (profile_count::dump): Dump new count types.
8617 (profile_count::differs_from_p): Check compatiblity.
8618 (profile_count::to_frequency): New function.
8619 (profile_count::to_cgraph_frequency): New function.
8620 * profile-count.h (struct function): Declare.
8621 (enum profile_quality): Add profile_guessed_local and
8622 profile_guessed_global0.
8623 (class profile_proability): Decrease number of bits to 29;
8624 update from_reg_br_prob_note and to_reg_br_prob_note.
8625 (class profile_count: Update comment; decrease number of bits
8626 to 61. Check compatibility.
8627 (profile_count::compatible_p): New private member function.
8628 (profile_count::ipa_p): New member function.
8629 (profile_count::operator<): Handle global zero correctly.
8630 (profile_count::operator>): Handle global zero correctly.
8631 (profile_count::operator<=): Handle global zero correctly.
8632 (profile_count::operator>=): Handle global zero correctly.
8633 (profile_count::nonzero_p): New member function.
8634 (profile_count::force_nonzero): New member function.
8635 (profile_count::max): New member function.
8636 (profile_count::apply_scale): Handle IPA scalling.
8637 (profile_count::guessed_local): New member function.
8638 (profile_count::global0): New member function.
8639 (profile_count::ipa): New member function.
8640 (profile_count::to_frequency): Declare.
8641 (profile_count::to_cgraph_frequency): Declare.
8642 * profile.c (OVERLAP_BASE): Delete.
8643 (compute_frequency_overlap): Delete.
8644 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8645 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8646 * sched-ebb.c (rank): Use counts only.
8647 * shrink-wrap.c (handle_simple_exit): Use counts only.
8648 (try_shrink_wrapping): Use counts only.
8649 (place_prologue_for_one_component): Use counts only.
8650 * tracer.c (find_best_predecessor): Use to_frequency.
8651 (find_trace): Use to_frequency.
8652 (tail_duplicate): Use to_frequency.
8653 * trans-mem.c (expand_transaction): Do not update frequency.
8654 * tree-call-cdce.c: Do not update frequency.
8655 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8656 (gimple_merge_blocks): Likewise.
8657 (gimple_split_edge): Likewise.
8658 (gimple_duplicate_sese_region): Likewise.
8659 (gimple_duplicate_sese_tail): Likewise.
8660 (move_sese_region_to_fn): Likewise.
8661 (gimple_account_profile_record): Likewise.
8662 (insert_cond_bb): Likewise.
8663 * tree-complex.c (expand_complex_div_wide): Likewise.
8664 * tree-eh.c (lower_resx): Update profile.
8665 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8667 (initialize_cfun): Do not initialize frequencies
8668 (freqs_to_counts): Delete.
8669 (copy_cfg_body): Ignore count parameter.
8670 (copy_body): Update.
8671 (expand_call_inline): Update count_max.
8672 (optimize_inline_calls): Update count_max.
8673 (tree_function_versioning): Update count_max.
8674 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8675 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8677 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8678 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8679 (try_peel_loop): Likewise.
8680 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8682 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8683 (tree_transform_and_unroll_loop): Do not use frequencies
8684 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8685 Use reliable prediction only.
8686 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8687 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8688 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8689 probability scaling.
8690 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8691 not update frequency
8692 (any_remaining_duplicated_blocks): Likewise.
8693 (update_profile): Likewise.
8694 (estimated_freqs_path): Delete.
8695 (freqs_to_counts_path): Delete.
8696 (clear_counts_path): Delete.
8697 (ssa_fix_duplicate_block_edges): Likewise.
8698 (duplicate_thread_path): Likewise.
8699 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8700 * tree-tailcall.c (decrease_profile): Do not update frequency.
8701 (eliminate_tail_call): Likewise.
8702 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8703 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8704 (optimize_mask_stores): Likewise.
8705 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8706 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8707 (ubsan_expand_ptr_ifn): Update profile.
8708 * value-prof.c (gimple_ic): Simplify.
8709 * value-prof.h (gimple_ic): Update prototype.
8710 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8711 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8713 (want_inline_self_recursive_call_p): Likewise.
8714 (resolve_noninline_speculation): Only cummulate defined counts.
8715 (inline_small_functions): Use nonzero_p.
8716 (ipa_inline): Do not access freed node.
8718 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8720 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8721 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8723 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8725 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8726 non-legitimate address.
8728 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8730 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8731 (*lt0_<mode>di, *lt0_<mode>si): New.
8733 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8735 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8736 TARGET_PAIRED_FLOAT.
8738 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8739 Jim Wilson <jim.wilson@linaro.org>
8741 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8742 * config/aarch64/aarch64-tune.md: Regenerated.
8743 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8744 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8746 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8748 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8749 (arc_expand_prologue): Restore blink for millicode.
8750 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8752 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8755 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8756 gen_vec_duplicate after forcing the scalar into a register.
8758 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8760 * combine (try_combine): Print the insns input to try_combine to the
8763 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8766 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8767 Remove second argument from aarch64_process_target_attr call.
8768 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8770 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8772 (aarch64_handle_attr_arch): Remove second argument.
8773 (aarch64_handle_attr_cpu): Ditto.
8774 (aarch64_handle_attr_tune): Ditto.
8775 (aarch64_handle_attr_isa_flags): Ditto.
8776 (aarch64_process_one_target_attr): Ditto.
8777 (aarch64_process_target_attr): Ditto.
8778 (aarch64_option_valid_attribute_p): Remove second argument.
8779 on aarch64_process_target_attr call.
8781 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8783 * diagnostic.c: Include "selftest-diagnostic.h".
8784 (selftest::assert_location_text): New function.
8785 (selftest::test_diagnostic_get_location_text): New function.
8786 (selftest::diagnostic_c_tests): Call it.
8788 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8790 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8791 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8792 (class selftest::test_diagnostic_context): Move to...
8793 * selftest-diagnostic.c: New file.
8794 * selftest-diagnostic.h: New file.
8796 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8798 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8799 offset range for FT32B.
8800 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8801 * config/ft32/ft32.md: Add TARGET_NOPM.
8802 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8803 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8805 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8807 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8809 2017-11-02 Jeff Law <law@redhat.com>
8811 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8812 virtual keyword on FINAL OVERRIDE members.
8814 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8815 virtuals together. Add virtual destructor.
8816 (substitute_and_fold_engine): Similarly.
8818 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8820 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8822 2017-11-02 Richard Biener <rguenther@suse.de>
8824 PR tree-optimization/82795
8825 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8827 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8829 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8831 (gcc_GAS_CHECK_FEATURE): Remove.
8832 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8833 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8834 * configure: Regenerate.
8836 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8838 * config/arc/arc.c (hwloop_optimize): Account for empty
8841 2017-11-02 Richard Biener <rguenther@suse.de>
8844 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8845 Truncate ARRAY_REF index and element size.
8847 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8849 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8851 2017-11-01 Jeff Law <law@redhat.com>
8853 * tree-ssa-ccp.c (ccp_folder): New class derived from
8854 substitute_and_fold_engine.
8855 (ccp_folder::get_value): New member function.
8856 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8857 (ccp_fold_stmt): Remove prototype.
8858 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8859 * tree-ssa-copy.c (copy_folder): New class derived from
8860 substitute_and_fold_engine.
8861 (copy_folder::get_value): Renamed from get_value.
8862 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8863 * tree-vrp.c (vrp_folder): New class derived from
8864 substitute_and_fold_engine.
8865 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8866 (vrp_folder::get_value): New member function.
8867 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8868 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8869 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8870 provide a class interface to folder/substitute routines.
8871 (ssa_prop_fold_stmt_fn): Remove typedef.
8872 (ssa_prop_get_value_fn): Likewise.
8873 (subsitute_and_fold): Remove prototype.
8874 (replace_uses_in): Likewise.
8875 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8876 Renamed from replace_uses_in. Call the virtual member function
8877 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8878 (substitute_and_fold_dom_walker): Remove initialization of
8879 data member entries for calbacks. Add substitute_and_fold_engine
8880 member and initialize it.
8881 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8882 member functions for get_value, replace_phi_args_in c
8883 replace_uses_in, and fold_stmt calls.
8884 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8885 substitute_and_fold. Remove assert. Update ctor call.
8887 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8888 (ssa_prop_visit_phi_fn): Likewise.
8889 (class ssa_propagation_engine): New class to provide an interface
8891 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8893 (ssa_prop_visit_phi): Likewise.
8894 (ssa_propagation_engine::simulate_stmt): Moved into class.
8895 Call visit_phi/visit_stmt from the class rather than via
8896 file scoped static variables.
8897 (ssa_propagation_engine::simulate_block): Moved into class.
8898 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8899 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8900 set file scoped statics for the visit_stmt/visit_phi callbacks.
8901 * tree-complex.c (complex_propagate): New class derived from
8902 ssa_propagation_engine.
8903 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8904 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8905 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8907 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8908 ssa_propagation_engine.
8909 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8910 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8911 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8912 * tree-ssa-copy.c (copy_prop): New class derived from
8913 ssa_propagation_engine.
8914 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8915 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8916 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8917 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8918 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8919 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8920 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8922 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8924 PR rtl-optimization/82778
8925 PR rtl-optimization/82597
8926 * compare-elim.c (struct comparison): Add in_a_setter field.
8927 (find_comparison_dom_walker::before_dom_children): Remove killed
8928 bitmap and df_simulate_find_defs call, instead walk the defs.
8929 Compute last_setter and initialize in_a_setter. Merge definitions
8930 with first initialization for a few variables.
8931 (try_validate_parallel): Use insn_invalid_p instead of
8932 recog_memoized. Return insn rather than just the pattern.
8933 (try_merge_compare): Fix up comment. Don't uselessly test if
8934 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8936 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8939 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8940 Alan Hayward <alan.hayward@arm.com>
8941 David Sherwood <david.sherwood@arm.com>
8943 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8944 aarch64_hard_regno_nregs to get the number of registers
8947 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8948 Alan Hayward <alan.hayward@arm.com>
8949 David Sherwood <david.sherwood@arm.com>
8951 * config/aarch64/constraints.md (Upl): Rename to...
8953 * config/aarch64/aarch64.md
8954 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8957 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8958 Alan Hayward <alan.hayward@arm.com>
8959 David Sherwood <david.sherwood@arm.com>
8961 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8962 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8965 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8966 Alan Hayward <alan.hayward@arm.com>
8967 David Sherwood <david.sherwood@arm.com>
8969 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8970 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8971 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8973 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8974 of a single element.
8975 * config/aarch64/iterators.md: Add a comment above the permute
8976 unspecs to say that they are generated directly by
8977 aarch64_expand_vec_perm_const.
8978 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8980 2017-11-01 Nathan Sidwell <nathan@acm.org>
8982 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8984 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8986 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8987 and medany code models, and describe what they do.
8989 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8991 Revert accidental duplicate:
8993 * combine.c (can_change_dest_mode): Reject changes in
8994 REGMODE_NATURAL_SIZE.
8996 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8998 PR rtl-optimization/64682
8999 PR rtl-optimization/69567
9000 PR rtl-optimization/69737
9001 PR rtl-optimization/82683
9002 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9003 register mentioned in the note, drop the note, unless it came from I3,
9004 in which case it should go to I3 again.
9006 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9008 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9009 and return false if not.
9010 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9012 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9014 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9015 range is known to be empty.
9017 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9018 Alan Hayward <alan.hayward@arm.com>
9019 David Sherwood <david.sherwood@arm.com>
9021 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9022 and CONST_VECTOR_NUNITS instead of computing the number of units from
9023 the byte sizes of the vector and element.
9024 (simplify_binary_operation_1): Likewise.
9025 (simplify_const_binary_operation): Likewise.
9026 (simplify_ternary_operation): Likewise.
9028 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9029 Alan Hayward <alan.hayward@arm.com>
9030 David Sherwood <david.sherwood@arm.com>
9032 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9033 (int_mem_offset): ...this new function.
9034 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9035 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9036 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9039 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9040 Alan Hayward <alan.hayward@arm.com>
9041 David Sherwood <david.sherwood@arm.com>
9043 * lower-subreg.c (interesting_mode_p): New function.
9044 (compute_costs, find_decomposable_subregs, decompose_register)
9045 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9046 (resolve_clobber, dump_choices): Use it.
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 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9054 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9055 Alan Hayward <alan.hayward@arm.com>
9056 David Sherwood <david.sherwood@arm.com>
9058 * alias.c (find_base_value, find_base_term): Only process integer
9059 truncations. Check the precision rather than the size.
9061 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9062 Alan Hayward <alan.hayward@arm.com>
9063 David Sherwood <david.sherwood@arm.com>
9065 * machmode.h (is_narrower_int_mode): New function
9066 * optabs.c (expand_float, expand_fix): Use it.
9067 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9069 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9070 Alan Hayward <alan.hayward@arm.com>
9071 David Sherwood <david.sherwood@arm.com>
9073 * rtl.h (narrower_subreg_mode): New function.
9074 * ira-color.c (update_costs_from_allocno): Use it.
9076 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9077 Alan Hayward <alan.hayward@arm.com>
9078 David Sherwood <david.sherwood@arm.com>
9080 * optabs-query.h (convert_optab_p): New function, split out from...
9081 (convert_optab_handler): ...here.
9082 (widening_optab_handler): Delete.
9083 (find_widening_optab_handler): Remove permit_non_widening parameter.
9084 (find_widening_optab_handler_and_mode): Likewise. Provide an
9085 override that operates on mode class wrappers.
9086 * optabs-query.c (widening_optab_handler): Delete.
9087 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9088 parameter. Assert that the two modes are the same class and that
9089 the "from" mode is narrower than the "to" mode. Use
9090 convert_optab_handler instead of widening_optab_handler.
9091 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9092 instead of widening_optab_handler.
9093 * expr.c (expand_expr_real_2): Update calls to
9094 find_widening_optab_handler.
9095 * optabs.c (expand_widen_pattern_expr): Likewise.
9096 (expand_binop_directly): Take the insn_code as a parameter.
9097 (expand_binop): Only call find_widening_optab_handler for
9098 conversion optabs; use optab_handler otherwise. Update calls
9099 to find_widening_optab_handler and expand_binop_directly.
9100 Use convert_optab_handler instead of widening_optab_handler.
9101 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9102 find_widening_optab_handler and use scalar_mode rather than
9104 (convert_plusminus_to_widen): Likewise.
9106 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9107 Alan Hayward <alan.hayward@arm.com>
9108 David Sherwood <david.sherwood@arm.com>
9110 * machmode.h (fixed_size_mode): New class.
9111 * rtl.h (get_pool_mode): Return fixed_size_mode.
9112 * gengtype.c (main): Add fixed_size_mode.
9113 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9114 (get_raw_arg_mode): Likewise.
9115 * doc/tm.texi: Regenerate.
9116 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9117 * targhooks.c (default_get_reg_raw_mode): Likewise.
9118 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9119 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9120 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9121 (msp430_get_raw_result_mode): Likewise.
9122 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9123 * dbxout.c (dbxout_parms): Require fixed-size modes.
9124 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9125 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9126 * omp-low.c (lower_oacc_reductions): Likewise.
9127 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9128 (simplify_subreg): Update accordingly.
9129 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9130 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9131 that aren't fixed-size.
9132 (get_pool_mode): Return a fixed_size_mode.
9133 (output_constant_pool_2): Take a fixed_size_mode.
9135 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9136 Alan Hayward <alan.hayward@arm.com>
9137 David Sherwood <david.sherwood@arm.com>
9139 * doc/rtl.texi (vec_series): Document.
9140 (const): Say that the operand can be a vec_series.
9141 * rtl.def (VEC_SERIES): New rtx code.
9142 * rtl.h (const_vec_series_p_1): Declare.
9143 (const_vec_series_p): New function.
9144 * emit-rtl.h (gen_const_vec_series): Declare.
9145 (gen_vec_series): Likewise.
9146 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9147 (gen_vec_series): Likewise.
9148 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9149 * simplify-rtx.c (simplify_unary_operation): Handle negations
9151 (simplify_binary_operation_series): New function.
9152 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9153 (test_vector_ops_series): New function.
9154 (test_vector_ops): Call it.
9155 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9156 gen_const_vec_series.
9157 (altivec_lvsr): Likewise.
9158 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9160 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9161 Alan Hayward <alan.hayward@arm.com>
9162 David Sherwood <david.sherwood@arm.com>
9164 * doc/rtl.texi (const): Update description of address constants.
9165 Say that vector constants are allowed too.
9166 * common.md (E, F): Use CONSTANT_P instead of checking for
9168 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9169 checking for CONST_VECTOR.
9170 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9172 * expr.c (expand_expr_real_2): Check for vector modes instead
9173 of checking for CONST_VECTOR.
9174 * rtl.h (const_vec_p): New function.
9175 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9176 (unwrap_const_vec_duplicate): Handle them here too.
9178 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9179 David Malcolm <dmalcolm@redhat.com>
9180 Alan Hayward <alan.hayward@arm.com>
9181 David Sherwood <david.sherwood@arm.com>
9183 * rtl.h (vec_duplicate_p): New function.
9184 * selftest-rtl.c (assert_rtx_eq_at): New function.
9185 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9186 (assert_rtx_eq_at): Declare.
9187 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9188 * selftest-run-tests.c (selftest::run_tests): Call it.
9189 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9190 (simplify_unary_operation_1): Recursively handle vector duplicates.
9191 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9193 (simplify_subreg): Handle subregs of vector duplicates.
9194 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9195 (selftest::simplify_rtx_c_tests): New functions.
9197 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9198 Alan Hayward <alan.hayward@arm.com>
9199 David Sherwood <david.sherwood@arm.com>
9201 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9202 (gen_vec_duplicate): Likewise.
9203 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9205 (gen_const_vector): ...here.
9206 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9207 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9208 whose elements are all equal.
9209 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9210 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9211 (simplify_relational_operation): Likewise.
9212 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9214 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9215 (aarch64_expand_vector_init): Likewise.
9216 * config/arm/arm.c (neon_vdup_constant): Likewise.
9217 (neon_expand_vector_init): Likewise.
9218 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9219 (arm_block_set_unaligned_vect): Likewise.
9220 (arm_block_set_aligned_vect): Likewise.
9221 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9222 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9223 (expand_vec_perm_even_odd_pack): Likewise.
9224 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9225 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9226 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9227 gen_const_vec_duplicate.
9228 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9229 * config/mips/mips.c (mips_gen_const_int_vector): Use
9230 gen_const_vec_duplicate.
9231 (mips_expand_vector_init): Use CONST0_RTX.
9232 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9233 (define_split): Use gen_const_vec_duplicate.
9234 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9235 (define_split): Use gen_const_vec_duplicate.
9236 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9237 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9238 * config/spu/spu.c (spu_const): Likewise.
9240 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9241 Alan Hayward <alan.hayward@arm.com>
9242 David Sherwood <david.sherwood@arm.com>
9244 * combine.c (can_change_dest_mode): Reject changes in
9245 REGMODE_NATURAL_SIZE.
9247 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9249 * configure.ac (--enable-libssp): New.
9250 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9251 trying to determine target-specific default. Adjust indentation.
9252 * configure: Regenerated.
9253 * doc/install.texi (Configuration): Expand --disable-libssp
9256 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9258 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9261 2017-10-31 Martin Jambor <mjambor@suse.cz>
9264 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9266 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9268 * auto-profile.c (autofdo_source_profile::read): Use
9269 UNKNOWN_LOCATION rather than 0.
9270 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9271 (warning_at): ...this overload.
9272 (warning_at_rich_loc_n): Rename to...
9273 (warning_n): ...this overload.
9274 (error_at_rich_loc): Rename to...
9275 (error_at): ...this overload.
9276 (pedwarn_at_rich_loc): Rename to...
9277 (pedwarn): ...this overload.
9278 (permerror_at_rich_loc): Rename to...
9279 (permerror): ...this overload.
9280 (inform_at_rich_loc): Rename to...
9281 (inform): ...this overload.
9282 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9283 (diagnostic_n_impl_richloc): Rename to...
9284 (diagnostic_n_impl): ...this rich_location *-based decl.
9285 (inform_at_rich_loc): Rename to...
9286 (inform): ...this, and add an assertion.
9287 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9288 (warning_at_rich_loc): Rename to...
9289 (warning_at): ...this, and add an assertion.
9290 (warning_at_rich_loc_n): Rename to...
9291 (warning_n): ...this, and add an assertion.
9292 (warning_n): Update location_t-based implementation for removal of
9293 location_t-based diagnostic_n_impl.
9294 (pedwarn_at_rich_loc): Rename to...
9295 (pedwarn): ...this, and add an assertion.
9296 (permerror_at_rich_loc): Rename to...
9297 (permerror): ...this, and add an assertion.
9298 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9299 (error_at_rich_loc): Rename to...
9300 (error_at): ...this, and add an assertion.
9301 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9302 (driver::do_spec_on_infiles): Likewise.
9303 * substring-locations.c (format_warning_va): Update for renaming
9304 of inform_at_rich_loc.
9306 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9308 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9309 _Float<N>X built-in functions so that the variant without the
9310 "__builtin_" prefix is only enabled for the GNU C and Objective C
9311 languages when they are in non-strict ANSI/ISO mode.
9312 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9313 * target.def (floatn_builtin_p): Add a target hook to control
9314 whether _Float<N> and _Float<N>X built-in functions without the
9315 "__builtin_" prefix are enabled, and return true for C and
9316 Objective C in the default hook. Include langhooks.h in
9318 * targhooks.h (default_floatn_builtin_p): Likewise.
9319 * targhooks.c (default_floatn_builtin_p): Likewise.
9320 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9321 floatn_builtin_p target hook.
9322 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9324 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9325 Eric Botcazou <ebotcazou@adacore.com>
9327 PR rtl-optimization/81803
9328 * lra-constraints.c (curr_insn_transform): Also reload the whole
9329 register for a strict subreg no wider than a word if this is for
9330 a WORD_REGISTER_OPERATIONS target.
9332 2017-10-31 Jason Merrill <jason@redhat.com>
9334 * gdbinit.in: Skip over inlines from timevar.h.
9336 2017-10-31 Martin Liska <mliska@suse.cz>
9338 * doc/gcov.texi: Document new option.
9339 * gcov.c (print_usage): Likewise print it.
9340 (process_args): Support the argument.
9341 (format_count): New function.
9342 (format_gcov): Use the function.
9344 2017-10-31 Martin Liska <mliska@suse.cz>
9346 * gcov.c (struct name_map): do not use typedef.
9347 Define operator== and operator<.
9348 (name_search): Remove.
9349 (name_sort): Remove.
9350 (main): Do not allocate names.
9351 (process_file): Add vertical space.
9352 (generate_results): Use std::find.
9353 (release_structures): Do not release memory.
9354 (find_source): Use std::find.
9356 2017-10-31 Martin Liska <mliska@suse.cz>
9358 * gcov.c (struct line_info): Remove it's typedef.
9359 (line_info::line_info): Add proper ctor.
9360 (line_info::has_block): Do not use a typedef.
9361 (struct source_info): Do not use typedef.
9362 (circuit): Likewise.
9363 (get_cycles_count): Likewise.
9364 (output_intermediate_file): Iterate via vector iterator.
9365 (add_line_counts): Use std::vector methods.
9366 (accumulate_line_counts): Likewise.
9367 (output_lines): Likewise.
9369 2017-10-31 Martin Liska <mliska@suse.cz>
9371 * gcov.c (struct source_info): Remove typedef.
9372 (source_info::source_info): Add proper ctor.
9373 (accumulate_line_counts): Use struct, not it's typedef.
9374 (output_gcov_file): Likewise.
9375 (output_lines): Likewise.
9376 (main): Do not allocate an array.
9377 (output_intermediate_file): Use size of vector container.
9378 (process_file): Resize the vector.
9379 (generate_results): Do not preallocate, use newly added vector
9381 (release_structures): Do not release sources.
9382 (find_source): Use vector methods.
9383 (add_line_counts): Do not use typedef.
9385 2017-10-31 Martin Liska <mliska@suse.cz>
9387 * doc/gcov.texi: Document that.
9388 * gcov.c (add_line_counts): Mark lines with a non-executed
9390 (output_line_beginning): Handle such lines.
9391 (output_lines): Pass new argument.
9392 (output_intermediate_file): Print it in intermediate format.
9394 2017-10-31 Martin Liska <mliska@suse.cz>
9396 * color-macros.h: New file.
9397 * diagnostic-color.c: Factor out color related to macros to
9399 * doc/gcov.texi: Document -k option.
9400 * gcov.c (INCLUDE_STRING): Include string.h.
9401 (print_usage): Add -k option.
9402 (process_args): Parse it.
9403 (pad_count_string): New function.
9404 (output_line_beginning): Likewise.
9405 (DEFAULT_LINE_START): New macro.
9406 (output_lines): Support color output.
9408 2017-10-31 Martin Liska <mliska@suse.cz>
9410 PR gcov-profile/82633
9411 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9412 their interaction with GCOV infrastructure.
9413 * configure.ac: Add -fkeep-{inline,static}-functions to
9415 * configure: Regenerate.
9417 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9420 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9422 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9425 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9426 into a register if it does not fit into an immediate offset field.
9428 2017-10-31 Olivier Hainque <hainque@adacore.com>
9430 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9432 2017-10-31 Julia Koval <julia.koval@intel.com>
9434 * config.gcc: Add gfniintrin.h.
9435 * config/i386/gfniintrin.h: New.
9436 * config/i386/i386-builtin-types.def
9437 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9438 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9439 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9440 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9441 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9442 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9443 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9444 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9445 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9446 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9447 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9448 * config/i386/immintrin.h: Include gfniintrin.h.
9449 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9451 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9453 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9455 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9457 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9458 (arm_ashldi3_1bit): Remove pattern.
9459 (ashrdi3): Remove shift by 1 expansion.
9460 (arm_ashrdi3_1bit): Remove pattern.
9461 (lshrdi3): Remove shift by 1 expansion.
9462 (arm_lshrdi3_1bit): Remove pattern.
9463 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9464 cost of ashldi3 by 1.
9465 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9466 (<shift>di3_neon): Likewise.
9468 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9470 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9471 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9472 and (*aarch64_simd_mov<VQ:mode>).
9473 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9474 pattern alternative.
9475 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9476 attributes to match pattern alternative.
9478 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9480 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9481 * config/rs6000/emmintrin.h: New file.
9482 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9484 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9486 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9487 * (movdi_vfp_cortexa8): Remove pattern.
9489 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9491 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9493 (Specific, alpha*-dec-osf5.1): Remove.
9494 (Specific, mips-sgi-irix5): Remove.
9495 (Specific, mips-sgi-irix6): Remove.
9497 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9500 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9501 arguments to clear_bit_region_be.
9503 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9505 * gimplify.c: Include memmodel.h.
9507 2017-10-30 Martin Jambor <mjambor@suse.cz>
9509 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9510 condition whether loop should be executed at all.
9512 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9514 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9515 gimple folding of vec_madd() intrinsics.
9516 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9517 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9518 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9520 2017-10-30 Richard Biener <rguenther@suse.de>
9522 PR tree-optimization/82762
9524 2017-10-23 Richard Biener <rguenther@suse.de>
9526 PR tree-optimization/82129
9528 2017-08-01 Richard Biener <rguenther@suse.de>
9530 PR tree-optimization/81181
9531 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9532 (compute_antic): ... end of iteration here.
9534 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9536 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9537 -std=iso9899:2017 and -std=gnu17.
9538 * doc/standards.texi (C Language): Document C17 support.
9539 * doc/cpp.texi (Overview): Mention -std=c17.
9540 (Standard Predefined Macros): Document C11 and C17 values of
9541 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9542 * doc/extend.texi (Inline): Do not list individual options for
9543 standards newer than C99.
9544 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9546 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9549 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9551 * asan.c (asan_finish_file): Align asan globals array by shadow
9554 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9557 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9558 (struct store_immediate_info): Add bitregion_start and bitregion_end
9560 (store_immediate_info::store_immediate_info): Add brs and bre
9561 arguments and initialize bitregion_{start,end} from those.
9562 (struct merged_store_group): Add bitregion_start, bitregion_end,
9563 align_base and mask fields. Drop unnecessary struct keyword from
9564 struct store_immediate_info. Add do_merge method.
9565 (clear_bit_region_be): Use memset instead of loop storing zeros.
9566 (merged_store_group::do_merge): New method.
9567 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9568 stores as long as the surrounding bitregions have no gaps.
9569 (merged_store_group::merge_overlapping): Use do_merge.
9570 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9571 is byte aligned, rather than requiring that start and width are
9572 byte aligned. Drop unnecessary struct keyword from
9573 struct store_immediate_info. Allocate and populate also mask array.
9574 Make start of the arrays relative to bitregion_start rather than
9575 start and size them according to bitregion_{end,start} difference.
9576 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9577 struct store_immediate_info.
9578 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9579 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9580 struct keyword from struct store_immediate_info.
9581 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9582 between stores as long as the surrounding bitregions have no gaps.
9584 (struct split_store): Add orig non-static data member.
9585 (split_store::split_store): Initialize orig to false.
9586 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9587 if there is exactly a single original stmt. Change stmts argument
9588 to pointer from reference, if NULL, don't push anything to it. Add
9589 first argument, use it to optimize skipping over orig stmts that
9590 are known to be before bitpos already. Simplify.
9591 (split_group): Return unsigned int count how many stores are or
9592 would be needed rather than a bool. Add allow_unaligned argument.
9593 Change split_stores argument from reference to pointer, if NULL,
9594 only do a dry run computing how many stores would be produced.
9595 Rewritten algorithm to use both alignment and misalign if
9596 !allow_unaligned and handle bitfield stores with gaps.
9597 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9598 from bitregion_start instead of start. Compute allow_unaligned
9599 here, if true, do 2 split_group dry runs to compute which one
9600 produces fewer stores and prefer aligned if equal. Punt if
9601 new count is bigger or equal than original before emitting any
9602 statements, rather than during that. Remove no longer needed
9603 new_ssa_names tracking. Replace num_stmts with
9604 split_stores.length (). Use 32-bit stack allocated entries
9605 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9606 if possible. Handle bitfields with gaps.
9607 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9608 Compute bitregion_{start,end} for the stores and construct
9609 store_immediate_info with that. Formatting fixes.
9611 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9614 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9615 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9617 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9619 * gimplify.c: Include tm_p.h.
9621 * common.opt (gcoff): Re-add as ignored option.
9622 (gcoff1, gcoff2, gcoff3): Likewise.
9624 * Makefile.in (OBJS): Delete sdbout.o.
9625 (GTFILES): Delete $(srcdir)/sdbout.c.
9626 * debug.h: Delete sdb_debug_hooks.
9627 * final.c: Delete sdbout.h include.
9628 (final_scan_insn): Delete SDB_DEBUG check.
9629 (rest_of_clean_state): Likewise.
9630 * output.h: Delete sdb_begin_function_line.
9633 * toplev.c: Delete sdbout.h include.
9634 (process_options): Delete SDB_DEBUG check.
9635 * tree-core.h (tree_type_common): Delete pointer field of
9637 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9638 TYPE_SYMTAB_POINTER.
9639 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9640 (TYPE_SYMTAB_IS_POINTER): Delete.
9641 (TYPE_SYMTAB_IS_DIE): Renumber.
9642 * xcoffout.c: Refer to former sdbout.c file.
9643 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9645 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9646 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9647 (Debugging Options): Delete -gcoff.
9648 (-gstabs): Delete SDB reference.
9650 (-gcoff@var{level}): Delete.
9651 * doc/passes.texi (Debugging information output): Delete SDB and
9652 sdbout.c references.
9653 * doc/tm.texi: Regenerate.
9654 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9655 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9657 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9658 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9659 (SDB_DEBUGGING_INFO): Delete.
9660 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9661 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9662 * target.def (output_source_filename): Delete COFF reference.
9664 * common.opt (gcoff): Delete.
9665 (gxcoff+): Update Negative chain.
9666 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9668 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9669 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9670 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9672 (expand_function_start): Change sdb reference to past tense.
9673 (expand_function_end): Change sdb reference to past tense.
9674 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9675 * opts.c (debug_type_names): Delete coff entry.
9676 (common_handle_option): Delete OPT_gcoff case.
9677 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9679 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9680 * config/cris/cris.h: Delete SDB reference in comment.
9681 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9682 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9683 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9684 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9686 (ix86_expand_prologue): Likewise.
9687 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9688 * config/ia64/ia64.h: Likewise.
9689 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9690 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9692 * config/mmix/mmix.h: Likewise.
9693 * config/nds32/nds32.c: Likewise.
9694 * config/stormy/storym16.h: Likewise.
9695 * config/visium/visium.h: Likewise.
9696 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9698 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9700 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9701 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9702 FRAME_POINTER_REGNUM point at high end of local var area.
9704 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9706 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9707 Move comment around. Do not reset best_edge for a copiable
9708 destination if the copy would cause a partition change.
9709 (better_edge_p): Remove redundant check.
9711 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9713 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9715 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9717 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9718 for math functions that have _Float<N> and _Float<N>X variants.
9719 (mathfn_built_in_2): Add support for math functions that have
9720 _Float<N> and _Float<N>X variants.
9721 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9722 (expand_builtin_mathfn_ternary): Add support for fma with
9723 _Float<N> and _Float<N>X variants.
9724 (expand_builtin): Likewise.
9725 (fold_builtin_3): Likewise.
9726 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9727 create math function _Float<N> and _Float<N>X variants as external
9729 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9730 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9731 the __builtin_ prefix and if not strict ansi, without the prefix.
9732 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9733 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9734 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9735 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9736 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9737 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9738 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9739 function signatures for fma _Float<N> and _Float<N>X variants.
9740 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9741 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9742 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9743 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9744 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9745 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9746 * gencfn-macros.c (print_case_cfn): Add support for math functions
9747 that have _Float<N> and _Float<N>X variants.
9748 (print_define_operator_list): Likewise.
9749 (fltfn_suffixes): Likewise.
9751 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9752 for math functions that have _Float<N> and _Float<N>X variants.
9753 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9754 and _Float<N>X variants.
9755 (COPYSIGN): Likewise.
9758 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9759 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9761 (integer_valued_read_call_p): Likewise.
9762 * fold-const-call.c (fold_const_call_ss): Likewise.
9763 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9764 _Float<N> and _Float<N>X variants.
9765 (fold_const_call_ssss): Add support for fma _Float<N> and
9766 _Float<N>X variants.
9767 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9768 support for copysign and fma _Float<N> and _Float<N>X variants.
9769 (backprop::process_builtin_call_use): Likewise.
9770 * tree-call-cdce.c (can_test_argument_range); Add support for
9771 sqrt _Float<N> and _Float<N>X variants.
9772 (edom_only_function): Likewise.
9773 (get_no_error_domain): Likewise.
9774 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9775 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9776 copysign _Float<N> and _Float<N>X variants.
9777 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9778 handled by machine independent code.
9779 (FMAF128): Likewise.
9780 * doc/cpp.texi (Common Predefined Macros): Document defining
9781 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9782 fma _Float<N> and _Float<N>X variants.
9784 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9787 * config/i386/i386-modes.def (CCFPU): Remove definition.
9788 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9789 (ix86_cc_modes_compatible): Ditto.
9790 (ix86_expand_carry_flag_compare): Ditto.
9791 (ix86_expand_int_movcc): Ditto.
9792 (ix86_expand_int_addcc): Ditto.
9793 (ix86_reverse_condition): Ditto.
9794 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9795 Return true/false for unordered/ordered fp comparisons.
9796 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9797 (ix86_prepare_fp_compare_args): Update for rename.
9798 (ix86_expand_fp_compare): Update for rename. Generate unordered
9799 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9800 (ix86_expand_sse_compare_and_jump): Ditto.
9801 * config/i386/predicates.md (fcmov_comparison_operator):
9802 Remove CCFPU mode handling.
9803 (ix86_comparison_operator): Ditto.
9804 (ix86_carry_flag_operator): Ditto.
9805 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9806 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9807 (*cmpu<mode>_cc_i387): Ditto.
9808 (FPCMP): Remove mode iterator.
9809 (unord): Remove mode attribute.
9810 (unord_subst): New define_subst transformation
9811 (unord): New define_subst attribute.
9813 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9814 (*cmpi<unord>xf_i387): Ditto.
9815 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9816 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9817 using unord_subst transformation.
9818 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9819 (round_saeonly): Also handle CCFP mode.
9820 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9821 Remove UNSPEC_SAHF unspec handling.
9823 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9825 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9827 2017-10-27 Jeff Law <law@redhat.com>
9829 * gimple-ssa-sprintf.c: Include domwalk.h.
9830 (class sprintf_dom_walker): New class, derived from dom_walker.
9831 (sprintf_dom_walker::before_dom_children): New function.
9832 (struct call_info): Moved into sprintf_dom_walker class
9833 (compute_formath_length, handle_gimple_call): Likewise.
9834 (sprintf_length::execute): Call the dominator walker rather
9835 than walking the statements.
9837 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9838 gimple statement locations.
9839 (check_array_bounds): Corresponding changes. Get the statement's
9840 location directly from wi->stmt.
9842 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9845 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9847 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9849 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9850 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9852 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9855 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9856 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9857 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9858 maybe_get_pool_constant.
9859 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9862 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9864 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9865 2.26 caveat. Update gas and gld versions.
9866 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9869 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9871 * cgraph.h (set_malloc_flag): Declare.
9872 * cgraph.c (set_malloc_flag_1): New function.
9873 (set_malloc_flag): Likewise.
9874 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9875 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9877 (read_ipa_call_summary): Add support for reading is_return_callee.
9878 (write_ipa_call_summary): Stream is_return_callee.
9879 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9880 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9881 ipa-prop.h, ipa-fnsummary.h.
9882 (pure_const_names): Change to static.
9883 (malloc_state_e): Define.
9884 (malloc_state_names): Define.
9885 (funct_state_d): Add field malloc_state.
9886 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9887 (check_retval_uses): New function.
9888 (malloc_candidate_p): Likewise.
9889 (analyze_function): Add support for malloc attribute.
9890 (pure_const_write_summary): Stream malloc_state.
9891 (pure_const_read_summary): Add support for reading malloc_state.
9892 (dump_malloc_lattice): New function.
9893 (propagate_malloc): New function.
9894 (warn_function_malloc): New function.
9895 (ipa_pure_const::execute): Call propagate_malloc and
9896 ipa_free_fn_summary.
9897 (pass_local_pure_const::execute): Add support for malloc attribute.
9898 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9899 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9901 2017-10-27 Martin Liska <mliska@suse.cz>
9903 PR gcov-profile/82457
9904 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9905 for fork-like functions to be properly instrumented.
9907 2017-10-27 Richard Biener <rguenther@suse.de>
9910 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9911 info when we redirected EH.
9913 2017-10-26 Michael Collison <michael.collison@arm.com>
9915 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9917 (<optab>_trunchf<GPI:mode>2: New pattern.
9918 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9919 * config/aarch64/iterators.md (wv): New mode attribute.
9920 (vf, VF): New mode attributes.
9921 (vgp, VGP): New mode attributes.
9922 (s): Update attribute with SImode and DImode prefixes.
9924 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9926 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9927 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9928 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9929 (nios2_option_overide): Initialize it. Don't allow R0-relative
9930 addressing with PIC.
9931 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9932 (nios2_symbolic_constant_p): Likewise.
9933 (nios2_legitimate_address_p): Likewise.
9934 (nios2_r0rel_section_name_p): New.
9935 (nios2_symbol_ref_in_r0rel_data_p): New.
9936 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9937 (r0rel_constant_p): New.
9938 (nios2_print_operand_address): Handle r0rel_constant_p.
9939 (nios2_cdx_narrow_form_p): Likewise.
9940 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9941 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9942 (Nios II Options): Document -mr0rel-sec.
9944 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9946 * config/nios2/nios2.c: Include xregex.h.
9947 (nios2_gprel_sec_regex): New.
9948 (nios2_option_overide): Initialize it. Don't allow GP-relative
9949 addressing with PIC.
9950 (nios2_small_section_name_p): Check for regex match.
9951 * config/nios2/nios2.opt (mgprel-sec=): New option.
9952 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9953 (Nios II Options): Document -mgprel-sec.
9955 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9957 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9959 2017-10-26 Tom de Vries <tom@codesourcery.com>
9961 PR tree-optimization/82707
9962 * gimple.c (gimple_copy): Fix unsharing of
9963 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9965 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9967 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9968 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9969 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9970 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9971 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9972 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9973 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9974 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9975 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9976 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9977 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9978 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9979 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9980 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9981 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9982 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9983 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9985 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9987 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9989 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9990 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9991 warning to rs6000.c. Remove the Undocumented flag, since it has
9993 (-mabi=ibmlongdouble): Likewise.
9994 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9995 already set, set the default format for long double.
9996 (rs6000_debug_reg_global): Print whether long double is IBM or
9998 (rs6000_option_override_internal): Rework setting long double
9999 format. Only warn if the user is changing the long double default
10000 and they did not use -Wno-psabi.
10001 * doc/invoke.texi (PowerPC options): Update the documentation for
10002 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10004 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10005 Alan Hayward <alan.hayward@arm.com>
10006 David Sherwood <david.sherwood@arm.com>
10008 * rtl.h (wider_subreg_mode): New function.
10009 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10010 rather than an unsigned int *.
10011 * ira-color.c (regno_max_ref_width): Replace with...
10012 (regno_max_ref_mode): ...this new variable.
10013 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10014 Use wider_subreg_mode.
10015 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10016 rather than an unsigned int *.
10017 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10018 (process_alt_operands): Likewise.
10019 (invariant_p): Likewise.
10020 * lra-spills.c (assign_mem_slot): Likewise.
10021 (add_pseudo_to_slot): Likewise.
10022 * lra.c (collect_non_operand_hard_regs): Likewise.
10023 (add_regs_to_insn_regno_info): Likewise.
10024 * reload1.c (regno_max_ref_width): Replace with...
10025 (regno_max_ref_mode): ...this new variable.
10026 (reload): Update accordingly. Update call to
10027 ira_sort_regnos_for_alter_reg.
10028 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10029 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10030 (scan_paradoxical_subregs): Likewise.
10032 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10034 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10035 (aarch64_frame): Add emit_frame_chain boolean.
10036 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10037 Move eh_return case to aarch64_layout_frame.
10038 (aarch64_layout_frame): Initialize emit_frame_chain.
10039 (aarch64_expand_prologue): Use emit_frame_chain.
10041 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10043 * config/aarch64/aarch64.c (aarch64_layout_frame):
10044 Ensure LR is always stored at the bottom of the callee-saves.
10045 Remove rarely used frame layout which saves callee-saves at top of
10046 frame, so the store of LR can be used as a valid probe in all cases.
10048 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10050 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10051 Improve unaligned TImode/TFmode base/offset split.
10053 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10054 Alan Hayward <alan.hayward@arm.com>
10055 David Sherwood <david.sherwood@arm.com>
10057 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10058 * combine.c (find_single_use_1): Likewise.
10059 (expand_field_assignment): Likewise.
10060 (move_deaths): Likewise.
10061 * lra-constraints.c (simplify_operand_subreg): Likewise.
10062 (curr_insn_transform): Likewise.
10063 * lra.c (collect_non_operand_hard_regs): Likewise.
10064 (add_regs_to_insn_regno_info): Likewise.
10065 * rtlanal.c (reg_referenced_p): Likewise.
10066 (covers_regno_no_parallel_p): Likewise.
10068 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10070 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10071 Don't print any bits outside the precision of the value.
10072 * wide-int.cc (test_printing): Add some new tests.
10074 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10076 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10077 supports -xbrace_comment option.
10078 * configure: Regenerate.
10079 * config.in: Regenerate.
10080 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10081 (ASM_CPU_SPEC): Use it.
10083 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10085 * target.def (static_rtx_alignment): New hook.
10086 * targhooks.h (default_static_rtx_alignment): Declare.
10087 * targhooks.c (default_static_rtx_alignment): New function.
10088 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10089 * doc/tm.texi: Regenerate.
10090 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10091 instead of targetm.constant_alignment. Remove call to
10092 set_mem_attributes.
10093 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10094 (cris_preferred_mininum_alignment): New function, split out from...
10095 (cris_constant_alignment): ...here.
10096 (cris_static_rtx_alignment): New function.
10097 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10099 (ix86_constant_alignment): ...here.
10100 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10101 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10102 (mmix_static_rtx_alignment): New function.
10103 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10104 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10106 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10109 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10110 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10112 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10114 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10115 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10116 mark as a sign-extending load.
10117 (local_pic_load_u): Define.
10119 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10121 PR middle-end/82062
10122 * fold-const.c (operand_equal_for_comparison_p): Also return true
10123 if ARG0 is a simple variant of ARG1 with narrower precision.
10124 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10126 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10128 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10130 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10131 scatter_static, scatter_per_elt.
10132 * x86-tune-costs.h: Add new cost entries.
10134 2017-10-25 Richard Biener <rguenther@suse.de>
10136 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10137 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10138 class pass_fre): Move to ...
10139 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10140 class pass_fre): ... here and adjust for statistics.
10142 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10145 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10146 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10147 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10148 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10151 2017-10-25 Richard Biener <rguenther@suse.de>
10153 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10154 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10155 eliminate_push_avail, eliminate_insert): Move inside...
10156 (class eliminate_dom_walker): ... this class in preparation
10158 (fini_eliminate): Remove by merging with ...
10159 (eliminate): ... this function. Adjust for class changes.
10160 (pass_pre::execute): Remove fini_eliminate call.
10161 (pass_fre::execute): Likewise.
10163 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10166 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10167 (VPERMI2, VPERMI2I): New mode iterators.
10168 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10169 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10171 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10172 mode iterator. Remove 3 old define_insn patterns.
10173 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10174 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10175 VPERMI2 mode iterator, remove the other two expanders.
10176 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10177 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10178 instructions, remove the other two patterns.
10179 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10180 mode iterator, remove the other two patterns.
10181 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10182 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10183 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10184 and adjust argument order accordingly.
10185 (ix86_expand_vec_perm): Adjust caller.
10186 (expand_vec_perm_1): Likewise.
10187 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10188 (expand_vec_perm_vpermt2_vpshub2): ... this.
10189 (ix86_expand_vec_perm_const_1): Adjust caller.
10190 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10193 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10194 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10195 (vec_shl_<mode>): Remove unused expander.
10196 (avx512bw_<shift_insn><mode>3): New define_insn.
10197 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10198 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10200 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10203 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10206 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10208 PR rtl-optimization/82396
10209 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10210 (autopref_multipass_init): Simplify initialization.
10211 (autopref_rank_data): Simplify sort order.
10212 * gcc/sched-int.h (autopref_multipass_data_): Remove
10213 multi_mem_insn_p, min_offset and max_offset.
10215 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10217 PR middle-end/60580
10218 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10219 Check special value of flag_omit_frame_pointer.
10220 (aarch64_can_eliminate): Likewise.
10221 (aarch64_override_options_after_change_1): Simplify handling of
10222 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10224 2017-10-24 Richard Biener <rguenther@suse.de>
10226 PR tree-optimization/82697
10227 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10228 zero for conditional load and unconditional store.
10230 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10232 * doc/install.texi: Document bootstrap-cet.
10234 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10237 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10238 ENDBR instruction at function entrance if function is only
10241 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10244 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10245 patterns to better describe from which operation the CF is computed.
10246 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10247 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10248 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10249 is 0, use _0 suffixed expanders instead of emitting a comparison
10252 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10254 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10255 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10257 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10259 PR middle-end/82569
10260 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10261 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10262 * loop-iv.c (iv_get_reaching_def): Likewise.
10263 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10264 variable is promoted and the partition contains undefined values.
10266 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10268 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10270 (nios2_address_cost): Define.
10271 (nios2_legitimize_address): Recognize (exp + constant) directly.
10272 (TARGET_ADDRESS_COST): Define.
10274 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10276 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10277 (nios2_symbolic_memory_operand_p): Declare.
10278 (nios2_split_large_constant): Declare.
10279 (nios2_split_symbolic_memory_operand): Declare.
10280 * config/nios2/nios2.c: Adjust includes.
10281 (nios2_symbolic_constant_allowed): New.
10282 (nios2_symbolic_constant_p): New.
10283 (nios2_plus_symbolic_constant_p): New.
10284 (nios2_valid_addr_expr_p): Recognize addresses involving
10285 symbolic constants.
10286 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10287 (nios2_symbolic_memory_operand_p): New.
10288 (nios2_large_constant_p): New.
10289 (nios2_split_large_constant): New.
10290 (nios2_split_plus_large_constant): New.
10291 (nios2_split_symbolic_memory_operand): New.
10292 (nios2_legitimize_address): Code refactoring. Handle addresses
10293 involving symbolic constants.
10294 (nios2_emit_move_sequence): Likewise.
10295 (nios2_print_operand): Improve error output.
10296 (nios2_print_operand_address): Handle LO_SUM.
10297 (nios2_cdx_narrow_form_p): Likewise.
10298 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10299 operands involving symbolic constants.
10300 (movhi_internal, movsi_internal): Likewise.
10301 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10302 (extendhisi2, extendqi<mode>2): Likewise.
10304 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10306 * tree-pass.h (PROP_rtl_split_insns): Define.
10307 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10309 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10311 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10313 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10316 * target.def (const_not_ok_for_debug_p): Default to
10317 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10318 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10319 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10320 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10321 which targetm.const_not_ok_for_debug_p returned true.
10322 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10324 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10326 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10327 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10328 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10330 (ix86_const_not_ok_for_debug_p): New function.
10331 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10332 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10334 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10337 * system.h: Conditionally include "unique-ptr.h" if
10338 INCLUDE_UNIQUE_PTR is defined.
10339 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10340 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10342 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10344 * config/rl78/rl78.md: New define_expand "subdi3".
10346 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10349 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10350 DF_REF_INSN if DF_REF_INSN_INFO is false.
10352 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10354 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10355 xmm_move instead of sse_move.
10356 (sse_store_index): New function.
10357 (ix86_register_move_cost): Be more sensible about mismatch stall;
10358 model AVX moves correctly; make difference between sse->integer and
10360 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10361 moves; make difference between SSE and AVX.
10362 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10363 and zmm_move. Increase size of sse load and store tables;
10364 add unaligned load and store tables; add ssemmx_to_integer.
10365 * x86-tune-costs.h: Update all entries according to real
10366 move latencies from Agner Fog's manual and chip documentation.
10368 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10371 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10372 * config/i386/constraints.md (Wf): New constraint.
10373 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10374 (cmp<dwi>_doubleword): Removed.
10375 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10376 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10377 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10378 expand with cmp<dwi>_doubleword. For LTU and GEU use
10379 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10381 * common.opt (gcolumn-info): Enable by default.
10382 * doc/invoke.texi (gcolumn-info): Document new default.
10384 2017-10-23 Richard Biener <rguenther@suse.de>
10386 PR tree-optimization/82672
10387 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10388 Fold the stmt if we propagated into it.
10390 2017-10-23 Richard Biener <rguenther@suse.de>
10392 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10393 (bitmap_remove_expr_from_set): ... this. All callers call this
10394 for non-constant values.
10395 (bitmap_set_subtract): Rename to...
10396 (bitmap_set_subtract_expressions): ... this. Adjust and
10398 (bitmap_set_contains_value): Remove superfluous check.
10399 (bitmap_set_replace_value): Inline into single caller ...
10400 (bitmap_value_replace_in_set): ... here and simplify.
10401 (dependent_clean): Merge into ...
10402 (clean): ... this using an overload. Adjust.
10403 (prune_clobbered_mems): Adjust.
10404 (compute_antic_aux): Likewise.
10405 (compute_partial_antic_aux): Likewise.
10407 2017-10-23 Richard Biener <rguenther@suse.de>
10409 PR tree-optimization/82129
10411 2017-08-01 Richard Biener <rguenther@suse.de>
10413 PR tree-optimization/81181
10414 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10415 (compute_antic): ... end of iteration here.
10417 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10419 * target.def (starting_frame_offset): New hook.
10420 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10421 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10422 * doc/tm.texi.in: Regenerate.
10423 * hooks.h (hook_hwi_void_0): Declare.
10424 * hooks.c (hook_hwi_void_0): New function.
10425 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10426 STARTING_FRAME_OFFSET.
10427 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10428 * reload1.c (reload): Likewise.
10429 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10430 instead of STARTING_FRAME_OFFSET.
10431 * function.c (try_fit_stack_local): Likewise.
10432 (assign_stack_local_1): Likewise
10433 (instantiate_virtual_regs): Likewise.
10434 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10435 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10436 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10437 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10438 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10439 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10440 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10441 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10442 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10443 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10444 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10445 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10446 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10447 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10448 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10449 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10450 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10451 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10452 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10453 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10454 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10455 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10456 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10457 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10458 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10459 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10460 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10461 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10462 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10463 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10464 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10465 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10466 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10467 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10468 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10469 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10470 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10471 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10472 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10473 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10474 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10475 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10476 return a HOST_WIDE_INT.
10477 (avr_builtin_setjmp_frame_value): Use it instead of
10478 STARTING_FRAME_OFFSET.
10479 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10480 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10481 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10483 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10484 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10485 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10486 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10487 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10488 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10489 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10490 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10491 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10492 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10493 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10494 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10496 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10497 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10498 * config/mips/mips.c (mips_compute_frame_info): Refer to
10499 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10500 (mips_starting_frame_offset): New function.
10501 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10502 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10503 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10504 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10505 and return a HOST_WIDE_INT.
10506 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10507 (mmix_initial_elimination_offset): Refer to
10508 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10509 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10510 * config/pa/pa.c (pa_starting_frame_offset): New function.
10511 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10512 (pa_expand_prologue): Likewise.
10513 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10514 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10515 !FRAME_GROWS_DOWNWARD handling to...
10516 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10517 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10518 !FRAME_GROWS_DOWNWARD handling to...
10519 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10520 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10521 !FRAME_GROWS_DOWNWARD handling to...
10522 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10523 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10525 (rs6000_starting_frame_offset): New function.
10526 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10527 !FRAME_GROWS_DOWNWARD handling to...
10528 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10529 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10530 !FRAME_GROWS_DOWNWARD handling to...
10531 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10532 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10533 !FRAME_GROWS_DOWNWARD handling to...
10534 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10535 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10536 (rs6000_starting_frame_offset): New function.
10537 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10538 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10539 * config/vax/vax.c (vax_starting_frame_offset): New function.
10540 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10541 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10542 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10543 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10544 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10545 * system.h (STARTING_FRAME_OFFSET): Poison.
10547 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10549 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10550 SCALAR_TYPE_MODE instead of TYPE_MODE.
10552 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10553 Alan Hayward <alan.hayward@arm.com>
10554 David Sherwood <david.sherwood@arm.com>
10556 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10558 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10559 Alan Hayward <alan.hayward@arm.com>
10560 David Sherwood <david.sherwood@arm.com>
10562 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10564 2017-10-23 Richard Biener <rguenther@suse.de>
10566 PR tree-optimization/82129
10567 * tree-ssa-pre.c (bitmap_set_and): Remove.
10568 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10569 canonicalizing expressions in the set to those with lowest
10570 ID rather than taking that from the first edge.
10572 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10574 * combine.c (rtx_equal_for_field_assignment_p): Use
10575 byte_lowpart_offset.
10577 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10578 Alan Hayward <alan.hayward@arm.com>
10579 David Sherwood <david.sherwood@arm.com>
10581 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10582 to a SUBREG_PROMOTED_VAR.
10584 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10585 Alan Hayward <alan.hayward@arm.com>
10586 David Sherwood <david.sherwood@arm.com>
10588 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10589 (expand_debug_source_expr): Likewise.
10590 * combine.c (combine_simplify_rtx): Likewise.
10591 * cse.c (fold_rtx): Likewise.
10592 * optabs.c (expand_float): Likewise.
10593 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10594 (simplify_binary_operation_1): Likewise.
10596 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10597 Alan Hayward <alan.hayward@arm.com>
10598 David Sherwood <david.sherwood@arm.com>
10600 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10601 (record_promoted_value): Likewise.
10602 * expr.c (expand_expr_real_2): Likewise.
10603 * ree.c (update_reg_equal_equiv_notes): Likewise.
10604 (combine_set_extension): Likewise.
10605 * rtlanal.c (low_bitmask_len): Likewise.
10606 * simplify-rtx.c (neg_const_int): Likewise.
10607 (simplify_binary_operation_1): Likewise.
10609 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10610 Alan Hayward <alan.hayward@arm.com>
10611 David Sherwood <david.sherwood@arm.com>
10613 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10614 * regcprop.c (maybe_mode_change): Likewise.
10615 * reload1.c (alter_reg): Likewise.
10617 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10619 * inchash.h (inchash::hash::add_wide_int): New function.
10620 * lto-streamer-out.c (hash_tree): Use it.
10622 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10624 * inchash.h (inchash::hash::add_wide_int): Rename to...
10625 (inchash::hash::add_hwi): ...this.
10626 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10627 (polymorphic_call_target_hasher::hash): Likewise.
10628 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10629 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10630 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10631 * lto-streamer-out.c (hash_tree): Likewise.
10632 * optc-save-gen.awk: Likewise.
10633 * tree.c (add_expr): Likewise.
10635 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10638 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10639 for ordered inequality comparisons even with TARGET_IEEE_FP.
10641 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10644 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10645 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10646 Expand with cmp<dwi>_doubleword.
10648 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10650 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10651 List CET intrinsics.
10652 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10653 specific to -fcf-protection option.
10655 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10657 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10658 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10659 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10660 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10661 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10662 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10663 (extra_objs): Add cet.o for Linux/x86 targets.
10664 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10665 * config/i386/cet.c: New file.
10666 * config/i386/cetintrin.h: Likewise.
10667 * config/i386/t-cet: Likewise.
10668 * config/i386/cpuid.h (bit_SHSTK): New.
10669 (bit_IBT): Likewise.
10670 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10671 pass IBT and SHSTK bits.
10672 * config/i386/i386-builtin-types.def
10673 (VOID_FTYPE_UNSIGNED_PVOID): New.
10674 (VOID_FTYPE_UINT64_PVOID): Likewise.
10675 * config/i386/i386-builtin.def: Add CET intrinsics.
10676 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10677 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10678 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10679 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10681 * config/i386/i386.c (rest_of_insert_endbranch): New.
10682 (pass_data_insert_endbranch): Likewise.
10683 (pass_insert_endbranch): Likewise.
10684 (make_pass_insert_endbranch): Likewise.
10685 (ix86_notrack_prefixed_insn_p): Likewise.
10686 (ix86_target_string): Add -mibt, -mshstk flags.
10687 (ix86_option_override_internal): Add flag_cf_protection
10689 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10690 (ix86_print_operand): Add 'notrack' prefix output.
10691 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10692 (ix86_expand_builtin): Expand CET intrinsics.
10693 (x86_output_mi_thunk): Add 'endbranch' instruction.
10694 * config/i386/i386.h (TARGET_IBT): New.
10695 (TARGET_IBT_P): Likewise.
10696 (TARGET_SHSTK): Likewise.
10697 (TARGET_SHSTK_P): Likewise.
10698 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10699 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10700 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10701 (builtin_setjmp_setup): New pattern.
10702 (builtin_longjmp): Likewise.
10703 (rdssp<mode>): Likewise.
10704 (incssp<mode>): Likewise.
10705 (saveprevssp): Likewise.
10706 (rstorssp): Likewise.
10707 (wrss<mode>): Likewise.
10708 (wruss<mode>): Likewise.
10709 (setssbsy): Likewise.
10710 (clrssbsy): Likewise.
10711 (nop_endbr): Likewise.
10712 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10714 * config/i386/immintrin.h: Include <cetintrin.h>.
10715 * config/i386/linux-common.h
10716 (file_end_indicate_exec_stack_and_cet): New prototype.
10717 (TARGET_ASM_FILE_END): New.
10719 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10721 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10722 latencies instead of having separate table; make difference between
10723 integer and float costs.
10724 * i386.h (processor_costs): Remove scalar_stmt_cost,
10725 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10726 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10728 * x86-tune-costs.h: Remove entries which has been removed in
10729 procesor_costs from all tables; make cond_taken_branch_cost
10730 and cond_not_taken_branch_cost COST_N_INSNS based.
10732 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10734 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10736 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10738 * config/i386/i386.md (isa): Remove fma_avx512f.
10739 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10740 <avx512>_fmadd_<mode>_mask3<round_name>,
10741 <avx512>_fmsub_<mode>_mask<round_name>,
10742 <avx512>_fmsub_<mode>_mask3<round_name>,
10743 <avx512>_fnmadd_<mode>_mask<round_name>,
10744 <avx512>_fnmadd_<mode>_mask3<round_name>,
10745 <avx512>_fnmsub_<mode>_mask<round_name>,
10746 <avx512>_fnmsub_<mode>_mask3<round_name>,
10747 <avx512>_fmaddsub_<mode>_mask<round_name>,
10748 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10749 <avx512>_fmsubadd_<mode>_mask<round_name>,
10750 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10751 (*vec_widen_umult_even_v16si<mask_name>,
10752 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10753 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10755 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10757 * extend.texi: Add 'nocf_check' documentation.
10758 * gimple.texi: Add second parameter to
10759 gimple_build_call_from_tree.
10760 * invoke.texi: Add -fcf-protection documentation.
10761 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10763 2017-10-20 Richard Biener <rguenther@suse.de>
10765 PR tree-optimization/82473
10766 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10767 the largest input type.
10769 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10771 * c-attribs.c (handle_nocf_check_attribute): New function.
10772 (c_common_attribute_table): Add 'nocf_check' handling.
10773 * gimple-parser.c: Add second argument NULL to
10774 gimple_build_call_from_tree.
10775 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10776 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10778 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10779 * common.opt: Add fcf-protection flag.
10780 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10781 * flag-types.h: Add enum cf_protection_level.
10782 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10783 Add 'nocf_check' attribute propagation to gimple call.
10784 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10785 (gimple_build_call_from_tree): Update prototype.
10786 (gimple_call_nocf_check_p): New function.
10787 (gimple_call_set_nocf_check): Likewise.
10788 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10789 * ipa-icf.c: Add nocf_check attribute in statement hash.
10790 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10791 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10792 * toplev.c (process_options): Add flag_cf_protection handling.
10794 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10796 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10798 2017-10-20 Richard Biener <rguenther@suse.de>
10800 PR tree-optimization/82603
10801 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10802 remove false predicated stores.
10804 2017-10-20 Richard Biener <rguenther@suse.de>
10806 * graphite-isl-ast-to-gimple.c
10807 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10808 Remove return value and simplify, dump copied stmt after lhs
10810 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10811 Reduce dump verbosity.
10812 (gsi_insert_earliest): Likewise.
10813 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10814 * graphite.c (print_global_statistics): Adjust dumping.
10815 (print_graphite_scop_statistics): Likewise.
10816 (print_graphite_statistics): Do not dump loops here.
10817 (graphite_transform_loops): But here.
10819 2017-10-20 Nicolas Roche <roche@adacore.com>
10821 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10822 * configure: Regenerate.
10824 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10827 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10828 functions when optimizing replace GIMPLE_RETURN stmts with
10829 calls to __builtin_unreachable ().
10832 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10833 for -fsanitize=thread link of executables.
10834 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10835 link of executables.
10838 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10839 New mode iterators.
10840 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10841 define_insns for logical vector shifts to use VI248_AVX512BW
10842 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10843 condition, useless isa and prefix attributes. Change the first
10844 2 of these define_insns to ...
10845 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10846 define_insn for avx512vl.
10847 (<shift_insn><mode>3): ... and this, new define_insn without
10848 masking for non-avx512vl.
10851 * config/i386/sse.md (*andnot<mode>3,
10852 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10853 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10854 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10855 not applied use empty suffix even for TARGET_AVX512VL.
10856 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10857 is applied, supply evex,evex or evex,evex,evex instead of just
10860 2017-10-20 Julia Koval <julia.koval@intel.com>
10862 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10863 (OPTION_MASK_ISA_GFNI_UNSET): New.
10864 (ix86_handle_option): Handle OPT_mgfni.
10865 * config/i386/cpuid.h (bit_GFNI): New.
10866 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10867 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10868 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10869 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10870 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10871 * config/i386/i386.opt: Add mgfni.
10873 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10875 * config/msp430/msp430.c (msp430_option_override): Disable
10876 -fdelete-null-pointer-checks.
10877 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10879 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10881 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10882 of x87 and SSE instructions.
10884 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10886 * asan.c (create_cond_insert_point): Do not update edge count.
10887 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10888 (afdo_propagate_circuit): Likewise.
10889 (afdo_calculate_branch_prob): Likewise.
10890 (afdo_annotate_cfg): Likewise.
10891 * basic-block.h (struct edge_def): Remove count.
10892 (edge_def::count): New accessor.
10893 * bb-reorder.c (rotate_loop): Update.
10894 (find_traces_1_round): Update.
10895 (connect_traces): Update.
10896 (sanitize_hot_paths): Update.
10897 * cfg.c (unchecked_make_edge): Update.
10898 (make_single_succ_edge): Update.
10899 (check_bb_profile): Update.
10900 (dump_edge_info): Update.
10901 (update_bb_profile_for_threading): Update.
10902 (scale_bbs_frequencies_int): Update.
10903 (scale_bbs_frequencies_gcov_type): Update.
10904 (scale_bbs_frequencies_profile_count): Update.
10905 (scale_bbs_frequencies): Update.
10906 * cfganal.c (connect_infinite_loops_to_exit): Update.
10907 * cfgbuild.c (compute_outgoing_frequencies): Update.
10908 (find_many_sub_basic_blocks): Update.
10909 * cfgcleanup.c (try_forward_edges): Update.
10910 (try_crossjump_to_edge): Update
10911 * cfgexpand.c (expand_gimple_cond): Update
10912 (expand_gimple_tailcall): Update
10913 (construct_exit_block): Update
10914 * cfghooks.c (verify_flow_info): Update
10915 (redirect_edge_succ_nodup): Update
10916 (split_edge): Update
10917 (make_forwarder_block): Update
10918 (duplicate_block): Update
10919 (account_profile_record): Update
10920 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10921 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10922 * cfgloopmanip.c (scale_loop_profile): Update.
10924 (lv_adjust_loop_entry_edge): Update.
10925 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10926 (force_nonfallthru_and_redirect): Update.
10927 (purge_dead_edges): Update.
10928 (rtl_flow_call_edges_add): Update.
10929 * cgraphunit.c (init_lowered_empty_function): Update.
10930 (cgraph_node::expand_thunk): Update.
10931 * gimple-pretty-print.c (dump_probability): Update.
10932 (dump_edge_probability): Update.
10933 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10934 * haifa-sched.c (sched_create_recovery_edges): Update.
10935 * hsa-gen.c (convert_switch_statements): Update.
10936 * ifcvt.c (dead_or_predicable): Update.
10937 * ipa-inline-transform.c (inline_transform): Update.
10938 * ipa-split.c (split_function): Update.
10939 * ipa-utils.c (ipa_merge_profiles): Update.
10940 * loop-doloop.c (add_test): Update.
10941 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10942 * lto-streamer-in.c (input_cfg): Update.
10943 (input_function): Update.
10944 * lto-streamer-out.c (output_cfg): Update.
10945 * modulo-sched.c (sms_schedule): Update.
10946 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10947 * predict.c (maybe_hot_edge_p): Update.
10948 (unlikely_executed_edge_p): Update.
10949 (probably_never_executed_edge_p): Update.
10950 (dump_prediction): Update.
10951 (drop_profile): Update.
10952 (propagate_unlikely_bbs_forward): Update.
10953 (determine_unlikely_bbs): Update.
10954 (force_edge_cold): Update.
10955 * profile.c (compute_branch_probabilities): Update.
10956 * reg-stack.c (better_edge): Update.
10957 * shrink-wrap.c (handle_simple_exit): Update.
10958 * tracer.c (better_p): Update.
10959 * trans-mem.c (expand_transaction): Update.
10960 (split_bb_make_tm_edge): Update.
10961 * tree-call-cdce.c: Update.
10962 * tree-cfg.c (gimple_find_sub_bbs): Update.
10963 (gimple_split_edge): Update.
10964 (gimple_duplicate_sese_region): Update.
10965 (gimple_duplicate_sese_tail): Update.
10966 (gimple_flow_call_edges_add): Update.
10967 (insert_cond_bb): Update.
10968 (execute_fixup_cfg): Update.
10969 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10970 * tree-complex.c (expand_complex_div_wide): Update.
10971 * tree-eh.c (lower_resx): Update.
10972 (unsplit_eh): Update.
10973 (cleanup_empty_eh_move_lp): Update.
10974 * tree-inline.c (copy_edges_for_bb): Update.
10975 (freqs_to_counts): Update.
10976 (copy_cfg_body): Update.
10977 * tree-ssa-dce.c (remove_dead_stmt): Update.
10978 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10979 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10980 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10981 (unloop_loops): Update.
10982 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10983 * tree-ssa-loop-split.c (connect_loops): Update.
10984 (split_loop): Update.
10985 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10986 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10987 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10988 * tree-ssa-reassoc.c (branch_fixup): Update.
10989 * tree-ssa-tail-merge.c (replace_block_by): Update.
10990 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10991 (compute_path_counts): Update.
10992 (update_profile): Update.
10993 (recompute_probabilities): Update.
10994 (update_joiner_offpath_counts): Update.
10995 (estimated_freqs_path): Update.
10996 (freqs_to_counts_path): Update.
10997 (clear_counts_path): Update.
10998 (ssa_fix_duplicate_block_edges): Update.
10999 (duplicate_thread_path): Update.
11000 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11001 (case_bit_test_cmp): Update.
11002 (collect_switch_conv_info): Update.
11003 (gen_inbound_check): Update.
11004 (do_jump_if_equal): Update.
11005 (emit_cmp_and_jump_insns): Update.
11006 * tree-tailcall.c (decrease_profile): Update.
11007 (eliminate_tail_call): Update.
11008 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11009 (vect_do_peeling): Update.
11010 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11011 * ubsan.c (ubsan_expand_null_ifn): Update.
11012 (ubsan_expand_ptr_ifn): Update.
11013 * value-prof.c (gimple_divmod_fixed_value): Update.
11014 (gimple_mod_pow2): Update.
11015 (gimple_mod_subtract): Update.
11016 (gimple_ic): Update.
11017 (gimple_stringop_fixed_value): Update.
11019 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11022 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11024 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11026 PR rtl-optimization/82395
11027 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11028 based on non_spilled_static_chain_regno_p.
11030 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11032 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11033 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11034 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11036 2017-10-19 Martin Sebor <msebor@redhat.com>
11038 PR tree-optimization/82596
11039 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11041 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11043 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11044 (get_mem_refs_of_builtin_call): Likewise.
11045 * builtins.c (expand_builtin_apply): Adjust call to
11046 allocate_dynamic_stack_space.
11047 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11048 the third argument to allocate_dynamic_stack_space, otherwise -1.
11049 (expand_builtin): Deal with all alloca variants.
11050 (is_inexpensive_builtin): Likewise.
11051 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11052 * calls.c (special_function_p): Deal with all alloca variants.
11053 (initialize_argument_information): Adjust call to
11054 allocate_dynamic_stack_space.
11055 (expand_call): Likewise.
11056 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11057 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11058 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11059 use it for the stack usage computation.
11060 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11061 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11062 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11063 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11064 (in_loop_p): Remove first argument and useless check.
11065 (pass_walloca::execute): Remove useless test and adjust call to above.
11066 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11067 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11068 (gimplify_call_expr): Deal with all alloca variants.
11069 * hsa-gen.c (gen_hsa_alloca): Likewise.
11070 (gen_hsa_insns_for_call): Likewise.
11071 * ipa-pure-const.c (special_builtin_state): Likewise.
11072 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11073 * tree-object-size.c (alloc_object_size): Likewise.
11074 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11075 (call_may_clobber_ref_p_1): Likewise.
11076 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11077 (ccp_fold_stmt): Likewise.
11078 (optimize_stack_restore): Likewise.
11079 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11080 (mark_all_reaching_defs_necessary_1): Likewise.
11081 (propagate_necessity): Likewise.
11082 (eliminate_unnecessary_stmts): Likewise.
11083 * tree.c (build_common_builtin_nodes): Build
11084 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11085 (build_alloca_call_expr): New function.
11086 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11087 (CASE_BUILT_IN_ALLOCA): Likewise.
11088 (build_alloca_call_expr): Declare.
11089 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11091 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11094 * dwarf2out.c (new_die_raw): New static inline function.
11095 (new_die): Use it to create the DIE.
11096 (add_AT_external_die_ref): Likewise.
11097 (clone_die): Likewise.
11098 (clone_as_declaration): Likewise.
11099 (dwarf2out_vms_debug_main_pointer): Likewise.
11100 (base_type_die): Likewise. Remove early return for corner cases.
11101 Do not call add_pubtype on the DIE here.
11102 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11103 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11104 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11105 native order exists for base types, attach the DIE manually and call
11106 add_pubtype on it. Do not equate a reverse order DIE to the type.
11108 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11110 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11111 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11113 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11115 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11117 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11118 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11119 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11120 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11121 * opts.c (sanitizer_opts): Add builtin.
11122 * ubsan.c (instrument_builtin): New function.
11123 (pass_ubsan::execute): Call it.
11124 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11125 * doc/invoke.texi: Document -fsanitize=builtin.
11127 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11128 builtins, store max (log2 (align), 0) into uchar field instead of
11129 align into uptr field.
11130 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11131 store uchar 0 field instead of uptr 0 field.
11132 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11133 instead of passing one address of struct with 2 locations pass
11134 two addresses of structs with 1 location each.
11135 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11136 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11137 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11138 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11139 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11140 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11141 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11142 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11144 2017-10-19 Martin Liska <mliska@suse.cz>
11147 * file-find.c (remove_prefix): Remove.
11148 * file-find.h (remove_prefix): Likewise.
11149 * gcc-ar.c: Remove smartness of lookup.
11151 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11153 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11154 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11155 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11157 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11160 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11161 (setcc + and to xor + setcc): New peephole2.
11163 2017-10-19 Tom de Vries <tom@codesourcery.com>
11165 * doc/sourcebuild.texi (Test Directives, Variants of
11166 dg-require-support): Add dg-require-stack-size.
11168 2017-10-19 Martin Liska <mliska@suse.cz>
11171 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11172 that have a large alignment.
11173 (gimplify_target_expr): Likewise.
11175 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11177 PR rtl-optimization/82602
11178 * ira.c (rtx_moveable_p): Return false for volatile asm.
11180 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11183 * config/i386/i386-modes.def (CCGZ): New CC mode.
11184 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11185 * config/i386/predicates.md (ix86_comparison_operator):
11187 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11188 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11189 with double-word subtraction.
11190 (put_condition_code): Handle CCGZmode.
11192 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11194 * wide-int.cc (debug (const wide_int &)): New.
11195 (debug (const wide_int *)): New.
11196 (debug (const widest_int &)): New.
11197 (debug (const widest_int *)): New.
11199 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11201 PR middle-end/82556
11202 * lra-constraints.c (curr_insn_transform): Use non-input operand
11203 instead of output one for matched reload.
11205 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11207 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11208 (tree-ssa-loop-ivopts.h): New header file.
11209 (struct builtin_info): New fields.
11210 (classify_builtin_1): Compute and record base and offset parts for
11211 memset builtin partition by calling strip_offset.
11212 (offset_cmp, fuse_memset_builtins): New functions.
11213 (finalize_partitions): Fuse adjacent memset partitions by calling
11215 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11216 Expose the interface.
11217 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11219 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11221 PR tree-optimization/82574
11222 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11223 that data reference must be executed exactly once per iteration
11224 against the outermost loop in nest.
11225 (classify_partition): Update call to above function.
11227 2017-10-18 Richard Biener <rguenther@suse.de>
11229 PR tree-optimization/82591
11230 * graphite.c (graphite_transform_loops): Move code gen message
11232 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11233 Here. Handle scop_to_isl_ast failing.
11234 (scop_to_isl_ast): Limit the number of ISL operations.
11236 2017-10-18 Richard Biener <rguenther@suse.de>
11238 * graphite-isl-ast-to-gimple.c
11239 (translate_isl_ast_to_gimple::set_rename): Simplify.
11240 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11241 (graphite_copy_stmts_from_block): ... here.
11242 (copy_bb_and_scalar_dependences): Simplify.
11243 (add_parameters_to_ivs_params): Canonicalize.
11244 (generate_entry_out_of_ssa_copies): Simplify.
11245 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11246 by passing in ISL dimension.
11247 (parameter_index_in_region_1): Rename to ...
11248 (parameter_index_in_region): ... this.
11249 (extract_affine): Adjust assert, pass down parameter index.
11250 (add_param_constraints): Use range-info when available.
11251 (build_scop_context): Adjust.
11252 * sese.c (new_sese_info): Adjust.
11253 (free_sese_info): Likewise.
11254 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11255 Remove unused typedefs.
11256 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11258 2017-10-18 Martin Liska <mliska@suse.cz>
11260 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11262 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11264 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11265 (s390_sched_init): Do not reset s390_sched_state if we entered the
11266 current basic block via a fallthru edge and all others are unlikely.
11268 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11270 * config/s390/s390.c (NUM_SIDES): New variable.
11271 (LONGRUNNING_THRESHOLD): New variable.
11272 (LATENCY_FACTOR): New variable.
11273 (s390_sched_score): Decrease score for long-running instructions on
11275 (s390_sched_variable_issue): Perform bookkeeping for long-running
11278 2017-10-18 Richard Biener <rguenther@suse.de>
11280 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11281 Simplify with removal of the parameter rename map.
11282 (set_rename): Likewise.
11283 (should_copy_to_new_region): Likewise.
11284 (graphite_copy_stmts_from_block): Likewise.
11285 (copy_bb_and_scalar_dependences): Remove initialization of
11286 unused copied_bb_map.
11287 (copy_def): Remove.
11288 (copy_internal_parameters): Likewise.
11289 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11290 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11291 Use INTEGRAL_TYPE_P.
11292 (parameter_index_in_region_1): Rename to ...
11293 (assign_parameter_index_in_region): ... this. Assert we have
11294 a parameter we handle.
11295 (scan_tree_for_params): Adjust.
11296 * sese.h (parameter_rename_map_t): Remove.
11297 (struct sese_info_t): Remove unused parameter_rename_map and
11298 copied_bb_map members.
11299 * sese.c (new_sese_info): Adjust.
11300 (free_sese_info): Likewise.
11302 2017-10-18 Martin Liska <mliska@suse.cz>
11305 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11306 on an abnormal edge.
11308 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11310 * doc/invoke.texi (ffunction-sections and fdata-sections):
11313 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11315 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11316 the use statement can throw internally.
11318 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11320 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11321 any RTX present on the RHS of a SET.
11322 * compare-elim.c (try_eliminate_compare): Restore comment.
11324 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11326 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11327 may be also called on tcc_type nodes.
11328 * langhooks.c (lhd_tree_size): Likewise.
11330 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11332 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11333 format_warning_at_substring.
11334 (maybe_warn): Convert source_range * param to a location_t. Pass
11335 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11336 (format_directive): Remove code to extract source_ranges and
11337 source_range * in favor of just a location_t.
11338 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11340 * substring-locations.c (format_warning_va): Convert
11341 source_range * param to a location_t.
11342 (format_warning_at_substring): Likewise.
11343 * substring-locations.h (format_warning_va): Likewise.
11344 (format_warning_at_substring): Likewise.
11346 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11348 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11350 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11351 and scatter/gather ops.
11353 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11354 vec_gather_load and vec_scatter_store.
11355 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11356 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11357 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11358 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11359 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11360 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11362 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11364 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11365 Detect FTST instruction and handle its register pops. Only pop
11366 second operand if can_pop_second_op is true.
11367 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11368 set can_pop_second_op to false in the compare_for_stack_reg call.
11370 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11371 output_fp_compare for stack register operands.
11372 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11373 instructions here. Do not emit stack register pops here. Assert
11374 that FCOMPP pops next to top stack register. Rewrite function.
11376 2017-10-17 Nathan Sidwell <nathan@acm.org>
11378 PR middle-end/82577
11379 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11380 use DECL_ASSEMBLER_NAME_RAW.
11382 PR middle-end/82546
11383 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11386 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11387 Wilco Dijkstra <wilco.dijkstra@arm.com>
11389 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11390 converstion to Pmode from the buf_addr.
11392 2017-10-17 Richard Biener <rguenther@suse.de>
11394 * graphite-dependences.c (scop_get_reads_and_writes): Change
11395 output parameters to references.
11397 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11399 PR 71026/tree-optimization
11400 * fold-const.c (distribute_real_division): Removed.
11401 (fold_binary_loc): Remove calls to distribute_real_divison.
11403 2017-10-17 Richard Biener <rguenther@suse.de>
11405 * graphite-scop-detection.c
11406 (scop_detection::stmt_has_simple_data_refs_p): Always use
11407 the full nest as region.
11408 (try_generate_gimple_bb): Likewise.
11409 * sese.c (scalar_evolution_in_region): Simplify now that
11410 SCEV can handle instantiation in regions.
11411 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11412 in the non-loop part of a function if requested.
11414 2017-10-17 Richard Biener <rguenther@suse.de>
11416 PR tree-optimization/82563
11417 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11419 (graphite_regenerate_ast_isl): Call it.
11420 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11422 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11424 PR tree-optimization/82549
11425 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11426 Formatting fixes. Instead of calling make_bit_field_ref with negative
11429 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11431 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11432 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11433 _mm_maskz_reduce_ss): New.
11434 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11435 __builtin_ia32_reducess_mask): Ditto..
11436 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11437 * config/i386/sse.md (reduces<mode>): Renamed to ...
11438 (reduces<mode><mask_scalar_name>): ... this.
11439 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11441 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11442 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11444 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11446 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11447 * selftest-run-tests.c (selftest::run_tests): Call
11448 selftest::unique_ptr_tests_cc_tests.
11449 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11450 * unique-ptr-tests.cc: New file.
11452 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11455 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11457 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11458 (make_hard_regno_born, make_hard_regno_dead): Update
11459 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11461 2017-10-16 Jeff Law <law@redhat.com>
11463 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11465 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11467 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11469 2017-10-16 Olivier Hainque <hainque@adacore.com>
11471 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11472 with_cpu if we were configured for an e500v2 target cpu name.
11474 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11476 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11477 * doc/invoke.texi: Document +nodsp as a valid extension for
11480 2017-10-16 Martin Liska <mliska@suse.cz>
11482 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11483 (test_set_range): Likewise.
11484 (test_range_functions): Rename to ...
11485 (test_bit_in_range): ... this.
11486 (sbitmap_c_tests): Add new test.
11488 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11490 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11492 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11493 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11495 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11497 * config/aarch64/aarch64-builtins.c
11498 (aarch64_types_quadopu_lane_qualifiers): New.
11499 (TYPES_QUADOPU_LANE): New.
11500 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11501 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11502 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11503 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11504 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11505 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11506 (Vdottype, DOTPROD): New.
11507 (sur): Add SDOT and UDOT.
11509 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11511 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11512 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11513 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11514 Add TARGET_DOTPROD.
11515 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11516 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11517 Enable TARGET_DOTPROD.
11518 (cortex-a75.cortex-a55): Likewise.
11519 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11521 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11523 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11524 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11526 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11528 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11529 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11530 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11531 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11532 * config/arm/types.md (neon_dot, neon_dot_q): New.
11533 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11535 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11537 * config/arm/arm.h (TARGET_DOTPROD): New.
11538 * config/arm/arm.c (arm_arch_dotprod): New.
11539 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11540 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11541 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11542 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11543 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11544 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11545 * doc/invoke.texi (armv8.2-a): Document dotprod
11547 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11549 * i386.c (ix86_vec_cost): New function.
11550 (ix86_rtx_costs): Handle vector operations better.
11551 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11552 * x86-tune-costs.h: Add new costs to all tables.
11554 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11556 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11558 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11559 divsd, sqrtss and sqrtsd
11560 * x86-tune-costs.h: Add new entries to all costs.
11561 (znver1_cost): Fix to match real instruction latencies.
11563 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11564 Michael Collison <michael.collison@arm.com>
11566 * compare-elim.c: Include emit-rtl.h.
11567 (can_merge_compare_into_arith): New function.
11568 (try_validate_parallel): Likewise.
11569 (try_merge_compare): Likewise.
11570 (try_eliminate_compare): Call the above when no previous clobber
11572 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11575 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11577 PR middle-end/62263
11578 PR middle-end/82498
11579 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11580 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11582 PR middle-end/62263
11583 PR middle-end/82498
11584 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11585 to be any operand_equal_p operands. For & (B - 1) require
11586 B to be power of 2. Recognize
11587 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11589 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11592 * optabs.c (expand_memory_blockage): Fix call of
11593 targetm.have_memory_blockage.
11595 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11598 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11599 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11600 objects to extra_objs instead of overwriting it.
11602 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11604 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11605 Use any_fp_register_operand as operand[3] predicate. Simplify
11606 equality test for operands[2] and operands[4] memory location.
11607 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11608 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11609 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11610 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11611 any_fp_register_operand as operand[1] predicate. Simplify
11612 equality test for operands[0] and operands[3] memory location.
11613 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11614 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11615 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11617 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11619 * target-insns.def: Add memory_blockage.
11620 * optabs.c (expand_memory_blockage): New function.
11621 (expand_asm_memory_barrier): Rename ...
11622 (expand_asm_memory_blockage): ... to this.
11623 (expand_mem_thread_fence): Call expand_memory_blockage
11624 instead of expand_asm_memory_barrier.
11625 (expand_mem_singnal_fence): Ditto.
11626 (expand_atomic_load): Ditto.
11627 (expand_atomic_store): Ditto.
11628 * doc/md.texi (Standard Pattern Names For Generation):
11629 Document memory_blockage instruction pattern.
11631 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11633 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11634 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11635 * config/rl78/rl78.md: New define_expand "adddi3".
11637 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11639 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11642 2017-10-13 Jeff Law <law@redhat.com>
11644 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11646 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11649 * internal-fn.c (expand_mul_overflow): If both operands have
11650 the same highpart of -1 or 0 and the topmost bit of lowpart
11651 is different, overflow is if res <= 0 rather than res < 0.
11653 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11655 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11656 TARGET_P9_VECTOR code for unaligned_load case.
11658 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11660 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11662 2017-10-13 Nathan Sidwell <nathan@acm.org>
11664 * tree-core.h (tree_contains_struct): Make bool.
11665 * tree.c (tree_contains_struct): Likewise.
11666 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11667 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11668 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11669 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11670 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11672 2017-10-13 Richard Biener <rguenther@suse.de>
11674 * graphite-isl-ast-to-gimple.c
11675 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11676 parameters and dominance check.
11677 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11678 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11679 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11680 Do not update SSA form here or do intermediate IL verification.
11681 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11682 (graphite_initialize): Remove check on the number of loops in
11683 the function and inline into graphite_transform_loops.
11684 (graphite_finalize): Inline into graphite_transform_loops.
11685 (graphite_transform_loops): Perform SSA update and IL verification
11687 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11689 2017-10-13 Richard Biener <rguenther@suse.de>
11691 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11692 graphite_expression_type_precision): Avoid global constructor
11694 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11695 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11696 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11697 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11698 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11699 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11701 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11704 * config/i386/i386.h (ix86_red_zone_size): New.
11705 * config/i386/i386.md (push peephole2s): Replace
11706 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11708 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11709 Alan Hayward <alan.hayward@arm.com>
11710 David Sherwood <david.sherwood@arm.com>
11712 * combine.c (can_change_dest_mode): Reject changes in
11713 REGMODE_NATURAL_SIZE.
11715 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11716 Alan Hayward <alan.hayward@arm.com>
11717 David Sherwood <david.sherwood@arm.com>
11719 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11720 (expand_debug_source_expr): Likewise.
11721 * combine.c (combine_simplify_rtx): Likewise.
11722 * cse.c (fold_rtx): Likewise.
11723 * fwprop.c (canonicalize_address): Likewise.
11724 * targhooks.c (default_shift_truncation_mask): Likewise.
11726 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11727 Alan Hayward <alan.hayward@arm.com>
11728 David Sherwood <david.sherwood@arm.com>
11730 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11731 (widened_mode): Likewise.
11732 (expand_unop): Likewise.
11733 * ree.c (transform_ifelse): Likewise.
11734 (merge_def_and_ext): Likewise.
11735 (combine_reaching_defs): Likewise.
11736 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11738 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11739 Alan Hayward <alan.hayward@arm.com>
11740 David Sherwood <david.sherwood@arm.com>
11742 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11743 * combine.c (gen_lowpart_for_combine): Likewise.
11744 * dwarf2out.c (rtl_for_decl_location): Likewise.
11745 * final.c (alter_subreg): Likewise.
11746 * rtlhooks.c (gen_lowpart_general): Likewise.
11747 (gen_lowpart_if_possible): Likewise.
11749 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11750 Alan Hayward <alan.hayward@arm.com>
11751 David Sherwood <david.sherwood@arm.com>
11753 * calls.c (expand_call): Use subreg_lowpart_offset.
11754 * cse.c (cse_insn): Likewise.
11755 * regcprop.c (copy_value): Likewise.
11756 (copyprop_hardreg_forward_1): Likewise.
11758 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11761 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11762 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11763 =Q constraints to +Q and into insn condition add check
11764 that operands[0] and operands[1] are equal.
11765 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11766 =Q constraints to +Q and into insn condition add check
11767 that operands[0] is equal to either operands[1] or operands[2].
11770 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11771 instead of handling MINUS_EXPR twice (once for each argument),
11772 canonicalize operand order and handle just once, use rtype where
11773 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11776 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11777 any values of __C while still being pattern recognizable as a simple
11778 rotate instruction.
11780 2017-10-13 Richard Biener <rguenther@suse.de>
11782 PR tree-optimization/82451
11784 2017-10-02 Richard Biener <rguenther@suse.de>
11786 PR tree-optimization/82355
11787 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11788 a mapping for the enclosing loop but avoid generating one for
11789 the loop tree root.
11790 (copy_bb_and_scalar_dependences): Remove premature codegen
11791 error on PHIs in blocks duplicated into multiple places.
11792 * graphite-scop-detection.c
11793 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11794 in the region use it as loop and nest to analyze the DR in.
11795 (try_generate_gimple_bb): Likewise.
11796 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11797 (add_loop_constraints): For blocks in a loop not in the region
11798 create a dimension with a single iteration.
11799 * sese.h (gbb_loop_at_index): Remove assert.
11801 * cfgloop.c (loop_preheader_edge): For the loop tree root
11802 return the single successor of the entry block.
11803 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11804 Reset the SCEV hashtable and niters.
11805 * graphite-scop-detection.c
11806 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11807 assert that we only have POLYNOMIAL_CHREC that vary in loops
11808 contained in the region.
11809 (scop_detection::graphite_can_represent_expr): Adjust.
11810 (scop_detection::stmt_has_simple_data_refs_p): For loops
11811 not in the region set loop to NULL. The nest is now the
11812 entry edge to the region.
11813 (try_generate_gimple_bb): Likewise.
11814 * sese.c (scalar_evolution_in_region): Adjust for
11815 instantiate_scev change.
11816 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11817 Make nest parameter the edge into the region.
11818 (create_data_ref): Likewise.
11819 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11820 entry edge into a region and adjust instantiate_scev calls.
11821 (create_data_ref): Likewise.
11822 (graphite_find_data_references_in_stmt): Likewise.
11823 (find_data_references_in_stmt): Pass the loop preheader edge
11824 from the nest argument.
11825 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11826 parameter the edge into the region.
11827 (instantiate_parameters): Use the loop preheader edge as entry.
11828 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11830 (get_instantiated_value_entry): Make instantiate_below parameter
11831 the edge into the region.
11832 (instantiate_scev_name): Likewise. Adjust dominance checks,
11833 when we cannot use loop-based instantiation instantiate by
11834 walking use-def chains.
11835 (instantiate_scev_poly): Adjust.
11836 (instantiate_scev_binary): Likewise.
11837 (instantiate_scev_convert): Likewise.
11838 (instantiate_scev_not): Likewise.
11839 (instantiate_array_ref): Remove.
11840 (instantiate_scev_3): Likewise.
11841 (instantiate_scev_2): Likewise.
11842 (instantiate_scev_1): Likewise.
11843 (instantiate_scev_r): Do not blindly handle N-operand trees.
11844 Do not instantiate array-refs. Handle all constants and invariants.
11845 (instantiate_scev): Make instantiate_below parameter
11846 the edge into the region.
11847 (resolve_mixers): Use the loop preheader edge for the region
11848 parameter to instantiate_scev_r.
11849 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11851 2017-10-13 Richard Biener <rguenther@suse.de>
11853 PR tree-optimization/82525
11854 * graphite-isl-ast-to-gimple.c
11855 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11857 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11858 Fail code generation when we cannot represent the isl integer.
11859 (binary_op_to_tree): Elide modulo operations that are no-ops
11860 in the type we code generate. Remove now superfluous code
11863 2017-10-13 Richard Biener <rguenther@suse.de>
11865 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11866 (scop_detection::harmful_loop_in_region): Remove premature
11867 IV type restriction.
11868 (scop_detection::graphite_can_represent_scev): We can handle
11869 pointer IVs just fine.
11871 2017-10-13 Alan Modra <amodra@gmail.com>
11873 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11874 "Clobbers and Scratch Registers". Add paragraph on
11875 alternative to clobbers for scratch registers and OpenBLAS
11878 2017-10-13 Alan Modra <amodra@gmail.com>
11880 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11881 example of a memory input for a string of known length. Move
11882 commentary out of table. Add a number of new examples
11883 covering array memory inputs.
11885 2017-10-12 Martin Liska <mliska@suse.cz>
11887 PR tree-optimization/82493
11888 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11889 (test_range_functions): New function.
11890 (sbitmap_c_tests): Likewise.
11891 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11892 * selftest.h (sbitmap_c_tests): New function.
11894 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11896 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11898 * config/rs6000/amo.h: Fix spacing issue.
11900 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11903 * config/i386/i386.md (*ashl<mode>3_mask_1,
11904 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11905 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11908 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11910 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11911 (profile_probability): Set max_probability
11912 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11914 * profile-count.c (profile_probability::differs_from_p): Do not
11915 rely on max_probaiblity == 10000
11917 2017-10-12 Jeff Law <law@redhat.com>
11919 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11922 2017-10-12 Martin Sebor <msebor@redhat.com>
11926 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11927 (handle_alias_pairs): Call it.
11928 * common.opt (-Wattribute-alias): New option.
11929 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11930 * doc/invoke.texi (-Wattribute-alias): Document.
11932 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11935 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11937 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11939 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11940 (make_hard_regno_born, make_hard_regno_dead): Update
11941 bb_killed_pseudos and bb_gen_pseudos.
11943 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11945 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11947 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11949 * config/alpha/alpha.c (alpha_split_conditional_move):
11950 Use std::swap instead of manually swapping.
11951 (alpha_stdarg_optimize_hook): Ditto.
11952 (alpha_canonicalize_comparison): Ditto.
11954 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11956 * tree-loop-distribution.c (struct builtin_info): New struct.
11957 (struct partition): Refactor fields into struct builtin_info.
11958 (partition_free): Free struct builtin_info.
11959 (build_size_arg_loc, build_addr_arg_loc): Delete.
11960 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11961 information from struct builtin_info.
11962 (find_single_drs): New function refactored from classify_partition.
11963 Also moved builtin validity checks to this function.
11964 (compute_access_range, alloc_builtin): New functions.
11965 (classify_builtin_st, classify_builtin_ldst): New functions.
11966 (classify_partition): Refactor code into functions find_single_drs,
11967 classify_builtin_st and classify_builtin_ldst.
11968 (distribute_loop): Don't do runtime alias check when distributing
11970 (find_seed_stmts_for_distribution): New function.
11971 (pass_loop_distribution::execute): Refactor code finding seed
11972 stmts into above function. Support distribution for the innermost
11973 two-level loop nest. Adjust dump information.
11975 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11977 * tree-loop-distribution.c: Adjust the general comment.
11978 (NUM_PARTITION_THRESHOLD): New macro.
11979 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11980 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11981 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11982 in call to build_partition_graph.
11983 (finalize_partitions): New parameter. Make loop distribution more
11984 conservative by fusing more partitions.
11985 (distribute_loop): Don't do runtime alias check in case of loop nest
11987 (find_seed_stmts_for_distribution): New function.
11988 (prepare_perfect_loop_nest): New function.
11989 (pass_loop_distribution::execute): Refactor code finding seed stmts
11990 and loop nest into above functions. Support loop nest distribution.
11991 Adjust dump information accordingly.
11993 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11995 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11996 and set PTYPE_SEQUENTIAL for merged partition.
11998 2017-10-12 Richard Biener <rguenther@suse.de>
12000 PR tree-optimization/69728
12002 2017-09-19 Richard Biener <rguenther@suse.de>
12004 PR tree-optimization/69728
12005 * graphite-sese-to-poly.c (schedule_error): New global.
12006 (add_loop_schedule): Handle empty domain by failing the
12008 (build_original_schedule): Handle schedule_error.
12010 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12011 domain by returning an unchanged schedule.
12013 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12015 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12016 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12018 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12020 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12023 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12026 * expr.c (store_field): Don't optimize away bitsize == 0 store
12027 from CALL_EXPR with addressable return type.
12029 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12031 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12032 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12033 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12034 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12035 TARGET_ISEL instead of TARGET_ISEL<sel>.
12037 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12039 * config/rs6000/rs6000.c
12040 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12042 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12044 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12046 (reg_or_cint_operand): Fix comment.
12047 (reg_or_zero_operand): New predicate.
12048 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12049 * config/rs6000/rs6000.c (output_isel): Delete.
12050 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12051 instead of reg_or_cint_operand. Output instruction directly (not via
12053 (isel_unsigned_<mode>): Ditto.
12054 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12055 gpc_reg_operand. Add an instruction alternative for this. Output
12056 instruction directly.
12057 (*isel_reversed_unsigned_<mode>): Ditto.
12059 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12061 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12062 (TARGET_CANONICALIZE_COMPARISON): Define.
12064 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12067 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12068 Check whether the dest is REG before adding REG_EQUIV note.
12070 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12073 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12075 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12076 (make_hard_regno_born, make_hard_regno_dead): Update
12077 bb_killed_pseudos and bb_gen_pseudos.
12079 2017-10-11 Nathan Sidwell <nathan@acm.org>
12081 * incpath.h (enum incpath_kind): Name enum, prefix values.
12082 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12083 * incpath.c (heads, tails): Use INC_MAX.
12084 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12085 (merge_include_chains, split_quote_chain,
12086 register_include_chains): Update incpath_kind names.
12087 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12088 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12090 (add_framework_path, darwin_register_objc_includes): Likewise.
12091 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12093 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12095 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12096 Do not use float_operator operator predicate.
12097 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12098 * config/i386/predicates.md (float_operator): Remove predicate.
12100 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12102 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12103 (*jccxf_i387): Ditto.
12104 (*jcc<mode>_i387): Ditto.
12105 (*jccu<mode>_i387): Ditto.
12106 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12107 (*jcc_*_i387 splitters): Remove.
12108 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12109 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12110 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12113 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12115 * profile-count.h (slow_safe_scale_64bit): New function.
12116 (safe_scale_64bit): New inline.
12117 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12118 * profile-count.c: Include wide-int.h
12119 (slow_safe_scale_64bit): New.
12121 2017-10-11 Nathan Sidwell <nathan@acm.org>
12123 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12124 HAS_DECL_ASSEMBLER_NAME_P.
12125 * gimple-expr.c (gimple_decl_printable_name: Check
12126 HAS_DECL_ASSEMBLER_NAME_P too.
12127 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12128 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12129 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12130 * passes.c (rest_of_decl_compilation): Check
12131 HAS_DECL_ASSEMBLER_NAME_P too.
12132 * recog.c (verify_changes): Likewise.
12133 * tree-pretty-print.c (dump_decl_name): Likewise.
12134 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12136 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12137 (SET_DECL_ASSEMBLER_NAME): Use it.
12138 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12139 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12140 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12142 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12144 * config.gcc (i386, x86_64): Add extra objects.
12145 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12146 (ix86_min_insn_size): Declare.
12147 (ix86_issue_rate): Declare.
12148 (ix86_adjust_cost): Declare.
12149 (ia32_multipass_dfa_lookahead): Declare.
12150 (ix86_macro_fusion_p): Declare.
12151 (ix86_macro_fusion_pair_p): Declare.
12152 (ix86_bd_has_dispatch): Declare.
12153 (ix86_bd_do_dispatch): Declare.
12154 (ix86_core2i7_init_hooks): Declare.
12155 (ix86_atom_sched_reorder): Declare.
12156 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12157 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12158 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12159 (rip_relative_addr_p): Rename to ...
12160 (ix86_rip_relative_addr_p): ... this one; export.
12161 (memory_address_length): Update.
12162 (ix86_issue_rate): Move to x86-tune-sched.c.
12163 (ix86_flags_dependent): Move to x86-tune-sched.c.
12164 (ix86_agi_dependent): Move to x86-tune-sched.c.
12165 (exact_dependency_1): Move to x86-tune-sched.c.
12166 (exact_store_load_dependency): Move to x86-tune-sched.c.
12167 (ix86_adjust_cost): Move to x86-tune-sched.c.
12168 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12169 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12170 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12171 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12172 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12173 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12174 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12175 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12176 (min_insn_size): Rename to ...
12177 (ix86_min_insn_size): ... this one; export.
12178 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12179 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12180 (core2i7_first_cycle_multipass_backtrack): Move to
12181 x86-tune-sched-core.c.
12182 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12183 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12184 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12185 (ix86_avoid_jump_mispredicts): Update.
12186 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12187 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12188 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12189 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12190 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12191 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12192 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12193 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12194 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12195 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12196 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12197 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12198 (BIG): Move to ix86-tune-sched-bd.c.
12199 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12200 (enum insn_path): Move to ix86-tune-sched-bd.c.
12201 (get_mem_group): Move to ix86-tune-sched-bd.c.
12202 (is_cmp): Move to ix86-tune-sched-bd.c.
12203 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12204 (is_branch): Move to ix86-tune-sched-bd.c.
12205 (is_prefetch): Move to ix86-tune-sched-bd.c.
12206 (init_window): Move to ix86-tune-sched-bd.c.
12207 (allocate_window): Move to ix86-tune-sched-bd.c.
12208 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12209 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12210 (process_end_window): Move to ix86-tune-sched-bd.c.
12211 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12212 (find_constant): Move to ix86-tune-sched-bd.c.
12213 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12214 (has_immediate): Move to ix86-tune-sched-bd.c.
12215 (get_insn_path): Move to ix86-tune-sched-bd.c.
12216 (get_insn_group): Move to ix86-tune-sched-bd.c.
12217 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12218 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12219 (add_insn_window): Move to ix86-tune-sched-bd.c.
12220 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12221 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12222 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12223 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12224 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12225 (do_dispatch): Move to ix86-tune-sched-bd.c.
12226 (has_dispatch): Move to ix86-tune-sched-bd.c.
12227 * config/i386/t-i386: Add new object files.
12228 * config/i386/x86-tune-costs.h: New file.
12229 * config/i386/x86-tune-sched-atom.c: New file.
12230 * config/i386/x86-tune-sched-bd.c: New file.
12231 * config/i386/x86-tune-sched-core.c: New file.
12232 * config/i386/x86-tune-sched.c: New file.
12234 2017-10-11 Liu Hao <lh_mouse@126.com>
12236 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12237 the generic version below instead.
12238 (should_colorize): Recognize Windows consoles as terminals
12240 * pretty-print.c [__MINGW32__] (write_all): New function.
12241 [__MINGW32__] (find_esc_head): Likewise.
12242 [__MINGW32__] (find_esc_terminator): Likewise.
12243 [__MINGW32__] (eat_esc_sequence): Likewise.
12244 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12246 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12249 2017-10-11 Richard Biener <rguenther@suse.de>
12251 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12252 Properly call analyze_scalar_evolution with the loop of the stmt.
12254 2017-10-11 Richard Biener <rguenther@suse.de>
12256 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12257 * tree-core.h (tree_base): Add chrec_var union member.
12258 * tree.h (CHREC_VAR): Remove.
12259 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12260 * tree-chrec.h (build_polynomial_chrec): Adjust.
12261 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12262 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12264 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12266 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12267 * match.pd: ... here.
12268 ((T) X == (T) Y): Relax condition.
12270 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12272 PR tree-optimization/82472
12273 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12275 (break_alias_scc_partitions): Update postorder number.
12277 2017-10-11 Martin Liska <mliska@suse.cz>
12280 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12282 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12284 2017-10-11 Martin Liska <mliska@suse.cz>
12286 * print-rtl.c (print_insn): Move declaration of idbuf
12287 to same scope as name.
12289 2017-10-11 Martin Liska <mliska@suse.cz>
12294 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12295 volatile arguments.
12297 2017-10-11 Martin Liska <mliska@suse.cz>
12300 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12301 volatile arguments.
12303 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12305 * config.gcc (default_gnu_indirect_function): Default to yes for
12306 arm*-*-linux* with glibc.
12308 2017-10-11 Richard Biener <rguenther@suse.de>
12310 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12311 default-defs and types we do not want to analyze.
12312 (interpret_loop_phi): Replace unreachable code with an assert.
12313 (compute_scalar_evolution_in_loop): Remove and inline ...
12314 (analyze_scalar_evolution_1): ... here, replacing condition with
12315 what makes the intent clearer. Remove handling of cases
12316 get_scalar_evolution now handles.
12318 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12320 PR rtl-optimization/81434
12321 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12322 comment for main loop. In sched_group_found if, also add checks for
12323 pass and min_cost_group.
12325 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12327 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12328 (rs6000_insn_cost): New function.
12329 * config/rs6000/rs6000.md (cost): New attribute.
12331 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12332 H.J. Lu <hongjiu.lu@intel.com>
12336 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12337 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12338 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12339 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12340 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12341 OPTION_MASK_ISA_MMX and some other bit set require both
12342 mmx and the ISAs without the mmx bit.
12343 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12344 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12345 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12346 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12347 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12348 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12349 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12350 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12351 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12352 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12353 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12354 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12355 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12356 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12357 Add OPTION_MASK_ISA_MMX.
12359 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12361 * config.gcc (armv7*-*-freebsd*): New target.
12362 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12364 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12366 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12367 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12370 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12372 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12373 a template parameter.
12374 (WIDE_INT_REF_FOR): Update accordingly.
12375 * tree.h (wi::int_traits <const_tree>): Delete.
12376 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12377 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12378 (wi::tree_to_wide_ref): New typedef.
12379 (wi::to_wide): New function.
12380 * calls.c (get_size_range): Use wi::to_wide when operating on
12381 trees as wide_ints.
12382 * cgraph.c (cgraph_node::create_thunk): Likewise.
12383 * config/i386/i386.c (ix86_data_alignment): Likewise.
12384 (ix86_local_alignment): Likewise.
12385 * dbxout.c (stabstr_O): Likewise.
12386 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12387 * expr.c (const_vector_from_tree): Likewise.
12388 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12389 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12390 (fold_negate_expr_1, int_const_binop_1, const_binop)
12391 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12392 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12393 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12394 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12395 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12396 (fold_not_const, round_up_loc): Likewise.
12397 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12398 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12399 (alloca_call_type): Likewise.
12400 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12401 * godump.c (go_output_typedef): Likewise.
12402 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12403 * internal-fn.c (get_min_precision): Likewise.
12404 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12405 * ipa-polymorphic-call.c
12406 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12407 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12408 (ipa_modify_call_arguments): Likewise.
12409 * match.pd: Likewise.
12410 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12411 * print-tree.c (print_node_brief, print_node): Likewise.
12412 * stmt.c (expand_case): Likewise.
12413 * stor-layout.c (layout_type): Likewise.
12414 * tree-affine.c (tree_to_aff_combination): Likewise.
12415 * tree-cfg.c (group_case_labels_stmt): Likewise.
12416 * tree-data-ref.c (dr_analyze_indices): Likewise.
12417 (prune_runtime_alias_test_list): Likewise.
12418 * tree-dump.c (dequeue_and_dump): Likewise.
12419 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12420 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12421 * tree-pretty-print.c (dump_generic_node): Likewise.
12422 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12423 (simple_iv_with_niters): Likewise.
12424 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12425 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12426 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12427 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12428 (refine_value_range_using_guard, number_of_iterations_ne_max)
12429 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12430 (get_cst_init_from_scev, record_nonwrapping_iv)
12431 (scev_var_range_cant_overflow): Likewise.
12432 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12433 * tree-ssa-pre.c (compute_avail): Likewise.
12434 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12435 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12436 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12437 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12438 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12439 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12440 (dump_case_nodes, try_switch_expansion): Likewise.
12441 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12442 (vect_do_peeling): Likewise.
12443 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12444 * tree-vect-stmts.c (vectorizable_load): Likewise.
12445 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12446 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12447 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12448 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12449 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12450 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12451 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12452 (evrp_dom_walker::before_dom_children): Likewise.
12453 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12454 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12455 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12456 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12457 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12458 (get_range_pos_neg): Likewise.
12459 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12460 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12461 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12462 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12463 * config/avr/avr.c (avr_fold_builtin): Likewise.
12464 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12465 * config/msp430/msp430.c (msp430_attr): Likewise.
12466 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12467 * config/powerpcspe/powerpcspe-c.c
12468 (altivec_resolve_overloaded_builtin): Likewise.
12469 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12470 (rs6000_expand_ternop_builtin): Likewise.
12471 * config/rs6000/rs6000-c.c
12472 (altivec_resolve_overloaded_builtin): Likewise.
12473 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12474 (rs6000_expand_ternop_builtin): Likewise.
12475 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12477 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12479 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12480 when copying loop nest with only one inner loop.
12482 2017-10-10 Richard Biener <rguenther@suse.de>
12484 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12485 blocks if SCEV is active.
12486 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12488 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12489 (scev_initialize): Assert we are not yet initialized.
12491 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12493 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12494 inner loop's exit stmt by making it always exit the loop, otherwise
12495 we would generate an infinite empty loop.
12497 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12499 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12500 renaming variables in new preheader if it's deleted.
12502 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12504 * tree-loop-distribution.c (struct partition): Remove unused field
12505 loops of the structure.
12506 (partition_alloc, partition_free): Ditto.
12507 (build_rdg_partition_for_vertex): Ditto.
12509 2017-10-09 Jeff Law <law@redhat.com>
12511 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12512 return type to match prototype and documentation.
12514 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12516 * config/rs6000/rs6000.c (processor_costs): Move to ...
12517 * config/rs6000/rs6000.h: ... here.
12518 (rs6000_cost): Declare.
12520 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12522 * except.c (setjmp_fn): New global variable.
12523 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12524 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12525 if DONT_USE_BUILTIN_SETJMP is defined.
12527 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12529 * target.def (insn_cost): New hook.
12530 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12531 * doc/tm.texi: Regenerate.
12532 * rtlanal.c (insn_cost): Use the new hook.
12534 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12536 * combine.c (combine_validate_cost): Compute the new insn_cost,
12537 not just pattern_cost.
12538 (try_combine): Adjust comment.
12540 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12542 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12544 * combine.c (uid_insn_cost): Adjust comment.
12545 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12547 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12548 * dse.c (find_shift_sequence): Ditto.
12549 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12550 (bb_valid_for_noce_process_p): Use pattern_cost.
12551 * rtl.h (insn_rtx_cost): Delete.
12552 (pattern_cost): New prototype.
12553 (insn_cost): New prototype.
12554 * rtlanal.c (insn_rtx_cost): Rename to...
12555 (pattern_cost): ... this.
12558 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12560 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12561 (*jcc<mode>_0_r_i387): Ditto.
12562 (*jccxf_r_i387): Ditto.
12563 (*jcc<mode>_r_i387): Ditto.
12564 (*jccu<mode>_r_i387): Ditto.
12565 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12566 (*jcc): Rename from *jcc_1.
12568 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12570 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12571 deferred rescans after the lvx/stvx recombination pre-pass.
12573 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12575 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12576 memory operation instruction support.
12577 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12578 (rs6000-ibm-aix[789]*): Likewise.
12579 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12580 Document new functions.
12582 2017-10-09 Richard Biener <rguenther@suse.de>
12584 PR tree-optimization/82397
12585 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12586 equality only for semantically equal trees.
12588 2017-10-09 Richard Biener <rguenther@suse.de>
12590 PR tree-optimization/82449
12591 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12592 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12593 allow constant addresses.
12594 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12597 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12599 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12602 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12605 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12608 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12611 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12613 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12616 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12617 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12619 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12621 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12622 (WI_BINARY_PREDICATE_RESULT): Likewise.
12623 (wi::binary_traits::operator_result): New type.
12624 (wi::binary_traits::predicate_result): Likewise.
12625 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12626 (generic_wide_int::operator==, generic_wide_int::operator!=)
12627 (generic_wide_int::operator&, generic_wide_int::and_not)
12628 (generic_wide_int::operator|, generic_wide_int::or_not)
12629 (generic_wide_int::operator^, generic_wide_int::operator+
12630 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12632 (operator~, unary operator-, operator==, operator!=, operator&)
12633 (operator|, operator^, operator+, binary operator-, operator*): New
12635 * expr.c (get_inner_reference): Use wi::bit_and_not.
12636 * fold-const.c (fold_binary_loc): Likewise.
12637 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12638 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12639 (bit_value_binop): Likewise.
12640 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12641 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12642 (extract_range_from_binary_expr_1): Likewise.
12643 (masked_increment): Likewise.
12644 (simplify_bit_ops_using_ranges): Likewise.
12646 2017-10-09 Martin Jambor <mjambor@suse.cz>
12649 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12650 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12651 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12653 (hsa_fixup_mov_insn_type): New function.
12654 (hsa_op_with_type::get_in_type): Use it.
12655 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12656 immediates in an assert.
12657 (hsa_op_with_type::extend_int_to_32bit): New method.
12658 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12659 types. Convert to dest type if necessary.
12660 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12661 (reg_for_gimple_ssa): Pass false as min32int to
12662 hsa_type_for_scalar_tree_type.
12663 (gen_hsa_addr): Fixup type when creating addresable temporary.
12664 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12665 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12666 necessary. Call hsa_fixup_mov_insn_type.
12667 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12668 extend operands and convert to dest type if necessary.
12669 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12670 to dest type if necessary.
12671 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12672 if conversion nt necessary and size matches.
12673 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12674 to dest type if necessary.
12675 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12676 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12678 (gen_hsa_clrsb): Likewise.
12679 (gen_hsa_ffs): Likewise.
12680 (gen_hsa_divmod): Extend operands and convert to dest type if
12682 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12684 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12686 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12687 Remove empty default arguments. Use a brace block as output
12689 (conditional return): Ditto.
12691 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12692 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12694 (group_ending_nop): Ditto.
12695 (doloop_end): Ditto.
12696 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12697 (splitters for those): Ditto.
12699 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12701 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12702 a conditional jump (and the compare for it) so that pc_rtx is the
12704 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12705 for the deleted and renamed ctr<mode>_internal[234] patterns.
12706 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12707 Delete second conditional return pattern.
12708 (ctr<mode>_internal2): Delete this second bdnz pattern.
12709 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12710 (ctr<mode>_internal4): Delete this second bdz pattern.
12712 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12714 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12715 (always_initialized_rtx_for_ssa_name_p): New predicate.
12716 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12717 (finish_out_of_ssa): Free new field of SA.
12718 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12719 * tree-ssa-coalesce.c: Include tree-ssa.h.
12720 (get_parm_default_def_partitions): Remove extern keyword.
12721 (get_undefined_value_partitions): New function.
12722 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12723 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12724 uninitialized bits.
12725 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12727 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12729 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12731 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12733 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12734 for targets that preffer 128bit.
12736 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12738 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12740 2017-10-08 Olivier Hainque <hainque@adacore.com>
12742 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12743 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12745 (thumb_set_return_address): Likewise.
12747 2017-10-08 Olivier Hainque <hainque@adacore.com>
12749 * common/config/arm/arm-common.c (arm_except_unwind_info):
12750 Handle DWARF2_UNWIND_INFO.
12752 2017-10-07 Michael Collison <michael.collison@arm.com>
12754 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12757 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12759 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12760 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12761 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12762 defined, force the creation of a new block for a dispatch label.
12764 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12766 * invoke.texi (Wsuggest-attribute=cold): Document.
12767 * common.opt (Wsuggest-attribute=cold): New
12768 * ipa-pure-const.c (warn_function_cold): New function.
12769 * predict.c (compute_function_frequency): Use it.
12770 * predict.h (warn_function_cold): Declare.
12772 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12774 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12777 2017-10-06 Martin Liska <mliska@suse.cz>
12779 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12780 keyword for member functions.
12781 (struct sanopt_tree_couple): New struct.
12782 (struct sanopt_tree_couple_hash): New function.
12783 (struct sanopt_ctx): Add new hash_map.
12784 (has_dominating_ubsan_ptr_check): New function.
12785 (record_ubsan_ptr_check_stmt): Likewise.
12786 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12787 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12788 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12790 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12793 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12794 aarch64_simd_valid_immediate on CONST_VECTORs.
12795 (aarch64_reg_or_bic_imm): Likewise.
12797 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12799 PR rtl-optimization/82396
12800 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12802 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12804 * graphite-dependences.c (scop_get_reads): Move code to...
12805 (scop_get_must_writes): Move code to...
12806 (scop_get_may_writes): Move code to...
12807 (scop_get_reads_and_writes): ... here.
12808 (scop_get_dependences): Call scop_get_reads_and_writes.
12810 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12812 PR tree-optimization/82434
12813 * fold-const.h (can_native_encode_type_p,
12814 can_native_encode_string_p): Remove.
12815 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12816 don't encode anything, just return what would be otherwise returned.
12817 (native_encode_fixed, native_encode_complex, native_encode_vector):
12819 (native_encode_string): Likewise. Inline by hand
12820 can_native_encode_string_p.
12821 (can_native_encode_type_p): Remove.
12822 (can_native_encode_string_p): Remove.
12823 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12824 STRING_CSTs using can_native_encode_string_p, test all
12825 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12826 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12827 argument from native_encode_expr.
12828 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12829 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12832 2017-10-06 Richard Biener <rguenther@suse.de>
12834 PR tree-optimization/82397
12835 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12836 operand_equal_p but rely on data_ref_compare_tree for detecting
12838 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12839 to match up with dr_group_sort_cmp.
12841 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12844 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12846 * config/s390/s390-builtin-types.def: Regenerate.
12848 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12851 * config/s390/s390-builtin-types.def: Regenerate.
12852 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12853 Change flag from B_VXE to B_VX.
12854 (s390_vec_min_dbl): Remove B_VXE flag.
12856 2017-10-06 Richard Biener <rguenther@suse.de>
12858 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12859 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12860 translate_isl_ast_to_gimple::is_valid_rename,
12861 translate_isl_ast_to_gimple::get_rename,
12862 translate_isl_ast_to_gimple::get_def_bb_for_const,
12863 translate_isl_ast_to_gimple::get_new_name,
12864 translate_isl_ast_to_gimple::collect_all_ssa_names,
12865 translate_isl_ast_to_gimple::copy_loop_phi_args,
12866 translate_isl_ast_to_gimple::collect_all_ssa_names,
12867 translate_isl_ast_to_gimple::copy_loop_phi_args,
12868 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12869 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12870 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12871 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12872 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12873 translate_isl_ast_to_gimple::copy_cond_phi_args,
12874 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12875 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12876 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12877 translate_isl_ast_to_gimple::rename_uses,
12878 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12879 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12880 (set_rename_for_each_def): Likewise.
12881 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12882 here. Handle rewriting SCEV analyzable uses here.
12883 (copy_bb_and_scalar_dependences): Generate code for PHI
12885 (graphite_regenerate_ast_isl): Adjust.
12886 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12887 (add_write, add_read): New functions.
12888 (build_cross_bb_scalars_def): Use it and simplify.
12889 (build_cross_bb_scalars_use): Likewise.
12890 (graphite_find_cross_bb_scalar_vars): Inline into...
12891 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12892 simulating out-of-SSA. Compute liveout and add dependencies.
12893 (build_scops): Force an empty entry block.
12894 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12896 (sese_build_liveouts): Declare.
12897 (sese_trivially_empty_bb_p): Likewise.
12898 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12899 compute liveout and debug_liveout.
12900 (sese_bad_liveouts_use): Remove.
12901 (sese_reset_debug_liveouts_bb): Likewise.
12902 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12903 (sese_build_liveouts): Build liveout and debug_liveout and store
12905 (new_sese_info): Adjust.
12906 (free_sese_info): Likewise.
12907 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12908 do not build liveout here.
12909 (move_sese_in_condition): Adjust region entry.
12910 (scev_analyzable_p): Match up with chrec_apply requirements.
12911 (sese_trivially_empty_bb_p): New.
12912 * tree-into-ssa.c (get_reaching_def): Properly support generating
12913 default-defs for incremental rewrite of anonymous names.
12915 2017-10-06 Richard Biener <rguenther@suse.de>
12917 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12918 precision do not perform modulo reduction.
12920 2017-10-06 Richard Biener <rguenther@suse.de>
12922 PR tree-optimization/82436
12923 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12924 conservatively choose the vectorization factor when checking
12925 whether we can perform the required load permutation.
12926 (vect_transform_slp_perm_load): Assert when we may not fail.
12928 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12930 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12931 message for incompatible -msdata=* and -mcall-* options.
12933 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12935 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12936 rate for post-reload scheduling.
12938 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12940 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12942 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12944 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12945 to improve monte carlo in scimark.
12947 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12949 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12950 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12951 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12952 pentium4_cost, nocona_cost): Set reassociation width to 1.
12953 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12954 width to 2 for fp operations and 1 otherwise.
12955 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12957 (atom_cost): Set reassociation width to 2.
12958 (slm_cost, generic_cost): Set fp reassociation width
12959 to 2 and 1 otherwise.
12960 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12961 (core_cost): Set fp reassociation width to 4 and vector to 2.
12962 (ix86_reassociation_width): Rewrite using cost table; special case
12963 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12964 and TARGET_AVX128_OPTIMAL.
12965 * config/i386/i386.h (processor_costs): Add
12966 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12967 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12968 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12969 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12970 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12971 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12973 2017-10-05 Nathan Sidwell <nathan@acm.org>
12975 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12977 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12979 * config/arm/arm.c (arm_test_fpu_data): New.
12980 (arm_run_selftests): Call arm_test_fpu_data.
12982 2017-10-04 Nathan Sidwell <nathan@acm.org>
12984 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12986 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12988 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12990 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12991 check type for aarch64_simd_valid_immediate.
12992 (aarch64_output_simd_mov_immediate): Update prototype.
12993 (aarch64_simd_valid_immediate): Update prototype.
12994 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12995 support for ORR-immediate.
12996 (and<mode>3): modified pattern to add support for BIC-immediate.
12997 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12998 now checks for valid immediate for BIC and ORR based on new enum
13000 (aarch64_output_simd_mov_immediate): Function now used to output
13001 BIC/ORR imm as well based on new enum argument.
13002 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13004 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13005 (aarch64_reg_or_bic_imm): Likewise.
13007 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13009 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13010 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13012 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13016 PR rtl-optimization/82396
13017 * haifa-sched.c (autopref_multipass_init): Simplify
13019 (autopref_rank_data): Simplify sort order.
13020 * sched-int.h (autopref_multipass_data_): Remove
13021 multi_mem_insn_p, min_offset and max_offset.
13023 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13025 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13027 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13029 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13030 vect_doubleint_cvt.
13032 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13034 * doc/sourcebuild.texi: Document vect_long_mult.
13036 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13038 PR tree-optimization/82413
13039 * fold-const.c (build_range_check): Use widest_int when comparing
13040 the maximum ETYPE value with HIGH.
13042 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13044 PR rtl-optimization/82396
13045 * haifa-sched.c (autopref_multipass_init): Simplify
13047 (autopref_rank_data): Simplify sort order.
13048 * sched-int.h (autopref_multipass_data_): Remove
13049 multi_mem_insn_p, min_offset and max_offset.
13051 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13053 PR tree-optimization/82381
13054 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13055 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13058 PR tree-optimization/82374
13059 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13060 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13061 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13062 current_function_decl to the new decl.
13064 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13066 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13067 helper macro for IEEE float128 hardware built-in functions.
13068 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13070 (TRUNCF128_ODD): Likewise.
13071 (ADDF128_ODD): Likewise.
13072 (SUBF128_ODD): Likewise.
13073 (MULF128_ODD): Likewise.
13074 (DIVF128_ODD): Likewise.
13075 (FMAF128_ODD): Likewise.
13076 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13077 UNSPEC_TRUNC_ROUND_TO_ODD.
13078 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13079 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13080 floating point round to odd instructions.
13081 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13082 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13083 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13084 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13085 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13086 (trunc<mode>sf2_hw): Change the truncate with round to odd
13087 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13088 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13089 to odd hardware instructions.
13090 (sub<mode>3_odd): Likewise.
13091 (mul<mode>3_odd): Likewise.
13092 (div<mode>3_odd): Likewise.
13093 (sqrt<mode>2_odd): Likewise.
13094 (fma<mode>4_odd): Likewise.
13095 (fms<mode>4_odd): Likewise.
13096 (nfma<mode>4_odd): Likewise.
13097 (nfms<mode>4_odd): Likewise.
13098 (trunc<mode>df2_odd): Change the truncate with round to odd
13099 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13101 * doc/extend.texi (PowerPC built-in functions): Update documentation
13102 for existing IEEE float128-bit built-in functions. Add built-in
13103 functions that generate the IEEE 128-bit floating point round to
13106 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13108 PR rtl-optimization/77729
13109 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13110 to (X&(C1&~C2))|C2 transformations.
13112 2017-10-03 Martin Jambor <mjambor@suse.cz>
13114 PR tree-optimization/82363
13115 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13116 mismatch, mark lacc written regardless of racc.
13118 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13120 PR tree-optimization/82381
13121 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13122 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13123 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13124 fallthrough into reassoc_stmt_dominates_stmt_p.
13127 * combine.c (combine_instructions): Don't combine in unreachable
13130 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13133 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13134 function to not use the have_cpu variable. Do not set cpu_index,
13135 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13136 or the default cpu.
13137 (rs6000_valid_attribute_p): Remove duplicate initializations of
13138 old_optimize and func_optimize.
13139 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13140 (rs6000_activate_target_options): Make global.
13141 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13144 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13146 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13147 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13148 Set *poffset to 0 and *psize and *pmax_size to -1 if
13149 *poffset + *psize overflows in HOST_WIDE_INT.
13151 PR tree-optimization/82387
13152 PR tree-optimization/82388
13153 PR tree-optimization/82389
13154 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13155 instead of live_bytes non-NULL.
13157 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13160 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13163 2017-10-02 Richard Biener <rguenther@suse.de>
13165 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13166 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13167 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13169 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13171 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13172 requested precision matches the type's.
13173 * calls.c (alloc_max_size): Calculate the new candidate size as
13174 a widest_int and use wi::to_widest when comparing it with the
13175 current candidate size.
13176 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13177 zero rather than integer_zero_node.
13178 * match.pd: Check for a no-op conversion before using wi::add
13179 rather than after. Use tree_to_uhwi when summing small shift
13180 counts into an unsigned int.
13182 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13183 Alan Hayward <alan.hayward@arm.com>
13184 David Sherwood <david.sherwood@arm.com>
13187 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13188 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13189 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13190 POINTER_AND_FP_REGS.
13192 2017-10-02 Richard Biener <rguenther@suse.de>
13194 PR tree-optimization/82355
13195 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13196 a mapping for the enclosing loop but avoid generating one for
13197 the loop tree root.
13198 (copy_bb_and_scalar_dependences): Remove premature codegen
13199 error on PHIs in blocks duplicated into multiple places.
13200 * graphite-scop-detection.c
13201 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13202 in the region use it as loop and nest to analyze the DR in.
13203 (try_generate_gimple_bb): Likewise.
13204 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13205 (add_loop_constraints): For blocks in a loop not in the region
13206 create a dimension with a single iteration.
13207 * sese.h (gbb_loop_at_index): Remove assert.
13209 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13211 * omp-expand.c (adjust_context_scope): New function.
13212 (expand_parallel_call): Call adjust_context_scope.
13214 2017-10-01 Jeff Law <law@redhat.com>
13216 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13217 dom_opt_dom_walker class with direct access to private members.
13218 Add comments. Call test_for_singularity.
13219 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13220 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13221 m_dummy_cond anymore.
13222 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13224 (pass_dominator:execute): Build the dummy_cond here and pass it
13225 to the dom_opt_dom_walker ctor.
13226 (test_for_singularity): New function.
13228 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13229 Maya Rashish <coypu@sdf.org>
13231 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13232 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13233 (alpha*-*-netbsd*) Use nbsd_tm_file.
13234 (arm*-*-netbsdelf*) Likewise.
13235 (i[34567]86-*-netbsdelf*) Likewise.
13236 (x86_64-*-netbsd*) Likewise.
13237 (mips*-*-netbsd*) Likewise.
13238 (powerpc-*-netbsd*) Likewise.
13239 (sh*-*-netbsd*) Likewise.
13240 (sparc-*-netbsdelf*) Likewise.
13241 (sparc64-*-netbsd*) Likewise.
13242 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13244 (vax-*-netbsdelf*) Likewise.
13245 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13246 (UINT_FAST8_TYPE) Likewise.
13247 (INT_FAST16_TYPE) Check CHAR_FAST16.
13248 (UINT_FAST16_TYPE) Likewise.
13250 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13253 * config/i386/i386.md
13254 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13255 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13256 *divmodsi4_zext_2): New define_insn_and_split.
13257 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13258 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13259 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13260 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13261 New define_insn_and_split.
13262 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13263 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13264 operands[1] having DImode when mode is SImode.
13266 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13267 always SImode for DIV and MOD in REG_EQUAL notes.
13269 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13271 PR middle-end/82319
13272 * match.pd: Fix handling of NaNs in pattern.
13274 2017-09-29 Jeff Law <law@redhat.com>
13276 * sbitmap.c (bitmap_bit_in_range_p): New function.
13277 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13278 * tree-ssa-dse.c (live_bytes_read): New function.
13279 (dse_classify_store): Ignore reads of dead bytes.
13281 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13282 typos and whitespace errors.
13283 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13284 * config/s390/s390.c (s390_emit_prologue): Likewise.
13286 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13289 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13290 with a symbol for LRA.
13292 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13294 PR rtl-optimization/82338
13295 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13297 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13299 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13300 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13301 (qsort_chk): Declare.
13302 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13303 (qsort_chk): New function.
13305 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13307 PR tree-optimization/82337
13308 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13309 phi definition if the PHI result appears in an abnormal PHI.
13310 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13311 basis appears in an abnormal PHI.
13313 2017-09-29 Richard Biener <rguenther@suse.de>
13315 * graphite-isl-ast-to-gimple.c
13316 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13317 (binary_op_to_tree): Use it.
13318 (get_rename_from_scev): Likewise.
13319 (copy_loop_phi_nodes): Likewise.
13320 (copy_bb_and_scalar_dependences): Likewise.
13321 (translate_pending_phi_nodes): Likewise.
13323 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13326 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13327 for movabsq $(i32 << shift), r64.
13329 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13331 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13332 index when encoding %esp as %rsp to avoid 0x67 prefix.
13334 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13336 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13337 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13339 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13341 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13342 Extensions with more than 16 double VFP registers.
13343 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13344 to_clear_mask and all code related to it. Replace the remaining
13345 entry by a sbitmap and adapt code accordingly.
13347 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13349 * brig-builtins.def: Change pure attributes to const.
13351 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13353 * config.gcc (default_gnu_indirect_function): Default to yes for
13354 sparc*-*-linux* with glibc.
13356 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13358 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13359 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13360 when creating .init_array and .fini_array sections with priority
13363 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13366 * config/aarch64/aarch64.c
13367 (aarch64_builtin_support_vector_misalignment): Always return false
13368 when misalignment is unknown.
13370 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13372 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13373 this function to return false if the definition used by the swap
13374 instruction is artificial, or if the memory address from which the
13375 constant value is loaded is not represented by a base address held
13376 in a register or if the base address register is a frame or stack
13377 pointer. Additionally, return false if the base address of the
13378 loaded constant is a SYMBOL_REF but is not considered to be a
13380 (replace_swapped_load_constant): New function.
13381 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13382 loaded constant vector with a load of a swapped constant vector.
13384 2017-09-27 Carl Love <cel@us.ibm.com>
13386 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13387 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13388 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13391 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13393 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13394 first, always call autopref_rank_data otherwise.
13396 2017-09-27 Richard Biener <rguenther@suse.de>
13398 * graphite-scop-detection.c (find_scop_parameters): Move
13399 loop bound handling ...
13400 (gather_bbs::before_dom_children): ... here, avoiding the need
13401 to build scop_info->loop_nest.
13402 (record_loop_in_sese): Remove.
13403 * sese.h (sese_info_t::loop_nest): Remove.
13404 * sese.c (new_sese_info): Do not allocate loop_nest.
13405 (free_sese_info): Do not free loop_nest.
13407 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13410 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13411 lhs from calls if the lhs has addressable type.
13413 2017-09-27 Richard Biener <rguenther@suse.de>
13415 * graphite.h (scop::max_alias_set): New member.
13416 * graphite-scop-detection.c: Remove references to non-existing
13417 --param in comments.
13418 (build_alias_sets): Record the maximum alias set used for drs.
13419 (build_scops): Support zero as unlimited for
13420 --param graphite-max-arrays-per-scop.
13421 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13422 and inline into ...
13423 (build_poly_sr_1): ... here. Compute alias set based on the
13424 maximum alias set used for drs rather than
13425 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13427 2017-09-27 Richard Biener <rguenther@suse.de>
13429 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13430 --param loop-block-tile-size=0 to disable tiling.
13432 2017-09-27 Richard Biener <rguenther@suse.de>
13434 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13435 (graphite-max-nb-scop-params): Document special value zero.
13436 * domwalk.h (dom_walker::STOP): New symbolical constant.
13437 (dom_walker::dom_walker): Add optional parameter for bb to
13439 (dom_walker::~dom_walker): Declare.
13440 (dom_walker::before_dom_children): Document STOP return value.
13441 (dom_walker::m_user_bb_to_rpo): New member.
13442 (dom_walker::m_bb_to_rpo): Likewise.
13443 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13444 mapping here if not provided by the user.
13445 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13446 provided by the user.
13447 (dom_walker::STOP): Define.
13448 (dom_walker::walk): Do not compute bb to RPO mapping here.
13449 Support STOP return value from before_dom_children to stop
13451 * graphite-optimize-isl.c (optimize_isl): If the schedule
13452 is the same still generate code if -fgraphite-identity
13453 or -floop-parallelize-all are given.
13454 * graphite-scop-detection.c: Include cfganal.h.
13455 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13457 (gather_bbs::before_dom_children): Return STOP for BBs
13459 (build_scops): Compute bb to RPO mapping and pass it to
13460 the domwalk. Treat --param graphite-max-nb-scop-params=0
13461 as not limiting the number of params.
13462 * graphite.c (graphite_initialize): Remove limit on the
13463 number of basic-blocks in a function.
13464 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13465 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13466 default value of 10.
13468 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13470 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13471 Adjust code to eliminate needing to do the shift right 32-bits
13472 operation after XSCVDPSPN.
13474 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13476 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13477 ((X / Y) != 0 -> X >= Y): Likewise.
13479 2017-09-26 Carl Love <cel@us.ibm.com>
13481 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13482 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13483 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13484 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13485 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13486 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13487 definitions and overloading.
13488 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13489 statement for P9V_BUILTIN_XST_LEN_R.
13490 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13491 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13492 define_expand and define_insn for the instructions and builtins.
13493 * doc/extend.texi: Update the built-in documentation file for the new
13494 built-in functions.
13495 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13496 define_insn for the instructions
13498 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13501 * gcc/config/netbsd-protos.h: New file.
13502 * gcc/config/netbsd.c: New file.
13503 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13504 * gcc/config/t-netbsd: New file.
13505 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13506 (tmake_file) Add t-netbsd.
13507 (extra_objs) Add netbsd.o.
13509 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13513 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13515 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13517 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13518 sign extension from a vector register to a GPR by doing a 32-bit
13519 direct move and then an EXTSW.
13520 (extendsi<mode>2 splitter): Likewise.
13521 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13522 right or vector extract after doing XSCVDPSPN. Use
13523 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13525 (movdi_from_sf_zero_ext): Likewise.
13526 (reload_gpr_from_vsxsf): Likewise.
13527 (p8_mfvsrd_4_disf): Delete, no longer used.
13528 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13529 then needing to move the SFmode to a GPR to use the XSCVDPSP
13530 instruction instead of FRSP and XSCVDPSPN.
13531 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13532 it is adjacent to the other XSCVSPDP insns.
13533 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13534 SFmode to be in traditional Altivec registers.
13535 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13536 (vsx_xscvspdpn): Likewise.
13537 (vsx_xscvspdpn_scalar): Likewise.
13539 2017-09-26 Martin Jambor <mjambor@suse.cz>
13541 * tree-sra.c (compare_access_positions): Put integral types first,
13542 stabilize sorting of integral types, remove conditions putting
13543 non-full-precision integers last.
13544 (sort_and_splice_var_accesses): Disable scalarization if a
13545 non-integert would be represented by a non-full-precision integer.
13547 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13549 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13550 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13551 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13552 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13553 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13554 conditionals inside the function instead of around it. Call
13555 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13556 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13558 2017-09-26 Richard Biener <rguenther@suse.de>
13560 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13562 (scop_detection::build_scop_breadth): ... this. Removed.
13563 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13564 (scop_detection::harmful_stmt_in_bb): Likewise.
13565 (scop_detection::graphite_can_represent_stmt): Likewise.
13566 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13567 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13568 (scop_detection::can_represent_loop_1): ... this. Removed.
13569 (scop_detection::harmful_loop_in_region): Simplify after inlining
13570 the above and remove more quadraticness.
13571 (build_scops): Adjust.
13572 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13575 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13578 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13579 REGNO (base) == SP_REG if base is a REG.
13581 PR middle-end/35691
13582 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13583 if it is different SSA_NAME.
13584 (optimize_range_tests_cmp_bitwise): New function.
13585 (optimize_range_tests): Call it.
13587 2017-09-26 Richard Biener <rguenther@suse.de>
13589 PR tree-optimization/82321
13590 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13591 for the def being inside the loop.
13593 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13595 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13597 * config/s390/s390-builtins.def: Fix constraint on op4.
13599 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13601 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13602 independent expanders.
13603 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13604 ("vec_ordered", "vec_unordered"): New expanders.
13606 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13608 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13611 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13613 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13614 vec_unpacks_lo_v16qi.
13615 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13617 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13619 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13620 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13621 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13623 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13625 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13627 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13629 ("*vec_slb<mode>"): New insn pattern.
13630 ("vec_shr_<mode>"): New expander.
13631 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13632 and force the shift count operand to V16QImode.
13633 ("vec_srb<mode>"): Set shift count mode to V16QI.
13635 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13637 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13638 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13639 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13641 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13644 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13646 2017-09-26 Richard Biener <rguenther@suse.de>
13648 PR tree-optimization/82320
13649 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13652 2017-09-25 Jeff Law <law@redhat.com>
13654 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13655 prototype for new argument.
13656 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13657 mostly extracted from rs6000_emit_allocate_stack.
13658 (rs6000_emit_probe_stack_range_stack_clash): New function.
13659 (rs6000_emit_allocate_stack): Call
13660 rs6000_emit_probe_stack_range_stack_clash as needed.
13661 (rs6000_emit_probe_stack_range): Add additional argument
13662 to call to gen_probe_stack_range{si,di}.
13663 (output_probe_stack_range): New.
13664 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13665 (output_probe_stack_range_stack_clash): New.
13666 (rs6000_emit_prologue): Emit notes into dump file as requested.
13667 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13668 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13669 Add additional operand and pass it to output_probe_stack_range.
13671 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13673 PR tree-optimization/82163
13674 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13675 (checking_verify_loop_closed_ssa): New parameter.
13676 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13677 (check_loop_closed_ssa_stmt): Delete.
13678 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13679 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13680 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13683 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13685 * brig-builtins.def: Treat HSAIL barrier builtins as
13686 setjmp/longjump style functions.
13688 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13690 * target.def (constant_alignment): New hook.
13691 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13692 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13693 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13694 * doc/tm.texi: Regenerate.
13695 * targhooks.h (default_constant_alignment): Declare.
13696 (constant_alignment_word_strings): Likewise.
13697 * targhooks.c (default_constant_alignment): New function.
13698 (constant_alignment_word_strings): Likewise.
13699 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13700 instead of CONSTANT_ALIGNMENT.
13701 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13702 (force_const_mem): Likewise.
13703 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13704 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13705 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13706 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13707 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13709 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13710 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13711 constant_alignment_word_strings.
13712 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13713 (CONSTANT_ALIGNMENT): Likewise.
13714 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13715 (arm_constant_alignment): New function.
13716 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13717 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13718 constant_alignment_word_strings.
13719 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13720 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13721 constant_alignment_word_strings.
13722 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13723 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13724 (cris_constant_alignment): New function.
13725 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13726 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13727 (epiphany_constant_alignment): New function.
13728 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13729 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13730 constant_alignment_word_strings.
13731 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13732 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13733 constant_alignment_word_strings.
13734 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13735 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13736 constant_alignment_word_strings.
13737 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13738 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13739 * config/i386/i386.c (ix86_constant_alignment): Make static.
13740 Use the same interface as the target hook.
13741 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13742 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13743 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13744 constant_alignment_word_strings.
13745 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13746 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13747 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13748 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13749 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13750 constant_alignment_word_strings.
13751 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13752 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13753 constant_alignment_word_strings.
13754 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13755 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13756 constant_alignment_word_strings.
13757 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13758 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13760 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13761 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13762 * config/mips/mips.c (mips_constant_alignment): New function.
13763 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13764 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13765 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13766 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13767 (mmix_constant_alignment): Make static. Use the same interface
13768 as the target hook.
13769 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13770 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13771 constant_alignment_word_strings.
13772 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13773 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13774 constant_alignment_word_strings.
13775 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13776 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13777 constant_alignment_word_strings.
13778 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13779 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13780 (rs6000_constant_alignment): New function.
13781 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13782 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13783 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13784 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13785 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13786 (rs6000_constant_alignment): New function.
13787 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13788 * config/s390/s390.c (s390_constant_alignment): New function.
13789 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13790 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13791 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13792 constant_alignment_word_strings.
13793 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13794 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13795 (sparc_constant_alignment): New function.
13796 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13797 * config/spu/spu.c (spu_constant_alignment): New function.
13798 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13799 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13800 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13801 constant_alignment_word_strings.
13802 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13803 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13804 constant_alignment_word_strings.
13805 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13806 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13807 constant_alignment_word_strings.
13808 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13809 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13810 (visium_constant_alignment): New function.
13811 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13812 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13813 (xtensa_constant_alignment): New function.
13814 * system.h (CONSTANT_ALIGNMENT): Poison.
13816 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13818 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13819 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13820 (rs6000_builtin_valid_without_lhs): New helper function.
13821 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13822 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13824 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13826 * target.h (vec_perm_indices): Use unsigned short rather than
13828 (auto_vec_perm_indices): Likewise.
13829 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13830 Use unsigned int rather than unsigned char.
13831 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13833 2017-09-25 Richard Biener <rguenther@suse.de>
13835 * cfgloop.h (sort_sibling_loops): Declare.
13836 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13837 (sort_sibling_loops): New function sorting the sibling loop list
13839 * graphite.c (graphite_transform_loops): Sort sibling loops.
13841 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13843 * target.def (vec_perm_const_ok): Change sel parameter to
13845 * optabs-query.c (can_vec_perm_p): Update accordingly.
13846 * doc/tm.texi: Regenerate.
13847 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13848 auto_vec_perm_indices and remove separate nelt field.
13849 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13850 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13851 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13852 (aarch64_expand_vec_perm_const): Update accordingly.
13853 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13854 to vec_perm_indices.
13855 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13856 auto_vec_perm_indices and remove separate nelt field.
13857 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13858 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13859 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13861 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13862 to vec_perm_indices.
13863 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13864 sel to vec_perm_indices.
13865 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13866 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13867 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13869 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13872 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13875 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13876 on the FUNCTION_DECL function context if it has a DIE that is a
13879 2017-09-25 Richard Biener <rguenther@suse.de>
13881 PR tree-optimization/82285
13882 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13885 2017-09-25 Tom de Vries <tom@codesourcery.com>
13889 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13892 2017-09-25 Richard Biener <rguenther@suse.de>
13894 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13895 ISL errors other than isl_error_quota happen. Dump if the
13896 schedule is the same.
13897 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13898 errors instead of aborting inside ISL.
13900 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13903 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13904 of libgcc_eh for m64.
13905 * config/i386/darwin64.h: Likewise.
13907 2017-09-25 Richard Biener <rguenther@suse.de>
13909 PR middle-end/82144
13910 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13911 attribute for incomplete types nor twice for complete ones.
13913 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13916 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13917 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13920 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13923 * config/i386/i386.opt (mprefer-avx256): Use
13924 ix86_target_flags variable.
13925 * config/i386/i386.c (ix86_target_string): Move
13926 -mprefer-avx256 to flag2_opts.
13928 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13930 PR middle-end/35691
13931 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13932 and x != -1 | y != -1 into (x & y) != -1.
13934 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13936 * config.gcc: Add new case statement to set
13937 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13938 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13939 s390x-*-linux* case statements. Added aarch64 to the list of
13940 supported architectures.
13942 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13944 PR tree-optimization/82289
13945 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13946 STMT_VINFO_RELEVANT_P.
13948 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13949 Alan Hayward <alan.hayward@arm.com>
13950 David Sherwood <david.sherwood@arm.com>
13952 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13953 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13954 (extract_range_from_binary_expr_1): Don't call
13955 extract_range_from_multiplicative_op_1 if !range_int_cst_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 * target.def (preferred_vector_alignment): New hook.
13962 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13964 * doc/tm.texi: Regenerate.
13965 * targhooks.h (default_preferred_vector_alignment): Declare.
13966 * targhooks.c (default_preferred_vector_alignment): New function.
13967 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13969 (DR_TARGET_ALIGNMENT): New macro.
13970 (aligned_access_p): Update commentary.
13971 (vect_known_alignment_in_bytes): New function.
13972 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13974 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13975 Calculate the misalignment based on the target alignment rather than
13977 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13978 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13979 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13980 the target alignment, rather than masking the element misalignment
13981 with the number of elements in a vector. Also use the target
13982 alignment when calculating the maximum number of peels.
13983 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13984 instead of TYPE_ALIGN_UNIT.
13985 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13986 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13987 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13988 (vect_create_data_ref_ptr): Likewise.
13989 (vect_setup_realignment): Realign by ANDing with
13990 -DR_TARGET_MISALIGNMENT.
13991 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13992 the number of peels based on DR_TARGET_ALIGNMENT.
13993 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13994 with the guaranteed alignment boundary when deciding whether
13996 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13997 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13998 (ensure_base_align): Remove stmt_info parameter. Get the
13999 target base alignment from DR_TARGET_ALIGNMENT.
14000 (vectorizable_store): Update call accordingly. Interpret
14001 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14003 (vectorizable_load): Likewise.
14005 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14006 Alan Hayward <alan.hayward@arm.com>
14007 David Sherwood <david.sherwood@arm.com>
14009 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14010 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14011 (vect_enhance_data_refs_alignment): Likewise.
14013 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14015 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14016 error. Only quit immediately if parsing is complete.
14017 (BEGIN): Initialize fatal_err and parse_done.
14018 (begin fpu, end fpu): Check number of arguments.
14019 (begin arch, end arch): Likewise.
14020 (begin cpu, end cpu): Likewise.
14021 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14022 (optalias): Likewise.
14024 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14026 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14027 * config/arm/arm-isa.h: Delete. Move definitions to ...
14028 * arm-cpus.in: ... here. Use new feature and fgroup values.
14029 * config/arm/arm.c (arm_option_override): Use lower case for feature
14031 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14032 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14033 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14035 (print_isa_bits_for): New function.
14036 (gen_isa): New function.
14037 (gen_comm_data): Use print_isa_bits_for.
14038 (define feature): New keyword.
14039 (define fgroup): New keyword.
14040 * config/arm/t-arm (TM_H): Remove.
14041 (GTM_H): Add arm-isa.h.
14042 (arm-isa.h): Add rule to generate file.
14043 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14044 case for feature bit names.
14046 2017-09-22 Richard Biener <rguenther@suse.de>
14048 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14050 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14051 print of no dependency loops ...
14052 * graphite.c (graphite_transform_loops): ... here.
14053 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14055 (same_close_phi_node, remove_duplicate_close_phi,
14056 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14057 (canonicalize_loop_closed_ssa): ... here and simplify.
14058 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14059 (optimize_isl): Use dump_printf_loc to tell when we stopped
14060 optimizing because of an ISL timeout.
14062 2017-09-22 Richard Biener <rguenther@suse.de>
14064 PR tree-optimization/82291
14065 * tree-if-conv.c (predicate_mem_writes): Make sure to
14066 remove writes in blocks predicated with false.
14068 2017-09-22 Richard Biener <rguenther@suse.de>
14070 * sese.c: Include cfganal.h.
14071 (if_region_set_false_region): Remove.
14072 (create_if_region_on_edge): Likewise.
14073 (move_sese_in_condition): Re-implement without destroying
14076 2017-09-22 Richard Biener <rguenther@suse.de>
14078 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14079 Verify both BBs contain loop PHI nodes before dispatching to
14080 copy_loop_phi_args.
14081 (graphite_regenerate_ast_isl): Do not recompute dominators,
14082 do not verify three times. Restructure for clarity.
14083 * graphite-scop-detection.c (same_close_phi_node,
14084 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14085 defined_in_loop_p, canonicalize_loop_closed_ssa,
14086 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14087 checking and SSA rewrite, move to ...
14088 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14089 (graphite_initialize): Do not pass in ctx, do not reset the
14090 SCEV cache, compute only dominators.
14091 (graphite_transform_loops): Allocate ISL ctx after
14092 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14093 Maintain post-dominators only around build_scops.
14094 * sese.c (if_region_set_false_region): Make static. Free
14095 and recompute dominators.
14096 (move_sese_in_condition): Assert we don't get called with
14097 post-dominators computed.
14098 * sese.h (if_region_set_false_region): Remove.
14100 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14102 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14103 mode attribute for TARGET_AVX512VL.
14105 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14107 * config/i386/i386.opt (mprefer-avx256): New option.
14108 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14110 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14111 for TARGET_PREFER_AVX256.
14112 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14114 2017-09-21 Jeff Law <law@redhat.com>
14116 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14117 Fix dump output if the only stack space is for pushed registers.
14119 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14121 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14124 2017-09-21 Martin Sebor <msebor@redhat.com>
14127 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14128 code (in C++) or code that triggers warnings.
14130 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14132 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14134 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14136 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14137 * sched-rgn.c: Ditto.
14138 * sel-sched-ir.c: Ditto.
14140 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14142 * toplev.h (set_random_seed): Adjust return type.
14143 * toplev.c (init_local_tick): Move eager initialization of random_seed
14144 to get_random_seed. Adjust comment.
14145 (init_random_seed): Inline to get_random_seed, delete.
14146 (get_random_seed): Initialize random_seed lazily.
14147 (set_random_seed): Do not return previous value.
14148 (print_switch_value): Do not call get_random_seed.
14150 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14152 * cgraph.c (delete_function_version): New, broken out from...
14153 (cgraph_node::delete_function_version): ...here. Rename to
14154 cgraph_node::delete_function_version_by_decl. Update all uses.
14155 (cgraph_node::remove): Call delete_function_version.
14157 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14160 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14161 VAR_DECLs to which addressable non-volatile parameters are mapped
14162 and for id->retvar after the return value assignment. Clear
14163 id->retval and id->retbnd after inlining.
14165 2017-09-21 Richard Biener <rguenther@suse.de>
14167 PR tree-optimization/82276
14168 PR tree-optimization/82244
14169 * tree-vrp.c (build_assert_expr_for): Set
14170 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14172 (remove_range_assertions): Revert earlier change.
14174 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14177 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14179 2017-09-21 Richard Biener <rguenther@suse.de>
14181 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14182 Restore valid IL after code generation errors.
14183 * graphite.c (graphite_transform_loops): Diagnose code
14184 generation issues as MSG_MISSED_OPTIMIZATION and continue
14185 with processing SCOPs.
14187 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14188 Alan Hayward <alan.hayward@arm.com>
14189 David Sherwood <david.sherwood@arm.com>
14191 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14192 rather than choosing between plus_constant and gen_rtx_<CODE>.
14193 * expr.c (emit_push_insn): Likewise.
14194 (expand_expr_real_2): Likewise.
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 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14202 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14203 Alan Hayward <alan.hayward@arm.com>
14204 David Sherwood <david.sherwood@arm.com>
14206 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14207 calling tree_to_uhwi.
14209 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14210 Alan Hayward <alan.hayward@arm.com>
14211 David Sherwood <david.sherwood@arm.com>
14213 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14214 INTEGER_CST rather than a negative test for ADDR_EXPR.
14216 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14217 Alan Hayward <alan.hayward@arm.com>
14218 David Sherwood <david.sherwood@arm.com>
14220 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14221 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14223 2017-09-21 Richard Biener <rguenther@suse.de>
14225 PR tree-optimization/71351
14226 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14227 graphite_create_new_loop_guard): Remove, fold remaining parts
14229 (translate_isl_ast_node_for): ... here and simplify.
14231 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14234 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14235 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14236 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14237 alternative always use QI mode, for -Os imov (=R,R) alternative
14238 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14241 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14242 Jeff Law <law@redhat.com>
14244 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14245 (allocate_stack_space): New function, partially extracted from
14246 s390_emit_prologue.
14247 (s390_emit_prologue): Track offset to most recent stack probe.
14248 Code to allocate space moved into allocate_stack_space.
14249 Dump actions when no stack is allocated.
14250 (s390_prologue_plus_offset): New function.
14251 (s390_emit_stack_probe): Likewise.
14253 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14255 * common.opt (Wa, Wl, Wp, g, gz=): Add
14257 (gno-column-info): Remove.
14258 (gcolumn-info): Drop RejectNegative.
14259 (gno-): New prefix.
14260 (gno-record-gcc-switches): Remove.
14261 (grecord-gcc-switches): Drop RejectNegative.
14262 (gno-split-dwarf): Remove.
14263 (gsplit-dwarf): Drop RejectNegative.
14264 (gno-strict-dwarf): Remove.
14265 (gstrict-dwarf): Drop RejectNegative.
14266 * config/darwin.opt (gfull, gused): Add RejectNegative.
14267 * dwarf2out.c (gen_producer_string): Drop
14268 gno-record-gcc-switches handler.
14269 * optc-gen.awk: Add g to prefixes with negative forms.
14270 * opts-common.c (remapping_prefix_p): New.
14271 (find_opt): Check it.
14272 (generate_canonical_option): Test g prefix.
14273 (option_map): Add -gno- mapping.
14274 (add_misspelling_candidates): Check remapping_prefix_p.
14276 2017-09-20 Jeff Law <law@redhat.com>
14278 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14279 thinko in stack clash protection support.
14281 * explow.c (compute_stack_clash_protection_loop_data): Use
14282 CONST_INT_P instead of explicit test. Verify object is a
14283 CONST_INT_P before looking at INTVAL.
14284 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14285 instead of explicit test.
14287 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14290 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14291 address instead of to r1 and r11.
14293 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14295 * config.gcc: Support "knm".
14296 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14297 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14299 * config/i386/i386.c (m_KNM): Define.
14300 (processor_target_table): Add "knm".
14302 (ix86_option_override_internal): Add "knm".
14303 (ix86_issue_rate): Add PROCESSOR_KNM.
14304 (ix86_adjust_cost): Ditto.
14305 (ia32_multipass_dfa_lookahead): Ditto.
14306 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14307 (fold_builtin_cpu): Add M_INTEL_KNM.
14308 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14309 (processor_type): Add PROCESSOR_KNM.
14310 * config/i386/x86-tune.def: Add m_KNM.
14311 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14313 2017-09-20 Richard Biener <rguenther@suse.de>
14315 PR tree-optimization/80213
14316 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14317 are allowed in empty BBs as well.
14318 (canonicalize_loop_closed_ssa): Also look for other complex
14320 (scop_detection::get_sese): Include the loop-closed PHI block
14322 (scop_detection::merge_sese): Remove code adding extra blocks.
14323 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14324 (build_scops): Assert the final returned scop is invalid.
14326 2017-09-20 Richard Biener <rguenther@suse.de>
14328 PR tree-optimization/82264
14329 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14331 (vn_phi_lookup): Likewise.
14332 (vn_phi_insert): Likewise.
14334 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14336 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14337 that fits into uhwi or shwi, add DW_AT_const_value regardless
14338 of early_dwarf without going through RTL, using add_AT_unsigned
14341 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14342 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14343 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14344 (DEBUG_MACRO_SECTION): Likewise.
14345 (DEBUG_LTO_MACRO_SECTION): Likewise.
14346 (DEBUG_STR_DWO_SECTION): Likewise.
14347 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14348 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14349 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14350 (DEBUG_LTO_LINE_STR_SECTION): Define.
14351 (init_sections_and_labels): Initialize debug_line_str_section
14352 variable. Initialize debug_loc_section for -gdwarf-5 to
14353 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14355 2017-09-20 Richard Biener <rguenther@suse.de>
14357 * graphite-sese-to-poly.c (extract_affine): Properly handle
14358 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14360 2017-09-20 Richard Biener <rguenther@suse.de>
14362 PR tree-optimization/81373
14363 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14364 Force SESE live-out defs to be handled even if they are
14367 2017-09-19 Jeff Law <law@redhat.com>
14369 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14370 nothing for stack adjustments with REG_STACK_CHECK.
14371 * sched-deps.c (parse_add_or_inc): Reject insns with
14372 REG_STACK_CHECK from dependency breaking.
14373 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14374 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14375 * reg-notes.def (STACK_CHECK): New note.
14377 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14378 (ix86_expand_prologue): Dump stack clash info as needed.
14379 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14381 * function.c (dump_stack_clash_frame_info): New function.
14382 * function.h (dump_stack_clash_frame_info): Prototype.
14383 (enum stack_clash_probes): New enum.
14385 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14386 flag_stack_clash_protection.
14387 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14388 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14389 (arm_frame_pointer_required): Likewise.
14390 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14391 (ia64_expand_prologue): Likewise.
14392 * config/mips/mips.c (mips_expand_prologue): Likewise.
14393 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14394 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14395 (sparc_flat_expand_prologue): Likewise.
14396 * config/spu/spu.c (spu_expand_prologue): Likewise.
14398 * explow.c: Include "params.h".
14399 (anti_adjust_stack_and_probe_stack_clash): New function.
14400 (get_stack_check_protect): Likewise.
14401 (compute_stack_clash_protection_loop_data): Likewise.
14402 (emit_stack_clash_protection_loop_start): Likewise.
14403 (emit_stack_clash_protection_loop_end): Likewise.
14404 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14405 Use anti_adjust_stack_and_probe_stack_clash.
14406 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14407 (emit_stack_clash_protection_loop_start): Likewise.
14408 (emit_stack_clash_protection_loop_end): Likewise.
14409 * rtl.h (get_stack_check_protect): Prototype.
14410 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14411 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14412 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14414 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14416 * doc/tm.texi: Rebuilt.
14417 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14418 get_stack_check_protect.
14419 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14420 * config/arm/arm.c (arm_expand_prologue): Likewise.
14421 (arm_frame_pointer_required): Likewise.
14422 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14423 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14424 * config/mips/mips.c (mips_expand_prologue): Likewise.
14425 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14426 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14427 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14428 (sparc_flat_expand_prologue): Likewise.
14430 * common.opt (-fstack-clash-protection): New option.
14431 * flag-types.h (enum stack_check_type): Note difference between
14432 -fstack-check= and -fstack-clash-protection.
14433 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14434 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14435 * toplev.c (process_options): Issue warnings/errors for cases
14436 not handled with -fstack-clash-protection.
14437 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14438 (-fstack-check): Note additional problem with -fstack-check=generic.
14439 Note that -fstack-check is primarily for Ada and refer users
14440 to -fstack-clash-protection for stack-clash-protection.
14441 Document new params for stack clash protection.
14443 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14445 * config/i386/i386.c (ix86_split_long_move): Do not handle
14446 address used for LEA in a special way.
14448 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14450 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14452 2017-09-19 Martin Sebor <msebor@redhat.com>
14455 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14456 of incompatible types.
14458 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14460 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14461 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14462 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14463 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14465 2017-09-19 Richard Biener <rguenther@suse.de>
14467 PR tree-optimization/82244
14468 * tree-vrp.c (remove_range_assertions): Do not propagate
14469 a constant to abnormals but replace the assert with a copy.
14471 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14473 PR rtl-optimization/57878
14474 PR rtl-optimization/68988
14475 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14476 avoidance test involving non_reload_pseudos. Move frequency test
14477 below the general fragmentation avoidance test.
14479 2017-09-19 Richard Biener <rguenther@suse.de>
14481 PR tree-optimization/69728
14482 * graphite-sese-to-poly.c (schedule_error): New global.
14483 (add_loop_schedule): Handle empty domain by failing the
14485 (build_original_schedule): Handle schedule_error.
14487 2017-09-19 Richard Biener <rguenther@suse.de>
14489 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14490 Do not iterate to sibling loops but only to siblings of inner
14493 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14496 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14497 that the registers are different.
14499 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14501 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14502 to processor_model and "amdfam17h" to arch_names_table.
14503 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14505 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14508 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14510 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14511 Alan Hayward <alan.hayward@arm.com>
14512 David Sherwood <david.sherwood@arm.com>
14514 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14516 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14518 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14519 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14520 vect_schedule_slp_instance.
14521 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14522 Update call to vect_slp_analyze_node_operations. Simplify return
14524 (vect_slp_analyze_bb_1): Update call accordingly.
14525 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14526 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14527 (vect_schedule_slp): Update call accordingly.
14529 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14530 Alan Hayward <alan.hayward@arm.com>
14531 David Sherwood <david.sherwood@arm.com>
14533 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14534 with types that aren't in fact scalar.
14536 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14538 * tree-vect-slp.c (vect_record_max_nunits): New function,
14540 (vect_build_slp_tree_1): ...here.
14541 (vect_build_slp_tree_2): Call it for phis too.
14543 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14545 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14546 to vect_get_vec_def_for_operand when getting the mask operand.
14548 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14549 Alan Hayward <alan.hayward@arm.com>
14550 David Sherwood <david.sherwood@arm.com>
14552 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14555 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14556 Alan Hayward <alan.hayward@arm.com>
14557 David Sherwood <david.sherwood@arm.com>
14559 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14560 calculation for vector booleans.
14562 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14563 Alan Hayward <alan.hayward@arm.com>
14564 David Sherwood <david.sherwood@arm.com>
14566 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14568 (vect_transform_stmt): ...here.
14569 (vect_analyze_stmt): Use it instead of calling
14570 vectorizable_live_operation directly.
14572 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14574 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14575 non-SIMT targets in acc vector loops.
14577 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14579 * configure.ac: Add arc and check if assembler supports gdwarf2.
14580 * configure: Regenerate.
14582 2017-09-18 Richard Biener <rguenther@suse.de>
14584 PR tree-optimization/82220
14585 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14586 epilogue niters from the min_profitable_iters compute.
14588 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14591 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14592 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14593 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14595 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14598 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14599 switching to a new text section.
14601 2017-09-18 Richard Biener <rguenther@suse.de>
14603 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14605 (build_alias_set): Reject aliases with no access function.
14607 2017-09-18 Richard Biener <rguenther@suse.de>
14609 PR tree-optimization/79622
14610 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14612 (build_cross_bb_scalars_use): Likewise.
14614 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14616 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14618 2017-09-18 Alan Modra <amodra@gmail.com>
14621 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14622 stack_pointer_rtx for count 0. Update comments. Break up
14623 large rtl expression.
14625 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14627 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14628 Increase to 20 bytes.
14629 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14630 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14631 or avx version of the stub.
14633 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14636 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14637 compute the minimum stack alignment. Also update preferred stack
14638 boundary for leaf functions.
14640 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14642 PR tree-optimization/82228
14643 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14646 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14648 * common/config/nds32/nds32-common.c
14649 (nds32_option_optimization_table): Refine formatting.
14650 (nds32_option_optimization_table): Use -fsched-pressure and
14651 -fomit-frame-pointer for specific optimization level.
14653 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14655 * config/nds32/nds32.c: Refine formatting and comments.
14656 * config/nds32/nds32.h: Likewise.
14657 * config/nds32/nds32.md: Likewise.
14658 * config/nds32/nds32-cost.c: Likewise.
14659 * config/nds32/nds32-isr.c: Likewise.
14660 * config/nds32/nds32-md-auxiliary.c: Likewise.
14661 * config/nds32/nds32-multiple.md: Likewise.
14662 * config/nds32/nds32-predicates.c: Likewise.
14664 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14665 Jakub Jelinek <jakub@redhat.com>
14667 Add support for -std=c++2a.
14668 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14670 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14672 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14675 * doc/extend.texi (Common Function Attributes): Add
14676 references to ARM, AArch64, and S/390 specific attributes.
14677 (Function Specific Option Pragmas): Add AArch64 and S/390
14678 to list of back ends that support the target pragma.
14680 2017-09-15 Nathan Sidwell <nathan@acm.org>
14682 * doc/standards.texi: Fix C++17 description. Update URLs for
14685 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14687 * common.opt (Wcast-align=strict): New warning option.
14688 * doc/invoke.texi: Document -Wcast-align=strict.
14690 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14692 * cgraph.h (cgraph_thunk_info): Add comments.
14693 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14694 assert for VIRTUAL_* arguments stricter.
14696 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14698 PR tree-optimization/71026
14699 * match.pd: Move RDIV patterns from fold-const.c
14700 * fold-const.c (distribute_real_division): Removed.
14701 (fold_binary_loc): Remove calls to distribute_real_divison.
14703 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14705 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14706 c++1z and gnu++1z as deprecated. Change other references to
14707 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14708 Change -Wc++1z-compat to -Wc++17-compat.
14709 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14710 * dwarf2out.c (highest_c_language): Handle C++17.
14711 (gen_compile_unit_die): Likewise.
14713 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14715 PR rtl-optimization/82192
14716 * combine.c (make_extraction): Don't look through non-paradoxical
14717 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14720 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14721 Alan Hayward <alan.hayward@arm.com>
14722 David Sherwood <david.sherwood@arm.com>
14724 * target.def (function_arg_offset): New hook.
14725 * targhooks.h (default_function_arg_offset): Declare.
14726 * targhooks.c (default_function_arg_offset): New function.
14727 * function.c (locate_and_pad_parm): Use
14728 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14729 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14730 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14731 * doc/tm.texi: Regenerate.
14732 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14733 * config/spu/spu.c (spu_function_arg_offset): New function.
14734 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14735 * system.h (FUNCTION_ARG_OFFSET): Poison.
14737 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14738 Alan Hayard <alan.hayward@arm.com>
14739 David Sherwood <david.sherwood@arm.com>
14741 * target.def (truly_noop_truncation): New hook.
14742 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14743 than TRULY_NOOP_TRUNCATION.
14744 * hooks.h (hook_bool_uint_uint_true): Declare.
14745 * hooks.c (hook_bool_uint_uint_true): New function.
14746 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14747 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14748 * doc/tm.texi: Regenerate.
14749 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14750 rather than TRULY_NOOP_TRUNCATION in comments.
14751 (simplify_comparison): Likewise.
14752 (record_truncated_value): Likewise.
14753 * expmed.c (extract_bit_field_1): Likewise.
14754 (extract_split_bit_field): Likewise.
14755 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14756 instead of TRULY_NOOP_TRUNCATION.
14757 * function.c (assign_parm_setup_block): Likewise.
14758 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14759 * rtlhooks.c: Include target.h.
14760 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14761 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14762 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14763 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14764 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14765 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14766 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14767 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14768 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14769 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14770 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14771 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14772 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14773 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14774 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14775 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14776 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14777 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14778 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14779 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14780 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14781 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14782 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14783 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14784 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14785 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14786 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14787 rather than TRULY_NOOP_TRUNCATION in comments.
14788 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14789 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14790 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14791 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14792 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14793 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14794 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14795 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14796 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14797 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14798 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14799 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14800 rather than TRULY_NOOP_TRUNCATION in comments.
14801 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14802 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14803 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14804 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14805 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14806 TRULY_NOOP_TRUNCATION condition.
14807 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14808 (TRULY_NOOP_TRUNCATION): Delete.
14809 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14810 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14811 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14812 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14813 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14814 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14815 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14816 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14817 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14818 rather than TRULY_NOOP_TRUNCATION in comments.
14819 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14820 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14821 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14822 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14823 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14824 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14826 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14829 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14830 (*cmp_ior): Likewise.
14831 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14832 (*ior_scc_scc_cmp): Likewise.
14833 (*and_scc_scc): Likewise.
14834 (*and_scc_scc_cmp): Likewise.
14836 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14837 Alan Hayard <alan.hayward@arm.com>
14838 David Sherwood <david.sherwood@arm.com>
14840 * target.def (can_change_mode_class): New hook.
14841 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14842 (hard_regno_nregs): Likewise.
14843 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14844 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14845 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14846 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14847 (LOAD_EXTEND_OP): Update accordingly.
14848 * doc/tm.texi: Regenerate.
14849 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14850 CANNOT_CHANGE_MODE_CLASS.
14851 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14852 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14853 * combine.c (simplify_set): Update accordingly.
14854 * emit-rtl.c (validate_subreg): Likewise.
14855 * recog.c (general_operand): Likewise.
14856 * regcprop.c (mode_change_ok): Likewise.
14857 * reload1.c (choose_reload_regs): Likewise.
14858 (inherit_piecemeal_p): Likewise.
14859 * rtlanal.c (simplify_subreg_regno): Likewise.
14860 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14861 instead of CANNOT_CHANGE_MODE_CLASS.
14862 (reload_cse_simplify_operands): Likewise.
14863 * reload.c (push_reload): Use targetm.can_change_mode_class
14864 instead of CANNOT_CHANGE_MODE_CLASS.
14865 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14866 REG_CANNOT_CHANGE_MODE_P.
14867 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14868 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14869 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14870 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14871 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14872 (arm_can_change_mode_class): New function.
14873 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14874 than CANNOT_CHANGE_MODE_CLASS in comments.
14875 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14876 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14877 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14878 (ix86_can_change_mode_class): ...this new function, inverting the
14879 sense of the return value.
14880 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14881 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14882 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14883 (ia64_can_change_mode_class): New function.
14884 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14885 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14886 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14887 (m32c_can_change_mode_class): ...this new function, inverting the
14888 sense of the return value.
14889 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14890 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14891 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14892 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14893 (mips_can_change_mode_class): ...this new function, inverting the
14894 sense of the return value.
14895 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14896 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14897 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14898 (msp430_can_change_mode_class): New function.
14899 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14900 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14901 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14902 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14903 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14904 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14905 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14906 (pa_cannot_change_mode_class): Replace with...
14907 (pa_can_change_mode_class): ...this new function, inverting the
14908 sense of the return value.
14909 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14910 than CANNOT_CHANGE_MODE_CLASS in comments.
14911 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14912 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14913 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14914 (pdp11_cannot_change_mode_class): Replace with...
14915 (pdp11_can_change_mode_class): ...this new function, inverting the
14916 sense of the return value.
14917 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14918 * config/powerpcspe/powerpcspe-protos.h
14919 (rs6000_cannot_change_mode_class_ptr): Delete.
14920 * config/powerpcspe/powerpcspe.c
14921 (rs6000_cannot_change_mode_class_ptr): Delete.
14922 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14923 (rs6000_option_override_internal): Assign to
14924 targetm.can_change_mode_class instead of
14925 rs6000_cannot_change_mode_class_ptr.
14926 (rs6000_cannot_change_mode_class): Replace with...
14927 (rs6000_can_change_mode_class): ...this new function, inverting the
14928 sense of the return value.
14929 (rs6000_debug_cannot_change_mode_class): Replace with...
14930 (rs6000_debug_can_change_mode_class): ...this new function.
14931 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14932 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14933 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14934 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14935 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14937 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14938 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14939 (rs6000_option_override_internal): Assign to
14940 targetm.can_change_mode_class instead of
14941 rs6000_cannot_change_mode_class_ptr.
14942 (rs6000_cannot_change_mode_class): Replace with...
14943 (rs6000_can_change_mode_class): ...this new function, inverting the
14944 sense of the return value.
14945 (rs6000_debug_cannot_change_mode_class): Replace with...
14946 (rs6000_debug_can_change_mode_class): ...this new function.
14947 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14948 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14949 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14950 (s390_can_change_mode_class): ...this new function, inverting the
14951 sense of the return value.
14952 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14953 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14954 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14955 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14956 (sh_cannot_change_mode_class): Replace with...
14957 (sh_can_change_mode_class): ...this new function, inverting the
14958 sense of the return value.
14959 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14960 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14961 (sparc_can_change_mode_class): New function.
14962 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14963 * config/spu/spu.c (spu_can_change_mode_class): New function.
14964 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14965 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14966 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14967 (visium_can_change_mode_class): New function.
14968 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14970 2017-09-15 Richard Biener <rguenther@suse.de>
14972 PR tree-optimization/82217
14973 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14974 but not undefined case.
14976 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14979 * postreload.c (reload_cse_simplify_operands): Skip
14980 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14982 2017-09-15 Richard Biener <rguenther@suse.de>
14984 PR tree-optimization/68823
14985 * graphite-scop-detection.c (build_alias_set): If we have a
14986 possible dependence check whether we can handle them by just
14987 looking at the DRs DR_ACCESS_FNs.
14988 (build_scops): If build_alias_set fails, fail the SCOP.
14990 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14992 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14993 to support float128 built-in functions that require the ISA 3.0
14995 (BU_FLOAT128_3_HW): Likewise.
14996 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14997 built-in functions.
14998 (FMAF128): Likewise.
15000 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15001 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15002 floating point instructions.
15003 (rs6000_invalid_builtin): Likewise.
15004 (rs6000_builtin_mask_names): Likewise.
15005 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15006 (RS6000_BTM_FLOAT128_HW): Likewise.
15007 (RS6000_BTM_COMMON): Likewise.
15008 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15010 * doc/extend.texi (RS/6000 built-in functions): Document the
15011 IEEE 128-bit floating point square root and fused multiply-add
15012 built-in functions.
15014 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15016 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15017 reg (r2) isn't in the set of registers defined in the prologue.
15019 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15020 Alan Hayward <alan.hayward@arm.com>
15021 David Sherwood <david.sherwood@arm.com>
15023 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15024 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15025 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15026 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15027 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15029 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15030 max_vectorization_factor.
15031 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15033 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15034 Alan Hayward <alan.hayward@arm.com>
15035 David Sherwood <david.sherwood@arm.com>
15037 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15038 (vect_worthwhile_without_simd_p): Declare.
15039 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15040 (vectorizable_reduction): Use it.
15041 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15042 (vectorizable_operation): Likewise.
15044 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15045 Alan Hayward <alan.hayward@arm.com>
15046 David Sherwood <david.sherwood@arm.com>
15048 * tree-vectorizer.h (vect_get_num_copies): New function.
15049 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15050 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15051 (vectorizable_induction): Likewise.
15052 (vectorizable_live_operation): Likewise.
15053 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15054 (vectorizable_bswap): Likewise.
15055 (vectorizable_call): Likewise.
15056 (vectorizable_conversion): Likewise.
15057 (vectorizable_assignment): Likewise.
15058 (vectorizable_shift): Likewise.
15059 (vectorizable_operation): Likewise.
15060 (vectorizable_store): Likewise.
15061 (vectorizable_load): Likewise.
15062 (vectorizable_condition): Likewise.
15063 (vectorizable_comparison): Likewise.
15064 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15066 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15067 Alan Hayward <alan.hayward@arm.com>
15068 David Sherwood <david.sherwood@arm.com>
15070 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15071 of vect_init_vector.
15073 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15074 Alan Hayward <alan.hayward@arm.com>
15075 David Sherwood <david.sherwood@arm.com>
15077 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15078 an inline wrapper that provides a location.
15079 (gimple_build_vector): Likewise.
15080 * gimple-fold.c (gimple_build_vector_from_val): New function.
15081 (gimple_build_vector): Likewise.
15082 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15083 functions to build the initial value. Always return a gimple value.
15084 (get_initial_defs_for_reduction): Likewise. Only compute
15086 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15087 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15088 (vectorizable_induction): Use gimple_build_vector rather than
15091 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15092 Alan Hayward <alan.hayward@arm.com>
15093 David Sherwood <david.sherwood@arm.com>
15095 * target.h (vec_perm_indices): New typedef.
15096 (auto_vec_perm_indices): Likewise.
15097 * optabs-query.h: Include target.h
15098 (can_vec_perm_p): Take a vec_perm_indices *.
15099 * optabs-query.c (can_vec_perm_p): Likewise.
15100 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15101 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15102 * tree-vect-generic.c (lower_vec_perm): Likewise.
15103 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15104 (vect_grouped_load_supported): Likewise.
15105 (vect_shift_permute_load_chain): Likewise.
15106 (vect_permute_store_chain): Use auto_vec_perm_indices.
15107 (vect_permute_load_chain): Likewise.
15108 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15109 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15110 Update uses of can_vec_perm_p.
15111 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15112 mode with a number of elements. Take a vec_perm_indices *.
15113 (vect_create_epilog_for_reduction): Update accordingly.
15114 Use auto_vec_perm_indices.
15115 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15116 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15117 (vect_transform_slp_perm_load): Likewise.
15118 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15119 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15120 (vect_gen_perm_mask_checked): Likewise.
15121 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15122 (vect_gen_perm_mask_checked): Likewise.
15123 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15124 (vectorizable_store): Likewise.
15125 (vectorizable_load): Likewise.
15126 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15127 (vectorizable_bswap): Likewise.
15129 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15130 Alan Hayward <alan.hayward@arm.com>
15131 David Sherwood <david.sherwood@arm.com>
15133 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15134 * tree.c (build_vector): Likewise.
15135 (build_vector_from_ctor): Update accordingly.
15136 (build_vector_from_val): Likewise.
15137 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15138 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15139 * tree-vect-generic.c (add_rshift): Likewise.
15140 (expand_vector_divmod): Likewise.
15141 (optimize_vector_constructor): Likewise.
15142 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15143 (vect_transform_slp_perm_load): Likewise.
15144 (vect_schedule_slp_instance): Likewise.
15145 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15146 (vectorizable_call): Likewise.
15147 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15148 * expmed.c (make_tree): Likewise.
15149 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15150 a vector passed to build_vector.
15151 (fold_convert_const): Likewise.
15152 (exact_inverse): Likewise.
15153 (fold_ternary_loc): Likewise.
15154 (fold_relational_const): Likewise.
15155 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15156 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15157 (const_unop): Likewise. Store the reduction accumulator in a
15158 variable rather than an array.
15159 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15160 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15161 the new vector, rather than constructing it after the input arrays.
15162 (native_interpret_vector): Use auto_vec<tree> when building
15163 a vector passed to build_vector. Add elements in order.
15164 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15165 auto_vec<tree> when building a vector passed to build_vector.
15166 (vect_create_epilog_for_reduction): Likewise.
15167 (vectorizable_induction): Likewise.
15168 (get_initial_def_for_reduction): Likewise. Fix indentation of
15170 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15172 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15173 passed to build_vector.
15175 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15176 Alan Hayward <alan.hayward@arm.com>
15177 David Sherwood <david.sherwood@arm.com>
15179 * tree-core.h (tree_base::u): Add an "nelts" field.
15180 (tree_vector): Use VECTOR_CST_NELTS as the length.
15181 * tree.c (tree_size): Likewise.
15182 (make_vector): Initialize VECTOR_CST_NELTS.
15183 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15184 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15185 TYPE_VECTOR_SUBPARTS.
15186 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15187 as the number of units, setting it from VECTOR_CST_NELTS.
15188 (const_vector_from_tree): Likewise.
15189 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15190 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15191 (fold_negate_expr_1): Likewise.
15192 (fold_convert_const): Likewise.
15193 (const_binop): Likewise. Differentiate the number of output and
15195 (const_unop): Likewise.
15196 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15197 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15198 in cases that did the opposite.
15200 2017-09-14 Richard Biener <rguenther@suse.de>
15202 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15205 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15207 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15209 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15212 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15213 if and where to split a bb, except for splitting before debug insn
15214 sequences followed by non-label real insn. Delete debug insns
15215 in between basic blocks.
15217 * combine.c (make_compound_operation_int): Formatting fixes.
15219 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15220 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15221 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15222 * config/sol2.h (LINK_EH_SPEC): Likewise.
15223 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15224 * config/s390/linux.h (LINK_SPEC): Likewise.
15225 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15226 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15227 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15228 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15229 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15230 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15231 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15232 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15234 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15236 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15238 (ENDFILE_LINUX_SPEC): Likewise.
15239 (LINK_EH_SPEC): Likewise.
15240 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15241 (LINK_OS_LINUX_SPEC32): Likewise.
15242 (LINK_OS_LINUX_SPEC64): Likewise.
15243 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15244 (LINK_OS_LINUX_SPEC): Likewise.
15246 2017-09-13 Martin Liska <mliska@suse.cz>
15248 PR middle-end/82154
15249 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15250 CASE_HIGH is NULL_TREE.
15252 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15253 Alan Hayward <alan.hayward@arm.com>
15254 David Sherwood <david.sherwood@arm.com>
15256 * target.def (secondary_memory_needed): New hook.
15257 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15258 instead of SECONDARY_MEMORY_NEEDED.
15259 (secondary_memory_needed_mode): Likewise.
15260 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15261 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15262 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15263 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15264 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15265 * doc/tm.texi: Regenerate.
15266 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15267 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15268 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15269 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15270 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15271 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15272 mode argument first and change the reg_class arguments to reg_class_t.
15273 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15274 Make static. Update the call to inline_secondary_memory_needed.
15275 (ix86_register_move_cost): Update the call to
15276 inline_secondary_memory_needed.
15277 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15278 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15280 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15281 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15283 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15284 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15285 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15286 and match hook interface. Add comment from mips.h.
15287 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15288 * config/mmix/mmix.md (truncdfsf2): Refer to
15289 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15291 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15292 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15293 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15294 (pa_secondary_memory_needed): New function.
15295 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15296 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15297 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15298 (pdp11_secondary_memory_needed): Make static and match hook interface.
15299 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15300 * config/powerpcspe/powerpcspe-protos.h
15301 (rs6000_secondary_memory_needed_ptr): Delete.
15302 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15304 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15305 (rs6000_option_override_internal): Assign to
15306 targetm.secondary_memory_needed rather than
15307 rs6000_secondary_memory_needed_ptr.
15308 (rs6000_secondary_memory_needed): Match hook interface.
15309 (rs6000_debug_secondary_memory_needed): Likewise.
15310 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15311 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15312 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15313 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15314 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15315 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15317 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15318 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15319 (rs6000_option_override_internal): Assign to
15320 targetm.secondary_memory_needed rather than
15321 rs6000_secondary_memory_needed_ptr.
15322 (rs6000_secondary_memory_needed): Match hook interface.
15323 (rs6000_debug_secondary_memory_needed): Likewise.
15324 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15325 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15326 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15327 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15328 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15329 (sparc_secondary_memory_needed): New function.
15330 * lra-constraints.c (check_and_process_move): Refer to
15331 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15333 (curr_insn_transform): Likewise.
15334 (process_alt_operands): Use targetm.secondary_memory_needed
15335 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15336 (check_secondary_memory_needed_p): Likewise.
15337 (choose_split_class): Likewise.
15338 * reload.c: Unconditionally include code that was previously
15339 conditional on SECONDARY_MEMORY_NEEDED.
15340 (push_secondary_reload): Use targetm.secondary_memory_needed
15341 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15342 (push_reload): Likewise.
15343 * reload1.c: Unconditionally include code that was previously
15344 conditional on SECONDARY_MEMORY_NEEDED.
15345 (choose_reload_regs): Use targetm.secondary_memory_needed
15346 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15347 (gen_reload): Likewise.
15348 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15350 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15351 Alan Hayward <alan.hayward@arm.com>
15352 David Sherwood <david.sherwood@arm.com>
15354 * target.def (secondary_memory_needed_mode): New hook:
15355 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15356 * targhooks.h (default_secondary_memory_needed_mode): New function.
15357 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15358 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15359 * doc/tm.texi: Regenerate.
15360 * lra-constraints.c (check_and_process_move): Use
15361 targetm.secondary_memory_needed_mode instead of
15362 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15363 (curr_insn_transform): Likewise.
15364 * reload.c (get_secondary_mem): Likewise.
15365 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15366 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15368 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15369 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15370 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15371 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15372 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15374 * config/powerpcspe/powerpcspe-protos.h
15375 (rs6000_secondary_memory_needed_mode): Delete.
15376 * config/powerpcspe/powerpcspe.c
15377 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15378 (rs6000_secondary_memory_needed_mode): Make static.
15379 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15380 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15382 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15384 (rs6000_secondary_memory_needed_mode): Make static.
15385 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15386 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15387 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15388 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15389 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15391 (sparc_secondary_memory_needed_mode): New function.
15392 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15394 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15396 * config/aarch64/constraints.md (Umq): New constraint.
15397 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15399 (mov<mode>): Update condition.
15401 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15403 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15404 when bitposition is the same.
15406 2017-09-13 Richard Biener <rguenther@suse.de>
15408 * dwarf2out.c (output_die_symbol): Remove.
15409 (output_die): Do not output a DIEs symbol.
15411 2017-09-13 Richard Biener <rguenther@suse.de>
15413 PR middle-end/82128
15414 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15415 default-def to avoid breaking iterator update with the weird
15416 interaction with cgraph_update_edges_for_call_stmt_node.
15418 2017-09-13 Richard Biener <rguenther@suse.de>
15420 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15421 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15422 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15423 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15425 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15427 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15428 Disable pc relative literal load irrespective of
15429 TARGET_FIX_ERR_A53_84341 for default.
15431 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15433 * config/sparc/sparc.c (output_return): Output the source location of
15434 the insn in the delay slot, if any.
15435 (output_sibcall): Likewise.
15437 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15440 * common.opt (-static-pie): New alias.
15441 (shared): Negate static-pie.
15442 (-no-pie): Update help text.
15444 (static-pie): New option.
15445 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15446 -static-pie support.
15447 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15448 (LINK_EH_SPEC): Likewise.
15449 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15450 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15451 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15452 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15453 (init_gcc_specs): Likewise.
15454 (init_spec): Likewise.
15455 (display_help): Update help message for -pie.
15456 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15459 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15461 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15462 (movdi_aarch64): Likewise.
15463 (movti_aarch64): Likewise.
15465 2017-09-12 Simon Wright <simon@pushface.org>
15468 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15469 calculation of the minor version, always output as 0.
15471 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15474 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15475 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15476 on it early, rather than manual conversion late. For
15477 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15478 instead of performing manual conversion.
15480 2017-09-12 Carl Love <cel@us.ibm.com>
15482 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15483 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15485 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15486 VMULOSW): Add definitions.
15487 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15488 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15489 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15490 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15491 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15493 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15495 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15497 (movti_aarch64): Likewise.
15498 (movdf_aarch64): Likewise.
15499 (movtf_aarch64): Likewise.
15500 (load_pairdi): Likewise.
15501 (store_pairdi): Likewise.
15502 (load_pairdf): Likewise.
15503 (store_pairdf): Likewise.
15504 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15505 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15506 (ldr_got_small_<mode>): Likewise.
15507 (ldr_got_small_28k_<mode>): Likewise.
15508 (ldr_got_tiny): Likewise.
15509 * config/aarch64/iterators.md (ldst_sz): New.
15510 (ldpstp_sz): Likewise.
15511 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15513 (thunderx_load): Split load_8 to load_16.
15514 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15516 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15517 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15518 (xgene1_store_pair): Split store_8 to store_16.
15519 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15520 (falkor_st_0_st_sd): Split store_8 to store_16.
15522 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15524 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15525 and store1/2/3/4 to store_4/8/12/16.
15526 * config/aarch64/aarch64.md: Update for rename.
15527 * config/arm/arm.md: Likewise.: Likewise.
15528 * config/arm/arm.c: Likewise.
15529 * config/arm/thumb1.md: Likewise.
15530 * config/arm/thumb2.md: Likewise.
15531 * config/arm/vfp.md: Likewise.
15532 * config/arm/arm-generic.md: Likewise.
15533 * config/arm/arm1020e.md: Likewise.
15534 * config/arm/arm1026ejs.md: Likewise.
15535 * config/arm/arm1136jfs.md: Likewise.
15536 * config/arm/arm926ejs.md: Likewise.
15537 * config/arm/cortex-a15.md: Likewise.
15538 * config/arm/cortex-a17.md: Likewise.
15539 * config/arm/cortex-a5.md: Likewise.
15540 * config/arm/cortex-a53.md: Likewise.
15541 * config/arm/cortex-a57.md: Likewise.
15542 * config/arm/cortex-a7.md: Likewise.
15543 * config/arm/cortex-a8.md: Likewise.
15544 * config/arm/cortex-a9.md: Likewise.
15545 * config/arm/cortex-m4.md: Likewise.
15546 * config/arm/cortex-m7.md: Likewise.
15547 * config/arm/cortex-r4.md: Likewise.
15548 * config/arm/exynos-m1.md: Likewise.
15549 * config/arm/fa526.md: Likewise.
15550 * config/arm/fa606te.md: Likewise.
15551 * config/arm/fa626te.md: Likewise.
15552 * config/arm/fa726te.md: Likewise.
15553 * config/arm/fmp626.md: Likewise.
15554 * config/arm/iwmmxt.md: Likewise.
15555 * config/arm/ldmstm.md: Likewise.
15556 * config/arm/marvell-pj4.md: Likewise.
15557 * config/arm/xgene1.md: Likewise.
15558 * config/aarch64/thunderx.md: Likewise.
15559 * config/aarch64/thunderx2t99.md: Likewise.
15560 * config/aarch64/falkor.md: Likewise.
15562 2017-09-12 Martin Liska <mliska@suse.cz>
15564 * attribs.c (private_lookup_attribute): New function.
15565 * attribs.h (private_lookup_attribute): Declared here.
15566 (lookup_attribute): Called from this place.
15568 2017-09-12 Richard Biener <rguenther@suse.de>
15570 PR tree-optimization/82157
15571 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15572 stmts with side-effects.
15574 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15575 Alan Hayward <alan.hayward@arm.com>
15576 David Sherwood <david.sherwood@arm.com>
15578 * target.def (hard_regno_nregs): New hook.
15579 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15580 * targhooks.h (default_hard_regno_nregs): Declare.
15581 * targhooks.c (default_hard_regno_nregs): New function.
15582 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15583 (TARGET_HARD_REGNO_NREGS): ...this hook.
15584 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15585 (CLASS_MAX_NREGS): Likewise.
15586 * doc/tm.texi: Regenerate.
15587 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15588 instead of HARD_REGNO_NREGS.
15589 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15590 HARD_REGNO_NREGS in the comment.
15591 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15592 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15593 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15594 Return an unsigned int.
15595 (TARGET_HARD_REGNO_NREGS): Redefine.
15596 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15597 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15598 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15599 (arc_hard_regno_nregs): New function.
15600 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15601 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15602 (arm_hard_regno_nregs): New function.
15603 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15604 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15605 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15606 (TARGET_HARD_REGNO_NREGS): Redefine.
15607 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15608 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15609 (HARD_REGNO_NREGS): Delete.
15610 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15611 (cr16_hard_regno_nregs): New function.
15612 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15613 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15614 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15615 (cris_hard_regno_nregs): New function.
15616 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15617 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15618 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15619 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15620 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15621 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15622 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15623 (frv_hard_regno_nregs): Make static. Take and return an
15625 (frv_class_max_nregs): Remove outdated copy of documentation.
15626 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15627 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15628 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15629 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15630 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15631 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15632 (TARGET_HARD_REGNO_NREGS): Redefine.
15633 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15634 (CLASS_MAX_NREGS): Update comment.
15635 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15636 (ia64_hard_regno_nregs): New function.
15637 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15638 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15639 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15640 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15641 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15643 (m32c_hard_regno_nregs): Likewise. Make static.
15644 (TARGET_HARD_REGNO_NREGS): Redefine.
15645 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15646 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15647 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15648 (m68k_hard_regno_nregs): New function.
15649 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15650 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15651 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15652 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15653 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15654 Take and return an unsigned int.
15655 (TARGET_HARD_REGNO_NREGS): Redefine.
15656 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15657 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15658 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15659 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15660 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15661 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15662 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15663 (msp430_hard_regno_nregs): Make static. Take and return an
15665 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15666 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15667 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15668 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15669 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15670 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15671 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15672 (TARGET_HARD_REGNO_NREGS): Redefine.
15673 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15674 (PA_HARD_REGNO_NREGS): ...this.
15675 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15676 (PA_HARD_REGNO_NREGS): ...this.
15677 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15678 (pa_hard_regno_nregs): New function.
15679 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15680 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15681 (pdp11_hard_regno_nregs): New function.
15682 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15683 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15684 (rs6000_hard_regno_nregs_hook): New function.
15685 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15686 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15687 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15688 Take and return an unsigned int. Move earlier in file.
15689 (TARGET_HARD_REGNO_NREGS): Redefine.
15690 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15691 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15692 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15693 (rl78_hard_regno_nregs): Make static. Take and return an
15695 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15696 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15697 (rs6000_hard_regno_nregs_hook): New function.
15698 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15699 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15700 (TARGET_HARD_REGNO_NREGS): Redefine.
15701 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15702 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15703 instead of HARD_REGNO_NREGS.
15704 (s390_hard_regno_nregs): New function.
15705 (s390_hard_regno_mode_ok): Add comment from s390.h.
15706 (TARGET_HARD_REGNO_NREGS): Redefine.
15707 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15708 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15709 (sh_hard_regno_nregs): New function.
15710 (sh_pass_in_reg_p): Use it.
15711 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15712 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15713 (sparc_hard_regno_nregs): New function.
15714 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15715 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15716 (spu_function_arg_advance): Use it, supplying a valid register number.
15717 (TARGET_HARD_REGNO_NREGS): Redefine.
15718 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15719 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15720 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15721 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15722 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15723 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15724 (CLASS_MAX_NREGS): Remove copy of old documentation.
15725 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15726 (visium_hard_regno_nregs): New function.
15727 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15728 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15729 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15730 xtensa_hard_regno_nregs): New function.
15731 * system.h (HARD_REGNO_NREGS): Poison.
15733 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15735 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15736 hard_regno_nregs instead of HARD_REGNO_NREGS.
15737 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15738 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15739 (c6x_expand_epilogue): Likewise.
15740 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15741 (frv_read_iacc_argument): Likewise.
15742 * config/sh/sh.c: Include regs.h.
15743 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15744 (regs_used): Likewise.
15745 (output_stack_adjust): Likewise.
15746 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15747 * expmed.c: Include regs.h.
15748 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15749 * ree.c: Include regs.h.
15750 (combine_reaching_defs): Use hard_regno_nregs instead of
15752 (add_removable_extension): Likewise.
15754 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15756 * regs.h (hard_regno_nregs): Turn into a function.
15757 (end_hard_regno): Update accordingly.
15758 * caller-save.c (setup_save_areas): Likewise.
15759 (save_call_clobbered_regs): Likewise.
15760 (replace_reg_with_saved_mem): Likewise.
15761 (insert_restore): Likewise.
15762 (insert_save): Likewise.
15763 * combine.c (can_change_dest_mode): Likewise.
15764 (move_deaths): Likewise.
15765 (distribute_notes): Likewise.
15766 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15767 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15768 (rs6000_split_multireg_move): Likewise.
15769 (rs6000_register_move_cost): Likewise.
15770 (rs6000_memory_move_cost): Likewise.
15771 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15772 (rs6000_split_multireg_move): Likewise.
15773 (rs6000_register_move_cost): Likewise.
15774 (rs6000_memory_move_cost): Likewise.
15775 * cselib.c (cselib_reset_table): Likewise.
15776 (cselib_lookup_1): Likewise.
15777 * emit-rtl.c (set_mode_and_regno): Likewise.
15778 * function.c (aggregate_value_p): Likewise.
15779 * ira-color.c (setup_profitable_hard_regs): Likewise.
15780 (check_hard_reg_p): Likewise.
15781 (calculate_saved_nregs): Likewise.
15782 (assign_hard_reg): Likewise.
15783 (improve_allocation): Likewise.
15784 (calculate_spill_cost): Likewise.
15785 * ira-emit.c (modify_move_list): Likewise.
15786 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15787 (ira_hard_reg_in_set_p): Likewise.
15788 * ira.c (setup_reg_mode_hard_regset): Likewise.
15789 (clarify_prohibited_class_mode_regs): Likewise.
15790 (check_allocation): Likewise.
15791 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15792 (lra_setup_reg_renumber): Likewise.
15793 (setup_try_hard_regno_pseudos): Likewise.
15794 (spill_for): Likewise.
15795 (assign_hard_regno): Likewise.
15796 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15797 * lra-constraints.c (in_class_p): Likewise.
15798 (lra_constraint_offset): Likewise.
15799 (simplify_operand_subreg): Likewise.
15800 (lra_constraints): Likewise.
15801 (split_reg): Likewise.
15802 (split_if_necessary): Likewise.
15803 (invariant_p): Likewise.
15804 (inherit_in_ebb): Likewise.
15805 * lra-lives.c (process_bb_lives): Likewise.
15806 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15807 (get_hard_regs): Likewise.
15808 (do_remat): Likewise.
15809 * lra-spills.c (assign_spill_hard_regs): Likewise.
15810 * mode-switching.c (create_pre_exit): Likewise.
15811 * postreload.c (reload_combine_recognize_pattern): Likewise.
15812 * recog.c (peep2_find_free_register): Likewise.
15813 * regcprop.c (kill_value_regno): Likewise.
15814 (set_value_regno): Likewise.
15815 (copy_value): Likewise.
15816 (maybe_mode_change): Likewise.
15817 (find_oldest_value_reg): Likewise.
15818 (copyprop_hardreg_forward_1): Likewise.
15819 * regrename.c (check_new_reg_p): Likewise.
15820 (regrename_do_replace): Likewise.
15821 * reload.c (push_reload): Likewise.
15822 (combine_reloads): Likewise.
15823 (find_dummy_reload): Likewise.
15824 (operands_match_p): Likewise.
15825 (find_reloads): Likewise.
15826 (find_equiv_reg): Likewise.
15827 (reload_adjust_reg_for_mode): Likewise.
15828 * reload1.c (count_pseudo): Likewise.
15829 (count_spilled_pseudo): Likewise.
15830 (find_reg): Likewise.
15831 (clear_reload_reg_in_use): Likewise.
15832 (free_for_value_p): Likewise.
15833 (allocate_reload_reg): Likewise.
15834 (choose_reload_regs): Likewise.
15835 (reload_adjust_reg_for_temp): Likewise.
15836 (emit_reload_insns): Likewise.
15837 (delete_output_reload): Likewise.
15838 * rtlanal.c (subreg_get_info): Likewise.
15839 * sched-deps.c (sched_analyze_reg): Likewise.
15840 * sel-sched.c (init_regs_for_mode): Likewise.
15841 (mark_unavailable_hard_regs): Likewise.
15842 (choose_best_reg_1): Likewise.
15843 (verify_target_availability): Likewise.
15844 * valtrack.c (dead_debug_insert_temp): Likewise.
15845 * var-tracking.c (track_loc_p): Likewise.
15846 (emit_note_insn_var_location): Likewise.
15847 * varasm.c (make_decl_rtl): Likewise.
15848 * reginfo.c (choose_hard_reg_mode): Likewise.
15849 (init_reg_modes_target): Refer directly to
15850 this_target_regs->x_hard_regno_nregs.
15852 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15854 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15855 instead of hard_regno_nregs.
15857 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15859 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15860 end_hard_regno instead of hard_regno_nregs.
15861 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15862 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15863 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15864 * ira-color.c (improve_allocation): Likewise.
15865 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15866 * lra-lives.c (mark_regno_live): Likewise.
15867 (mark_regno_dead): Likewise.
15868 * lra-remat.c (operand_to_remat): Likewise.
15869 * lra.c (collect_non_operand_hard_regs): Likewise.
15870 * postreload.c (reload_combine_note_store): Likewise.
15871 (move2add_valid_value_p): Likewise.
15872 * reload.c (regno_clobbered_p): Likewise.
15874 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15876 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15878 * config/v850/v850.c (v850_reorg): Likewise.
15879 * reload.c (refers_to_regno_for_reload_p): Likewise.
15880 (find_equiv_reg): Likewise.
15881 * reload1.c (reload_reg_reaches_end_p): Likewise.
15883 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15885 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15887 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15888 * config/arm/arm.c (output_move_neon): Likewise.
15889 (arm_attr_length_move_neon): Likewise.
15890 (neon_split_vcombine): Likewise.
15891 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15892 (c6x_mark_reg_written): Likewise.
15893 (c6x_dwarf_register_span): Likewise.
15894 * config/i386/i386.c (ix86_save_reg): Likewise.
15895 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15896 (rws_access_reg): Likewise.
15897 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15898 * mode-switching.c (create_pre_exit): Likewise.
15899 * ree.c (combine_reaching_defs): Likewise.
15900 (add_removable_extension): Likewise.
15901 * regcprop.c (find_oldest_value_reg): Likewise.
15902 (copyprop_hardreg_forward_1): Likewise.
15903 * reload.c (reload_inner_reg_of_subreg): Likewise.
15904 (push_reload): Likewise.
15905 (combine_reloads): Likewise.
15906 (find_dummy_reload): Likewise.
15907 (reload_adjust_reg_for_mode): Likewise.
15908 * reload1.c (find_reload_regs): Likewise.
15909 (forget_old_reloads_1): Likewise.
15910 (reload_reg_free_for_value_p): Likewise.
15911 (reload_adjust_reg_for_temp): Likewise.
15912 (emit_reload_insns): Likewise.
15913 (delete_output_reload): Likewise.
15914 * sel-sched.c (choose_best_reg_1): Likewise.
15915 (choose_best_pseudo_reg): Likewise.
15917 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15918 Alan Hayward <alan.hayward@arm.com>
15919 David Sherwood <david.sherwood@arm.com>
15921 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15922 * target.def (slow_unaligned_access): New hook.
15923 * targhooks.h (default_slow_unaligned_access): Declare.
15924 * targhooks.c (default_slow_unaligned_access): New function.
15925 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15926 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15927 * doc/tm.texi: Regenerate.
15928 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15929 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15930 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15932 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15933 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15935 (rs6000_slow_unaligned_access): New function.
15936 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15937 (expand_block_compare): Likewise.
15938 (expand_strn_compare): Likewise.
15939 (rs6000_rtx_costs): Likewise.
15940 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15941 (riscv_slow_unaligned_access): Likewise.
15942 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15943 (riscv_slow_unaligned_access_p): ...this and make static.
15944 (riscv_option_override): Update accordingly.
15945 (riscv_slow_unaligned_access): New function.
15946 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15947 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15948 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15949 (rs6000_slow_unaligned_access): New function.
15950 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15951 (rs6000_rtx_costs): Likewise.
15952 * config/rs6000/rs6000-string.c (expand_block_compare)
15953 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15954 of SLOW_UNALIGNED_ACCESS.
15955 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15956 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15957 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15958 of SLOW_UNALIGNED_ACCESS.
15959 * expmed.c (simple_mem_bitfield_p): Likewise.
15960 * expr.c (alignment_for_piecewise_move): Likewise.
15961 (emit_group_load_1): Likewise.
15962 (emit_group_store): Likewise.
15963 (copy_blkmode_from_reg): Likewise.
15964 (emit_push_insn): Likewise.
15965 (expand_assignment): Likewise.
15966 (store_field): Likewise.
15967 (expand_expr_real_1): Likewise.
15968 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15969 * lra-constraints.c (simplify_operand_subreg): Likewise.
15970 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15971 * gimple-ssa-store-merging.c: Likewise in block comment at start
15973 * tree-ssa-strlen.c: Include target.h.
15974 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15975 of SLOW_UNALIGNED_ACCESS.
15976 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15978 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15980 PR rtl-optimization/82185
15981 * expmed.c (emit_store_flag_int): Only test tem if it has been
15984 2017-09-12 Richard Biener <rguenther@suse.de>
15986 PR middle-end/82149
15987 * match.pd ((FTYPE) N CMP CST): Fix typo.
15989 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15991 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15993 (mips_near_type_p): Add 'short_call' attribute as a synonym
15995 * doc/extend.texi (short_call): Document new function attribute.
15997 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16000 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16001 assertion check that in the condition.
16002 (get_atomic_generic_size): Likewise. Before testing if parameter
16003 has pointer type, if it has array type, call for C++
16004 default_conversion to perform array-to-pointer conversion.
16006 2017-09-12 Richard Biener <rguenther@suse.de>
16008 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16009 for operations we cannot scalarize.
16011 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16013 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16014 vectors heap vectors. Clean up comments.
16015 Make visited_bbs a reference.
16016 (profitable_jump_thread_path): Make GC
16017 vectors heap vectors. Clean up comments.
16019 (convert_and_register_jump_thread_path): Make GC vectors heap
16021 (check_subpath_and_update_thread_path): Same. Clean up comments.
16022 Make visited_bbs a reference.
16023 (handle_phi): Abstract common code to to
16024 register_jump_thread_path_if_profitable.
16025 Rename VAR_BB to DEF_BB.
16027 Make GC vectors heap vectors.
16028 Make visited_bbs a reference.
16029 (handle_assignment): Same.
16030 (register_jump_thread_path_if_profitable): New.
16031 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16033 Make GC vectors heap vectors. Clean up comments.
16034 Make visited_bbs a reference.
16035 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16036 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16039 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16042 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16043 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16045 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16047 Revert r251800 and r251799.
16049 2017-09-11 Martin Jambor <mjambor@suse.cz>
16052 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16053 arguments in advance.
16054 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16055 use it to find predecessor edges.
16056 (naive_outof_ssa): Collect vector of predecessors.
16058 2017-09-08 Jason Merrill <jason@redhat.com>
16060 PR c++/70029 - ICE with ref-qualifier and -flto
16061 * langhooks.h (struct lang_hooks_for_types): Add
16062 copy_lang_qualifiers.
16063 * attribs.c (build_type_attribute_qual_variant): Use it.
16064 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16066 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16067 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16069 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16072 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16073 (*mulsi3_sp64): New instruction.
16074 (mulsi3): New expander.
16076 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16078 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16080 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16082 * sancov.c: Include memmodel.h.
16084 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16087 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16090 2017-09-07 Carl Love <cel@us.ibm.com>
16092 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16093 the sldi instruction.
16095 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16097 * sancov.c: Include tm_p.h.
16099 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16102 * output.h (switch_to_other_text_partition): New declaration.
16103 * varasm.c (switch_to_other_text_partition): New function.
16104 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16105 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16106 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16107 to the other text partition before emitting LCL label and switch back
16108 after emitting the word after it.
16110 2017-09-07 Richard Biener <rguenther@suse.de>
16112 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16113 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16114 critical edges here, after loop init.
16115 (pass_data_pre): Remove PROP_no_crit_edges flags.
16116 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16117 for valueization of call args to avoid leaking VN_TOP.
16118 (visit_use): Assert we do not visit default defs.
16119 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16120 Use error_mark_node to more easily detect leaking VN_TOP.
16121 All default-defs are varying, not VN_TOP. Mark them visited.
16122 (run_scc_vn): Make code match comment.
16124 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16126 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16127 OPTION_MASK_FLOAT128_KEYWORD.
16128 (POWERPC_MASKS): Likewise.
16129 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16130 support for the -mfloat128-type option, and make -mfloat128
16131 default on PowerPC Linux systems. Define or undefine
16132 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16133 Define __float128 to be __ieee128 if IEEE 128-bit support is
16134 enabled, or undefine it.
16135 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16136 Delete defining __FLOAT128_TYPE__.
16137 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16138 -mfloat128-type option and make -mfloat128 default on PowerPC
16140 (TARGET_FLOAT128_TYPE): Likewise.
16141 (-mfloat128-type): Likewise.
16142 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16143 Delete the -mfloat128-type option and make -mfloat128 default on
16144 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16145 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16146 128-bit floating point is enabled. Change tests from using
16147 -mfloat128-type to -mfloat128.
16148 (rs6000_mangle_type): Use the correct mangling for the __float128
16149 type even if normal long double is restricted to 64-bits.
16150 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16152 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16153 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16154 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16156 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16157 documentation for -mfloat128.
16159 2017-09-06 Olivier Hainque <hainque@adacore.com>
16161 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16163 2017-09-06 Wish Wu <wishwu007@gmail.com>
16164 Jakub Jelinek <jakub@redhat.com>
16166 * asan.c (initialize_sanitizer_builtins): Add
16167 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16168 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16169 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16170 BT_FN_VOID_UINT64_PTR variables.
16171 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16172 (BT_FN_VOID_UINT16_UINT16): Likewise.
16173 (BT_FN_VOID_UINT32_UINT32): Likewise.
16174 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16175 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16176 (BT_FN_VOID_UINT64_PTR): Likewise.
16177 * common.opt (flag_sanitize_coverage): New variable.
16178 (fsanitize-coverage=trace-pc): Remove.
16179 (fsanitize-coverage=): Add.
16180 * flag-types.h (enum sanitize_coverage_code): New enum.
16181 * fold-const.c (fold_range_test): Disable non-short-circuit
16182 optimization if flag_sanitize_coverage.
16183 (fold_truth_andor): Likewise.
16184 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16185 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16186 (coverage_sanitizer_opts): New array.
16187 (get_closest_sanitizer_option): Add OPTS argument, handle also
16188 OPT_fsanitize_coverage_.
16189 (parse_sanitizer_options): Adjusted to also handle
16190 OPT_fsanitize_coverage_.
16191 (common_handle_option): Add OPT_fsanitize_coverage_.
16192 * sancov.c (instrument_comparison, instrument_switch): New function.
16193 (sancov_pass): Add trace-cmp support.
16194 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16195 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16196 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16197 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16198 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16199 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16200 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16201 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16202 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16203 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16205 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16207 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16208 error. Only quit immediately if parsing is complete.
16209 (BEGIN): Initialize fatal_err and parse_done.
16210 (begin fpu, end fpu): Check number of arguments.
16211 (begin arch, end arch): Likewise.
16212 (begin cpu, end cpu): Likewise.
16213 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16214 (optalias): Likewise.
16216 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16218 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16219 * config/arm/arm-isa.h: Delete. Move definitions to ...
16220 * arm-cpus.in: ... here. Use new feature and fgroup values.
16221 * config/arm/arm.c (arm_option_override): Use lower case for feature
16223 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16224 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16225 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16227 (print_isa_bits_for): New function.
16228 (gen_isa): New function.
16229 (gen_comm_data): Use print_isa_bits_for.
16230 (define feature): New keyword.
16231 (define fgroup): New keyword.
16232 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16233 (arm-isa.h): Add rule to generate file.
16234 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16235 case for feature bit names.
16237 2017-09-06 Richard Biener <rguenther@suse.de>
16239 * tree-ssa-pre.c (NECESSARY): Remove.
16240 (create_expression_by_pieces): Do not touch pass-local flags.
16241 (insert_into_preds_of_block): Likewise.
16242 (do_pre_regular_insertion): Likewise.
16243 (eliminate_insert): Likewise.
16244 (eliminate_dom_walker::before_dom_children): Likewise.
16245 (fini_eliminate): Do not look at inserted_exprs.
16246 (mark_operand_necessary): Remove.
16247 (remove_dead_inserted_code): Replace with simple work-list
16248 algorithm based on inserted_exprs and SSA uses.
16249 (pass_pre::execute): Re-order fini_eliminate and
16250 remove_dead_inserted_code.
16252 2017-09-06 Olivier Hainque <hainque@adacore.com>
16254 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16255 for VxWorks 7. Adjust surrounding comments.
16257 2017-09-06 Richard Biener <rguenther@suse.de>
16259 * gimple-ssa-strength-reduction.c
16260 (find_candidates_dom_walker::before_dom_children): Also allow
16263 2017-09-06 Richard Biener <rguenther@suse.de>
16265 PR tree-optimization/82108
16266 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16267 for gap in the non-permutation SLP case.
16269 2017-09-06 Martin Jambor <mjambor@suse.cz>
16271 PR tree-optimization/82078
16272 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16273 add_access_to_work_queue...
16274 (build_accesses_from_assign): ...here.
16275 (propagate_all_subaccesses): Make sure racc is the group
16276 representative, if there is one.
16278 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16280 PR middle-end/82095
16281 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16284 2017-09-06 Richard Biener <rguenther@suse.de>
16286 * gimple-ssa-strength-reduction.c
16287 (find_candidates_dom_walker::before_doom_children): Use a
16288 type and not a mode check.
16290 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16293 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16295 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16297 2017-09-05 Jeff Law <law@redhat.com>
16299 PR tree-optimization/64910
16300 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16301 cases where we have 3 or more operands.
16303 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16305 PR middle-end/81768
16306 * omp-low.c (lower_omp_for): Recompute tree invariant if
16307 gimple_omp_for_initial/final is ADDR_EXPR.
16309 PR middle-end/81768
16310 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16311 into gimple val before gimplification fo the COND_EXPR.
16313 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16315 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16316 REGION_COPY argument.
16317 (thread_through_all_blocks): Remove unused argument to
16318 duplicate_thread_path.
16320 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16321 Alan Hayward <alan.hayward@arm.com>
16322 David Sherwood <david.sherwood@arm.com>
16324 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16325 Take a scalar_mode rather than a machine_mode.
16326 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16327 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16328 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16329 (aarch64_gen_adjusted_ldpstp): Likewise.
16330 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16332 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16333 Alan Hayward <alan.hayward@arm.com>
16334 David Sherwood <david.sherwood@arm.com>
16336 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16337 Take a scalar_int_mode instead of a machine_mode.
16338 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16339 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16340 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16341 (aarch64_simd_attr_length_rglist): Delete.
16342 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16343 a scalar_int_mode instead of a machine_mode.
16344 (aarch64_add_offset): Likewise.
16345 (aarch64_internal_mov_immediate): Likewise
16346 (aarch64_add_constant_internal): Likewise.
16347 (aarch64_add_constant): Likewise.
16348 (aarch64_movw_imm): Likewise.
16349 (aarch64_rtx_arith_op_extract_p): Likewise.
16350 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16351 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16352 Remove assert that the mode isn't a vector.
16353 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16354 (aarch64_expand_mov_immediate): Update calls after above changes.
16355 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16356 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16357 (aarch64_move_imm): Likewise.
16358 (aarch64_can_const_movi_rtx_p): Likewise.
16359 (aarch64_strip_extend): Likewise.
16360 (aarch64_extr_rtx_p): Likewise.
16361 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16362 a CONST_INT when the mode parameter is VOIDmode.
16363 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16365 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16367 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16368 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16369 (bitwise_type_for_mode): Update accordingly.
16371 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16373 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16374 * stor-layout.c (mode_for_size_tree): Likewise.
16375 (mode_for_array): Update accordingly.
16376 (layout_decl): Likewise.
16377 (compute_record_mode): Likewise. Only set the mode once.
16379 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16381 * target.def (get_mask_mode): Change return type to opt_mode.
16383 * doc/tm.texi: Regenerate.
16384 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16385 * targhooks.c (default_get_mask_mode): Likewise.
16386 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16387 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16388 targetm.get_mask_mode.
16389 * tree.c (build_truth_vector_type): Likewise.
16391 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16393 * machmode.h (mode_for_vector): Return an opt_mode.
16394 * stor-layout.c (mode_for_vector): Likewise.
16395 (mode_for_int_vector): Update accordingly.
16396 (layout_type): Likewise.
16397 * config/i386/i386.c (emit_memmov): Likewise.
16398 (ix86_expand_set_or_movmem): Likewise.
16399 (ix86_expand_vector_init): Likewise.
16400 (ix86_get_mask_mode): Likewise.
16401 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16403 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16404 * expmed.c (extract_bit_field_1): Likewise.
16405 * expr.c (expand_expr_real_2): Likewise.
16406 * optabs-query.c (can_vec_perm_p): Likewise.
16407 (can_vec_mask_load_store_p): Likewise.
16408 * optabs.c (expand_vec_perm): Likewise.
16409 * targhooks.c (default_get_mask_mode): Likewise.
16410 * tree-vect-stmts.c (vectorizable_store): Likewise.
16411 (vectorizable_load): Likewise.
16412 (get_vectype_for_scalar_type_and_size): Likewise.
16414 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16416 * machmode.h (mode_for_int_vector): New function.
16417 * stor-layout.c (mode_for_int_vector): Likewise.
16418 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16419 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16420 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16421 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16422 (s390_expand_vcond): Likewise.
16424 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16426 * machmode.h (opt_machine_mode): New type.
16427 (opt_mode<T>): Allow construction from anything that can be
16429 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16430 (mode_for_size): Return an opt_machine_mode.
16431 * stor-layout.c (mode_for_size): Likewise.
16432 (mode_for_size_tree): Update call accordingly.
16433 (bitwise_mode_for_mode): Likewise.
16434 (make_fract_type): Likewise.
16435 (make_accum_type): Likewise.
16436 * caller-save.c (replace_reg_with_saved_mem): Update call
16438 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16439 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16440 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16441 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16442 * expmed.c (extract_bit_field_1): Likewise.
16443 * reload.c (get_secondary_mem): Likewise.
16444 * varasm.c (assemble_integer): Likewise.
16445 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16448 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16450 * machmode.h (decimal_float_mode_for_size): New function.
16451 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16452 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16453 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16454 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16455 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16456 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16458 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16460 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16461 (get_builtin_sync_mode): Likewise.
16462 (expand_ifn_atomic_compare_exchange): Likewise.
16463 (expand_builtin_atomic_clear): Likewise.
16464 (expand_builtin_atomic_test_and_set): Likewise.
16465 (fold_builtin_atomic_always_lock_free): Likewise.
16466 * calls.c (compute_argument_addresses): Likewise.
16467 (emit_library_call_value_1): Likewise.
16468 (store_one_arg): Likewise.
16469 * combine.c (combine_instructions): Likewise.
16470 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16471 * config/arm/arm.c (arm_function_value): Likewise.
16472 (aapcs_allocate_return_reg): Likewise.
16473 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16474 * config/i386/i386.c (construct_container): Likewise.
16475 (ix86_gimplify_va_arg): Likewise.
16476 (ix86_expand_sse_cmp): Likewise.
16477 (emit_memmov): Likewise.
16478 (emit_memset): Likewise.
16479 (expand_small_movmem_or_setmem): Likewise.
16480 (ix86_expand_pextr): Likewise.
16481 (ix86_expand_pinsr): Likewise.
16482 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16483 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16485 * config/mips/mips.c (mips_function_value_1) Likewise.
16486 (mips_block_move_straight): Likewise.
16487 (mips_expand_ins_as_unaligned_store): Likewise.
16488 * config/powerpcspe/powerpcspe.c
16489 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16490 (rs6000_darwin64_record_arg_flush): Likewise.
16491 * config/rs6000/rs6000.c
16492 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16493 (rs6000_darwin64_record_arg_flush): Likewise.
16494 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16495 (sparc_function_value_1): Likewise.
16496 * config/spu/spu.c (adjust_operand): Likewise.
16497 (spu_emit_branch_or_set): Likewise.
16498 (arith_immediate_p): Likewise.
16499 * emit-rtl.c (gen_lowpart_common): Likewise.
16500 * expr.c (expand_expr_real_1): Likewise.
16501 * function.c (assign_parm_setup_block): Likewise.
16502 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16503 * reload1.c (alter_reg): Likewise.
16504 * stor-layout.c (mode_for_vector): Likewise.
16505 (layout_type): Likewise.
16507 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16509 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16510 (spu_convert_move): Likewise.
16511 * lower-subreg.c (resolve_simple_move): Likewise.
16513 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16516 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16517 define_insn to a define_expand.
16518 (altivec_vsum2sws_direct): New define_insn.
16519 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16521 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16523 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16526 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16531 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16532 realign stack if stack alignment needed is less than incoming
16535 2017-09-05 Marek Polacek <polacek@redhat.com>
16538 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16541 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16543 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16545 2017-09-05 Richard Biener <rguenther@suse.de>
16547 PR tree-optimization/82084
16548 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16550 2017-09-05 Richard Biener <rguenther@suse.de>
16552 PR tree-optimization/82102
16553 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16555 2017-09-05 Martin Liska <mliska@suse.cz>
16557 PR tree-optimization/82032
16558 * tree-cfg.c (generate_range_test): New function.
16559 * tree-cfg.h (generate_range_test): Declared here.
16560 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16561 (cleanup_control_expr_graph): Use it.
16562 * tree-switch-conversion.c (try_switch_expansion): Remove
16564 (emit_case_nodes): Use generate_range_test.
16566 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16569 * config/i386/i386.md (*<btsc><mode>_mask): Add
16570 TARGET_USE_BT to insn constraint.
16571 (*btr<mode>_mask): Ditto.
16573 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16575 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16576 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16578 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16581 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16582 TARGET_NEON and TARGET_IWMMXT.
16583 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16584 TARGET_NEON and TARGET_IWMMXT.
16585 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16587 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16589 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16590 (ix86_rewrite_tls_address): Ditto.
16591 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16592 (ix86_rewrite_tls_address_1): Ditto.
16593 (ix86_rewrite_tls_address): Ditto.
16594 * config/i386/predicates.md (tls_address_pattern): New predicate.
16595 * config/i386/i386.md (TLS address splitter): New splitter.
16597 2017-09-04 Richard Biener <rguenther@suse.de>
16599 PR tree-optimization/82084
16600 * fold-const.h (can_native_encode_string_p): Declare.
16601 * fold-const.c (can_native_encode_string_p): Factor out from ...
16602 (native_encode_string): ... here.
16603 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16604 vectorizing stores from constants we later cannot handle.
16606 2017-09-04 Marek Polacek <polacek@redhat.com>
16609 * doc/invoke.texi: Update -Wtautological-compare documentation.
16611 2017-09-04 Jeff Law <law@redhat.com>
16613 PR tree-optimization/64910
16614 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16615 swap the first and last operand if the last is a constant.
16617 2017-09-04 Marek Polacek <polacek@redhat.com>
16620 * convert.c (do_narrow): When sanitizing signed integer overflows,
16621 bail out for signed types.
16622 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16624 2017-09-04 Richard Biener <rguenther@suse.de>
16626 PR tree-optimization/82060
16627 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16628 Move devirtualization after stmt folding and before EH/AB/noreturn
16629 cleanup to get the stmt refs canonicalized. Use a bool instead
16630 of gimple_modified_p since that doesn't work for NOPs. Schedule
16631 NOPs generated by folding for removal.
16633 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16634 Alan Hayward <alan.hayward@arm.com>
16635 David Sherwood <david.sherwood@arm.com>
16637 * coretypes.h (pad_direction): New enum.
16638 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16639 (FUNCTION_ARG_PADDING): Likewise.
16640 * target.def (function_arg_padding): New hook.
16641 * targhooks.h (default_function_arg_padding): Declare.
16642 * targhooks.c (default_function_arg_padding): New function.
16643 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16644 (TARGET_FUNCTION_ARG_PADDING): ...this.
16645 * doc/tm.texi: Regenerate.
16646 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16647 instead of direction.
16648 (compute_argument_addresses): Likewise.
16649 (load_register_parameters): Likewise.
16650 (emit_library_call_value_1): Likewise.
16651 (store_one_arg): Use targetm.calls.function_arg_padding instead
16652 of FUNCTION_ARG_PADDING.
16653 (must_pass_in_stack_var_size_or_pad): Likewise.
16654 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16655 (emit_group_store): Likewise.
16656 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16657 instead of FUNCTION_ARG_PADDING.
16658 (emit_push_insn): Likewise, and propagate enum change throughout
16660 * function.h (direction): Delete.
16661 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16663 * function.c (assign_parm_find_stack_rtl): Likewise.
16664 (assign_parm_setup_block_p): Likewise.
16665 (assign_parm_setup_block): Likewise.
16666 (gimplify_parameters): Likewise.
16667 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16668 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16670 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16671 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16672 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16673 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16674 (aarch64_function_arg_padding): ...this new function.
16675 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16676 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16677 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16678 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16679 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16680 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16681 (arm_pad_arg_upward): Replace with...
16682 (arm_function_arg_padding): ...this new function.
16683 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16685 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16686 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16687 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16688 (ia64_hpux_function_arg_padding): Replace with...
16689 (ia64_function_arg_padding): ...this new function. Use pad_direction
16690 instead of direction. Check for TARGET_HPUX.
16691 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16692 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16693 (iq2000_function_arg_padding): New function.
16694 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16695 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16696 (mips_function_arg_padding): ...this new function.
16697 (mips_pad_reg_upward): Update accordingly.
16698 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16699 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16700 targetm.calls.function_arg_padding.
16701 (FUNCTION_ARG_PADDING): Delete.
16702 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16703 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16704 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16705 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16706 (nios2_block_reg_padding): Return pad_direction instead of direction.
16707 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16708 instead of direction.
16709 (nios2_function_arg_padding): Likewise. Make static.
16710 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16711 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16712 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16713 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16714 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16715 (pa_function_arg_padding): Make static. Return pad_direction instead
16717 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16718 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16719 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16720 instead of direction. Use targetm.calls.function_arg_padding.
16721 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16722 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16723 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16724 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16725 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16727 (function_arg_padding): Rename to...
16728 (rs6000_function_arg_padding): ...this. Make static. Return
16729 pad_direction instead of direction.
16730 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16731 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16732 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16733 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16734 instead of direction. Use targetm.calls.function_arg_padding.
16735 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16736 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16737 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16738 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16739 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16740 (function_arg_padding): Rename to...
16741 (rs6000_function_arg_padding): ...this. Make static. Return
16742 pad_direction instead of direction.
16743 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16744 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16745 * config/s390/s390.c (s390_function_arg_padding): New function.
16746 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16747 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16748 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16749 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16750 (function_arg_padding): Rename to...
16751 (sparc_function_arg_padding): ...this. Make static. Return
16752 pad_direction instead of direction.
16753 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16754 * config/spu/spu.c (spu_function_arg_padding): New function.
16755 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16756 * system.h (FUNCTION_ARG_PADDING): Poison.
16758 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16759 Alan Hayward <alan.hayward@arm.com>
16760 David Sherwood <david.sherwood@arm.com>
16762 * target.def (modes_tieable_p): New hook.
16763 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16764 (TARGET_MODES_TIEABLE_P): ...this.
16765 * doc/tm.texi.in: Regenerate.
16766 * hooks.h (hook_bool_mode_mode_true): Declare.
16767 * hooks.c (hook_bool_mode_mode_true): New function.
16768 * combine.c (subst): Use targetm.modes_tieable_p instead of
16770 * dse.c (find_shift_sequence): Likewise.
16771 * expmed.c (extract_low_bits): Likewise.
16772 * lower-subreg.c: Include target.h.
16773 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16775 * rtlanal.c (rtx_cost): Likewise.
16776 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16777 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16778 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16779 (TARGET_MODES_TIEABLE_P): Redefine.
16780 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16781 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16782 (TARGET_MODES_TIEABLE_P): Redefine.
16783 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16784 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16785 (arc_modes_tieable_p): New function.
16786 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16787 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16788 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16789 (arm_modes_tieable_p): Make static.
16790 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16791 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16792 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16793 (TARGET_MODES_TIEABLE_P): Redefine.
16794 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16795 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16796 (TARGET_MODES_TIEABLE_P): Redefine.
16797 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16798 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16799 (cr16_modes_tieable_p): New function.
16800 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16801 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16802 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16803 (TRULY_NOOP_TRUNCATION): Update comment.
16804 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16805 (TRULY_NOOP_TRUNCATION): Update comment.
16806 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16807 (frv_modes_tieable_p): New function.
16808 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16809 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16810 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16811 (TARGET_MODES_TIEABLE_P): Redefine.
16812 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16813 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16814 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16815 (TARGET_MODES_TIEABLE_P): Redefine.
16816 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16817 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16818 (ia64_modes_tieable_p): New function.
16819 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16820 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16821 (iq2000_modes_tieable_p): New function.
16822 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16823 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16824 (lm32_modes_tieable_p): New function.
16825 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16826 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16827 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16828 (TARGET_MODES_TIEABLE_P): Redefine.
16829 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16830 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16831 (m32r_modes_tieable_p): New function.
16832 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16833 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16834 (m68k_modes_tieable_p): New function.
16835 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16836 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16837 (mcore_modes_tieable_p): New function.
16838 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16839 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16841 (TARGET_MODES_TIEABLE_P): Redefine.
16842 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16843 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16844 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16845 (TARGET_MODES_TIEABLE_P): Redefine.
16846 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16847 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16848 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16849 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16850 (mn10300_modes_tieable_p): ...this and make static.
16851 (TARGET_MODES_TIEABLE_P): Redefine.
16852 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16853 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16854 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16855 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16856 (msp430_modes_tieable_p): Make static.
16857 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16858 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16859 (TARGET_MODES_TIEABLE_P): Redefine.
16860 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16861 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16862 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16863 (TARGET_MODES_TIEABLE_P): Redefine.
16864 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16865 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16866 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16867 (TARGET_MODES_TIEABLE_P): Redefine.
16868 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16869 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16870 (pdp11_modes_tieable_p): New function.
16871 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16872 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16873 (rs6000_modes_tieable_p): New function.
16874 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16875 * config/powerpcspe/powerpcspe.md: Update comment.
16876 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16877 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16878 (TARGET_MODES_TIEABLE_P): Redefine.
16879 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16880 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16881 (rl78_modes_tieable_p): New function.
16882 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16883 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16884 (rs6000_modes_tieable_p): New function.
16885 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16886 * config/rs6000/rs6000.md: Update comment.
16887 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16888 * config/rx/rx.c (rx_modes_tieable_p): New function.
16889 (TARGET_MODES_TIEABLE_P): Redefine.
16890 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16891 * config/s390/s390.c (s390_modes_tieable_p): New function.
16892 (TARGET_MODES_TIEABLE_P): Redefine.
16893 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16894 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16895 (sh_modes_tieable_p): New function.
16896 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16897 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16898 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16899 (sparc_modes_tieable_p): Make static.
16900 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16901 * config/spu/spu.c (spu_modes_tieable_p): New function.
16902 (TARGET_MODES_TIEABLE_P): Redefine.
16903 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16904 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16905 (TARGET_MODES_TIEABLE_P): Redefine.
16906 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16907 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16908 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16909 * config/v850/v850.c (v850_modes_tieable_p): New function.
16910 (TARGET_MODES_TIEABLE_P): Redefine.
16911 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16912 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16913 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16914 (visium_modes_tieable_p): New function.
16915 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16916 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16917 (xtensa_modes_tieable_p): New function.
16918 * system.h (MODES_TIEABLE_P): Poison.
16920 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16921 Alan Hayward <alan.hayward@arm.com>
16922 David Sherwood <david.sherwood@arm.com>
16924 * target.def (hard_regno_mode_ok): New hook.
16925 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16926 (TARGET_HARD_REGNO_MODE_OK): ...this.
16927 * doc/tm.texi.in: Regenerate.
16928 * hooks.h (hook_bool_uint_mode_true): Declare.
16929 * hooks.c (hook_bool_uint_mode_true): New function.
16930 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16931 HARD_REGNO_MODE_OK.
16932 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16933 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16934 instead of HARD_REGNO_MODE_OK.
16935 * caller-save.c: Include target.h.
16936 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16937 HARD_REGNO_MODE_OK.
16938 * combine.c (can_combine_p): Likewise.
16939 (combinable_i3pat): Likewise.
16940 (can_change_dest_mode): Likewise.
16941 * expr.c (init_expr_target): Likewise.
16942 (convert_move): Likewise.
16943 (convert_modes): Likewise.
16944 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16945 (setup_prohibited_mode_move_regs): Likewise.
16946 * ira.h (target_ira): Likewise.
16947 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16948 * lra-constraints.c (process_alt_operands): Likewise.
16949 (split_reg): Likewise.
16950 * recog.c (peep2_find_free_register): Likewise.
16951 * ree.c (combine_reaching_defs): Likewise.
16952 * regcprop.c (maybe_mode_change): Likewise.
16953 * reginfo.c (init_reg_sets_1): Likewise.
16954 (choose_hard_reg_mode): Likewise.
16955 (simplifiable_subregs): Likewise.
16956 * regrename.c (check_new_reg_p): Likewise.
16957 * reload.c (find_valid_class): Likewise.
16958 (find_valid_class_1): Likewise.
16959 (reload_inner_reg_of_subreg): Likewise.
16960 (push_reload): Likewise.
16961 (combine_reloads): Likewise.
16962 (find_dummy_reload): Likewise.
16963 (find_reloads): Likewise.
16964 * reload1.c (find_reg): Likewise.
16965 (set_reload_reg): Likewise.
16966 (allocate_reload_reg): Likewise.
16967 (choose_reload_regs): Likewise.
16968 (reload_adjust_reg_for_temp): Likewise.
16969 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16970 (simplify_subreg_regno): Likewise.
16971 * sel-sched.c (init_regs_for_mode): Likewise.
16972 * varasm.c (make_decl_rtl): Likewise.
16973 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16974 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16975 HARD_REGNO_MODE_OK.
16976 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16977 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16978 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16979 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16980 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16981 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16982 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16983 (arc_mode_class): Delete.
16984 (HARD_REGNO_MODE_OK): Delete.
16985 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16986 (arc_hard_regno_mode_ok): Rename old array to...
16987 (arc_hard_regno_mode_ok_modes): ...this.
16988 (arc_conditional_register_usage): Update accordingly.
16989 (arc_mode_class): Make static.
16990 (arc_hard_regno_mode_ok): New function.
16991 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16992 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16993 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16994 (arm_hard_regno_mode_ok): Make static.
16995 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16996 HARD_REGNO_MODE_OK.
16997 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16998 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16999 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17002 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17003 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17004 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17005 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17006 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17007 * config/bfin/predicates.md (valid_reg_operand): Use
17008 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17009 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17010 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17011 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17012 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17013 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17014 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17015 (cr16_hard_regno_mode_ok): Make static and return a bool.
17016 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17017 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17018 (cris_hard_regno_mode_ok): New function.
17019 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17020 (epiphany_mode_class): Delete.
17021 (HARD_REGNO_MODE_OK): Delete.
17022 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17023 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17024 (hard_regno_mode_ok): Rename to...
17025 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17026 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17027 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17028 HARD_REGNO_MODE_OK.
17029 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17030 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17031 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17032 (frv_hard_regno_mode_ok): Make static and return a bool.
17033 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17034 HARD_REGNO_MODE_OK.
17035 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17036 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17037 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17038 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17040 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17041 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17042 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17043 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17045 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17046 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17047 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17048 (ia64_hard_regno_mode_ok): New function.
17049 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17050 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17051 (iq2000_hard_regno_mode_ok): New function.
17052 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17053 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17054 (lm32_hard_regno_mode_ok): New function.
17055 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17056 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17057 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17058 instead of HARD_REGNO_MODE_OK.
17059 (m32c_hard_regno_ok): Rename to...
17060 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17061 (m32c_cannot_change_mode_class): Update accordingly.
17062 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17063 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17064 (m32r_mode_class): Delete.
17065 (HARD_REGNO_MODE_OK): Delete.
17066 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17067 (m32r_hard_regno_mode_ok): Rename to...
17068 (m32r_hard_regno_modes): ...this.
17069 (m32r_mode_class): Make static.
17070 (m32r_hard_regno_mode_ok): New function.
17071 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17072 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17073 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17074 (m68k_hard_regno_mode_ok): Make static.
17075 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17076 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17077 (mcore_hard_regno_mode_ok): New function.
17078 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17079 (HARD_REGNO_MODE_OK): Delete.
17080 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17082 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17083 (microblaze_hard_regno_mode_ok): New function.
17084 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17085 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17086 (mips_hard_regno_mode_ok): Delete.
17087 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17088 (mips_hard_regno_mode_ok_p): ...this and make static.
17089 (mips_hard_regno_mode_ok_p): Rename to...
17090 (mips_hard_regno_mode_ok_uncached): ...this.
17091 (mips_hard_regno_mode_ok): New function.
17092 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17093 of HARD_REGNO_MODE_OK.
17094 (mips_option_override): Update after above name changes.
17095 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17096 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17097 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17098 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17099 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17100 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17101 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17102 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17103 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17104 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17105 (msp430_hard_regno_mode_ok): Make static and return a bool.
17106 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17107 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17108 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17110 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17111 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17112 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17113 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17114 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17115 (PA_HARD_REGNO_MODE_OK): ...this
17116 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17117 (PA_HARD_REGNO_MODE_OK): ...this.
17118 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17119 (pa_hard_regno_mode_ok): New function.
17120 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17121 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17122 (pdp11_hard_regno_mode_ok): New function.
17123 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17124 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17126 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17128 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17129 (rs6000_hard_regno_mode_ok): Rename to...
17130 (rs6000_hard_regno_mode_ok_uncached): ...this.
17131 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17132 (rs6000_hard_regno_mode_ok): New function.
17133 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17134 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17135 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17136 (riscv_hard_regno_mode_ok): ...this and make static.
17137 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17138 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17139 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17140 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17141 (rl78_hard_regno_mode_ok): Make static and return bool.
17142 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17143 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17145 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17146 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17147 (rs6000_hard_regno_mode_ok): Rename to...
17148 (rs6000_hard_regno_mode_ok_uncached): ...this.
17149 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17150 (rs6000_hard_regno_mode_ok): New function.
17151 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17152 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17153 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17154 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17155 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17156 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17157 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17158 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17159 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17160 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17161 (sh_hard_regno_mode_ok): Make static.
17162 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17163 instead of HARD_REGNO_MODE_OK.
17164 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17165 (sparc_mode_class): Delete.
17166 (HARD_REGNO_MODE_OK): Delete.
17167 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17168 (hard_regno_mode_classes): Make static.
17169 (sparc_mode_class): Likewise.
17170 (sparc_hard_regno_mode_ok): New function.
17171 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17172 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17173 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17175 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17176 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17177 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17178 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17179 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17180 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17181 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17182 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17183 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17184 (visium_hard_regno_mode_ok): New function.
17185 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17186 instead of HARD_REGNO_MODE_OK.
17187 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17188 (HARD_REGNO_MODE_OK): Delete.
17189 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17190 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17191 (xtensa_option_override): Update accordingly.
17192 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17193 (xtensa_hard_regno_mode_ok): New function.
17194 * system.h (HARD_REGNO_MODE_OK): Poison.
17196 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17197 Alan Hayward <alan.hayward@arm.com>
17198 David Sherwood <david.sherwood@arm.com>
17200 * target.def (hard_regno_call_part_clobbered): New hook.
17201 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17202 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17203 * doc/tm.texi: Regenerate.
17204 * hooks.h (hook_bool_uint_mode_false): Declare.
17205 * hooks.c (hook_bool_uint_mode_false): New function.
17206 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17207 * cselib.c (cselib_process_insn): Use
17208 targetm.hard_regno_call_part_clobbered instead of
17209 HARD_REGNO_CALL_PART_CLOBBERED.
17210 * ira-conflicts.c (ira_build_conflicts): Likewise.
17211 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17212 * lra-constraints.c (need_for_call_save_p): Likewise.
17213 * lra-lives.c: Include target.h.
17214 (check_pseudos_live_through_calls): Use
17215 targetm.hard_regno_call_part_clobbered instead of
17216 HARD_REGNO_CALL_PART_CLOBBERED.
17217 * regcprop.c: Include target.h.
17218 (copyprop_hardreg_forward_1): Use
17219 targetm.hard_regno_call_part_clobbered instead of
17220 HARD_REGNO_CALL_PART_CLOBBERED.
17221 * reginfo.c (choose_hard_reg_mode): Likewise.
17222 * regrename.c (check_new_reg_p): Likewise.
17223 * reload.c (find_equiv_reg): Likewise.
17224 * reload1.c (emit_reload_insns): Likewise.
17225 * sched-deps.c (deps_analyze_insn): Likewise.
17226 * sel-sched.c (init_regs_for_mode): Likewise.
17227 (mark_unavailable_hard_regs): Likewise.
17228 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17229 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17230 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17232 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17233 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17234 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17236 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17238 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17239 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17240 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17242 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17243 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17244 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17246 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17247 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17249 * config/powerpcspe/powerpcspe.c
17250 (rs6000_hard_regno_call_part_clobbered): New function.
17251 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17252 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17253 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17255 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17256 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17257 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17259 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17260 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17261 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17263 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17264 Alan Hayward <alan.hayward@arm.com>
17265 David Sherwood <david.sherwood@arm.com>
17267 * rtl.h (subreg_memory_offset): Declare.
17268 * emit-rtl.c (subreg_memory_offset): New function.
17269 * expmed.c (store_bit_field_1): Use it.
17270 * expr.c (undefined_operand_subword_p): Likewise.
17271 * simplify-rtx.c (simplify_subreg): Likewise.
17273 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17275 PR rtl-optimization/57448
17278 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17279 using atomic_load pattern.
17280 (expand_atomic_store): Likewise.
17282 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17285 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17286 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17287 handling. Use replace_call_with_value with NULL instead of
17288 gsi_replace, unlink_stmt_vdef and release_defs.
17290 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17293 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17295 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17298 * rtl.h (emit_library_call_value_1): Declare.
17299 (emit_library_call): Replace declaration with a series of overloads.
17300 Remove the parameter count argument.
17301 (emit_library_call_value): Likewise.
17302 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17303 with an "rtx_mode_t *".
17304 (emit_library_call_value): Delete.
17305 (emit_library_call): Likewise.
17306 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17307 (asan_emit_allocas_unpoison): Likewise.
17308 * builtins.c (expand_builtin_powi): Likewise.
17309 (expand_asan_emit_allocas_unpoison): Likewise.
17310 * cfgexpand.c (expand_main_function): Likewise.
17311 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17312 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17313 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17314 * config/arm/arm.c (arm_trampoline_init): Likewise.
17315 (arm_call_tls_get_addr): Likewise.
17316 (arm_expand_divmod_libfunc): Likewise.
17317 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17318 (smulsi3_highpart): Likewise.
17319 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17320 (c6x_expand_compare): Likewise.
17321 (c6x_expand_movmem): Likewise.
17322 * config/frv/frv.c (frv_trampoline_init): Likewise.
17323 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17324 (ix86_expand_divmod_libfunc): Likewise.
17325 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17326 (ia64_expand_compare): Likewise.
17327 (ia64_profile_hook): Likewise.
17328 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17329 (nonlocal_goto): Likewise.
17330 (restore_stack_nonlocal): Likewise.
17331 * config/m32r/m32r.c (block_move_call): Likewise.
17332 (m32r_trampoline_init): Likewise.
17333 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17334 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17335 (m68k_call_m68k_read_tp): Likewise.
17336 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17337 (microblaze_expand_divide): Likewise.
17338 * config/mips/mips.h (mips_args): Likewise.
17339 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17340 (MIPS_ICACHE_SYNC): Likewise.
17341 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17342 (nios2_trampoline_init): Likewise.
17343 * config/pa/pa.c (hppa_tls_call): Likewise.
17344 (pa_trampoline_init): Likewise.
17345 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17346 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17347 (expand_strn_compare): Likewise.
17348 (rs6000_generate_compare): Likewise.
17349 (rs6000_expand_float128_convert): Likewise.
17350 (output_profile_hook): Likewise.
17351 (rs6000_trampoline_init): Likewise.
17352 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17353 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17354 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17355 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17356 (rs6000_generate_compare): Likewise.
17357 (rs6000_expand_float128_convert): Likewise.
17358 (output_profile_hook): Likewise.
17359 (rs6000_trampoline_init): Likewise.
17360 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17361 * config/sh/sh.c (sh_trampoline_init): Likewise.
17362 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17363 (sparc_emit_float_lib_cmp): Likewise.
17364 (sparc32_initialize_trampoline): Likewise.
17365 (sparc64_initialize_trampoline): Likewise.
17366 (sparc_profile_hook): Likewise.
17367 * config/spu/spu.c (ea_load_store): Likewise.
17368 * config/spu/spu.md (floatunssidf2): Likewise.
17369 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17370 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17371 * config/visium/visium.c (expand_block_move_4): Likewise.
17372 (expand_block_move_2): Likewise.
17373 (expand_block_move_1): Likewise.
17374 (expand_block_set_4): Likewise.
17375 (expand_block_set_2): Likewise.
17376 (expand_block_set_1): Likewise.
17377 (visium_trampoline_init): Likewise.
17378 (visium_profile_hook): Likewise.
17379 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17380 (xtensa_setup_frame_addresses): Likewise.
17381 (xtensa_trampoline_init): Likewise.
17382 * except.c (sjlj_emit_function_enter): Likewise.
17383 (sjlj_emit_function_exit): Likewise.
17384 * explow.c (allocate_dynamic_stack_space): Likewise.
17385 (probe_stack_range): Likewise.
17386 * expr.c (convert_mode_scalar): Likewise.
17387 * optabs.c (expand_binop): Likewise.
17388 (expand_twoval_binop_libfunc): Likewise.
17389 (expand_unop): Likewise.
17390 (prepare_cmp_insn): Likewise.
17391 (prepare_float_lib_cmp): Likewise.
17392 (expand_float): Likewise.
17393 (expand_fix): Likewise.
17394 (expand_fixed_convert): Likewise.
17395 (maybe_emit_sync_lock_test_and_set): Likewise.
17396 (expand_atomic_compare_and_swap): Likewise.
17397 (expand_mem_thread_fence): Likewise.
17398 (expand_atomic_fetch_op): Likewise.
17400 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17402 * doc/generic.texi (OpenACC): Adjust URL.
17403 * doc/invoke.texi (C Dialect Options): Ditto.
17405 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17407 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17408 predicate for operand 1. Add (m,<S>) constraint.
17409 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17410 Prevent memory operand 1 with register operand 2.
17412 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17414 PR rtl-optimization/82024
17415 * combine.c (try_combine): If the combination result is a PARALLEL,
17416 and we only need to retain the SET in there that would be placed
17417 at I2, check that we can place that at I3 instead, before doing so.
17419 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17422 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17424 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17425 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17428 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17429 Jeff Law <law@redhat.com>
17431 * varasm.c (bss_initializer_p): Do not put constants into .bss
17432 (categorize_decl_for_section): Handle bss_initializer_p returning
17433 false when DECL_INITIAL is NULL.
17435 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17438 * config/s390/s390.c (s390_can_inline_p): New function.
17440 2017-09-01 Jeff Law <law@redhat.com>
17442 PR tree-optimization/82052
17443 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17444 Always initialize the returned slot after a hash table miss
17445 when INSERT is true.
17447 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17449 * config/s390/s390.md (mem_signal_fence): Remove.
17450 * doc/md.texi (mem_signal_fence): Remove.
17451 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17453 * target-insns.def (mem_signal_fence): Remove.
17455 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17458 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17461 * asan.c (create_odr_indicator): Strip name encoding from assembler
17462 name before appending it after __odr_asan_.
17464 2017-09-01 Martin Liska <mliska@suse.cz>
17466 PR tree-optimization/82059
17467 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17468 frequency only when an edge is redirected.
17470 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17472 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17473 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17474 (arc_conditional_register_usage): Remove ARC600 lp_count
17476 (arc_file_start): Emit Tag_ARC_CPU_variation.
17477 (arc_can_use_doloop_p): New conditions to use ZOLs.
17478 (hwloop_fail): New function.
17479 (hwloop_optimize): Likewise.
17480 (hwloop_pattern_reg): Likewise.
17481 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17482 (arc_reorg_loops): New function, calls reorg_loops.
17483 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17484 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17486 (arc_loop_hazard): Remove function, functionality moved into
17488 (arc_hazard): Remove arc_loop_hazard call.
17489 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17490 into hwloop_optimize.
17491 (arc_label_align): Remove ZOL handling.
17492 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17493 * config/arc/arc.md (doloop_begin): Remove pattern.
17494 (doloop_begin_i): Likewise.
17495 (doloop_end_i): Likewise.
17496 (doloop_fallback): Likewise.
17497 (doloop_fallback_m): Likewise.
17498 (doloop_end): Reimplement expand.
17499 (arc_lp): New pattern for LP instruction.
17500 (loop_end): New pattern.
17501 (loop_fail): Likewise.
17502 (decrement_and_branch_until_zero): Likewise.
17503 * config/arc/arc.opt (mlpc-width): New option.
17504 * doc/invoke.texi (mlpc-width): Document option.
17506 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17508 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17509 (arc_ccfsm_advance): Fix checking for delay slots.
17510 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17512 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17514 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17516 (movhi_insn): Update store instruction constraint which are saving
17517 6-bit short immediates.
17518 (movsi_insn): Consider also short scaled load operations.
17519 (zero_extendhisi2_i): Use Usd constraint instead of T.
17520 (extendhisi2_i): Add q constraint.
17521 (arc_clzsi2): Add type and length attributes.
17522 (arc_ctzsi2): Likewise.
17523 * config/arc/constraints.md (Usc): Update constraint, the
17524 assembler can parse two relocations for a single instruction.
17526 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17528 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17529 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17531 2017-08-31 Olivier Hainque <hainque@adacore.com>
17533 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17534 match as powerpc-wrs-vxworks*.
17536 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17538 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17539 register constraint for by-element operand.
17540 (aarch64_mls_elt_merge<mode>): Likewise.
17542 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17544 * config/arc/arc.c (arc_can_follow_jump): Check for short
17547 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17549 * config.gcc: Use g.opt for arc.
17550 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17551 functionality moved to ...
17552 (legitimate_scaled_address_p): New function, ...here.
17553 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17554 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17555 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17557 (arc_override_options): Handle G option.
17558 (arc_output_pic_addr_const): Correct function definition.
17559 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17560 (arc_decl_anon_ns_mem_p): Delete.
17561 (arc_in_small_data_p): Overhaul this function to take into
17562 consideration the value given via G option.
17563 (arc_rewrite_small_data_1): Renamed and corrected old
17564 arc_rewrite_small_data function.
17565 (arc_rewrite_small_data): New function.
17566 (small_data_pattern): Don't use pic_offset_table_rtx.
17567 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17568 * config/arc/simdext.md (movmisalignv2hi): Use
17569 prepare_move_operands function.
17571 (movmisalign*): Likewise.
17572 * doc/invoke.texi (ARC options): Document -G option.
17574 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17576 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17578 * config/arc/arc.c (arc_print_operand): Output scalled address for
17579 sdata whenever is possible.
17580 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17581 load/stores are available.
17582 (compact_sda_memory_operand): Check for the alignment required by
17583 code density instructions.
17584 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17586 * config/arc/constraints.md (Usd): Update constraint.
17587 (Us0): New constraint.
17588 (Usc): Update constraint.
17590 2017-08-31 Richard Biener <rguenther@suse.de>
17592 PR middle-end/82054
17593 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17594 function only once.
17596 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17598 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17599 Resize type_signature.
17601 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17602 Alan Hayward <alan.hayward@arm.com>
17603 David Sherwood <david.sherwood@arm.com>
17605 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17606 subregs whose inner modes can be stored in GPRs.
17607 (aarch64_classify_index): Likewise.
17609 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17610 Alan Hayward <alan.hayward@arm.com>
17611 David Sherwood <david.sherwood@arm.com>
17613 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17614 (V_INT_EQUIV): ...this.
17615 (v_cmp_result): Rename to...
17616 (v_int_equiv): ...this.
17617 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17618 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17619 (copysign<mode>3): Likewise.
17620 (aarch64_simd_bsl<mode>_internal): Likewise.
17621 (aarch64_simd_bsl<mode>): Likewise.
17622 (vec_cmp<mode><mode>): Likewise.
17623 (vcond<mode><mode>): Likewise.
17624 (vcond<v_cmp_mixed><mode>): Likewise.
17625 (vcondu<mode><v_cmp_mixed>): Likewise.
17626 (aarch64_cm<optab><mode>): Likewise.
17627 (aarch64_cmtst<mode>): Likewise.
17628 (aarch64_fac<optab><mode>): Likewise.
17629 (vec_perm_const<mode>): Likewise.
17630 (vcond_mask_<mode><v_cmp_result>): Rename to...
17631 (vcond_mask_<mode><v_int_equiv>): ...this.
17632 (vec_cmp<mode><v_cmp_result>): Rename to...
17633 (vec_cmp<mode><v_int_equiv>): ...this.
17635 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17636 Alan Hayward <alan.hayward@arm.com>
17637 David Sherwood <david.sherwood@arm.com>
17639 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17641 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17642 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17643 (UNSPEC_LD4_DREG): New unspecs.
17644 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17645 (aarch64_ld2<mode>_dreg_be): Replace with...
17646 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17648 (aarch64_ld3<mode>_dreg_le)
17649 (aarch64_ld3<mode>_dreg_be): Replace with...
17650 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17652 (aarch64_ld4<mode>_dreg_le)
17653 (aarch64_ld4<mode>_dreg_be): Replace with...
17654 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17657 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17659 PR tree-optimization/81987
17660 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17661 insert an initializer in a location not dominated by the stride
17664 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17666 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17667 on the entire header of the finally block in the fallthru case.
17669 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17671 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17673 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17675 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17676 rs6000_emit_move_from_cr and call renamed function.
17677 (rs6000_emit_prologue): Call renamed functions.
17678 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17679 movesi_from_cr, remove volatile CRs.
17681 2017-08-30 Jon Beniston <jon@beniston.com>
17682 Richard Biener <rguenther@suse.de>
17684 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17685 of VECTOR_MODE_P check.
17686 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17687 element vector types.
17689 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17691 * df.h (df_read_modify_subreg_p): Remove in favor of...
17692 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17693 const_rtx instead of an rtx.
17694 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17695 * df-problems.c (df_word_lr_mark_ref): Likewise.
17696 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17697 (mark_pseudo_reg_dead): Likewise.
17698 (mark_ref_dead): Likewise.
17699 * reginfo.c (init_subregs_of_mode): Likewise.
17700 * sched-deps.c (sched_analyze_1): Likewise.
17701 * df-scan.c (df_def_record_1): Likewise.
17702 (df_uses_record): Likewise.
17703 (df_read_modify_subreg_p): Remove in favor of...
17704 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17705 const_rtx instead of an rtx.
17707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17708 Alan Hayward <alan.hayward@arm.com>
17709 David Sherwood <david.sherwood@arm.com>
17711 * rtl.h (partial_subreg_p): New function.
17712 * caller-save.c (save_call_clobbered_regs): Use it.
17713 * calls.c (expand_call): Likewise.
17714 * combine.c (combinable_i3pat): Likewise.
17715 (simplify_set): Likewise.
17716 (make_extraction): Likewise.
17717 (make_compound_operation_int): Likewise.
17718 (gen_lowpart_or_truncate): Likewise.
17719 (force_to_mode): Likewise.
17720 (make_field_assignment): Likewise.
17721 (reg_truncated_to_mode): Likewise.
17722 (record_truncated_value): Likewise.
17723 (move_deaths): Likewise.
17724 * cse.c (record_jump_cond): Likewise.
17725 (cse_insn): Likewise.
17726 * cselib.c (cselib_lookup_1): Likewise.
17727 * expmed.c (extract_bit_field_using_extv): Likewise.
17728 * function.c (assign_parm_setup_reg): Likewise.
17729 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17730 * ira-build.c (create_insn_allocnos): Likewise.
17731 * lra-coalesce.c (merge_pseudos): Likewise.
17732 * lra-constraints.c (match_reload): Likewise.
17733 (simplify_operand_subreg): Likewise.
17734 (curr_insn_transform): Likewise.
17735 * lra-lives.c (process_bb_lives): Likewise.
17736 * lra.c (new_insn_reg): Likewise.
17737 (lra_substitute_pseudo): Likewise.
17738 * regcprop.c (mode_change_ok): Likewise.
17739 (maybe_mode_change): Likewise.
17740 (copyprop_hardreg_forward_1): Likewise.
17741 * reload.c (push_reload): Likewise.
17742 (find_reloads): Likewise.
17743 (find_reloads_subreg_address): Likewise.
17744 * reload1.c (alter_reg): Likewise.
17745 (eliminate_regs_1): Likewise.
17746 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17748 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17750 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17751 back to if statements, including unpack.
17753 2017-08-30 Martin Liska <mliska@suse.cz>
17755 PR inline-asm/82001
17756 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17758 (func_checker::compare_asm_inputs_outputs): ... this function.
17759 (func_checker::compare_gimple_asm): Use the function to compare
17760 also ASM constrains.
17761 * ipa-icf-gimple.h: Rename the function.
17763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17764 Alan Hayward <alan.hayward@arm.com>
17765 David Sherwood <david.sherwood@arm.com>
17767 * coretypes.h (complex_mode): New type.
17768 * gdbhooks.py (build_pretty_printer): Handle it.
17769 * machmode.h (complex_mode): New class.
17770 (complex_mode::includes_p): New function.
17771 (is_complex_int_mode): Likewise.
17772 (is_complex_float_mode): Likewise.
17773 * genmodes.c (get_mode_class): Handle complex mode classes.
17774 * function.c (expand_function_end): Use is_complex_int_mode.
17776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17777 Alan Hayward <alan.hayward@arm.com>
17778 David Sherwood <david.sherwood@arm.com>
17780 * coretypes.h (scalar_mode_pod): New typedef.
17781 * gdbhooks.py (build_pretty_printer): Handle it.
17782 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17783 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17784 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17785 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17786 as_a <scalar_mode>.
17788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17789 Alan Hayward <alan.hayward@arm.com>
17790 David Sherwood <david.sherwood@arm.com>
17792 * machmode.h (mode_for_vector): Take a scalar_mode instead
17794 * stor-layout.c (mode_for_vector): Likewise.
17795 * explow.c (promote_mode): Use as_a <scalar_mode>.
17796 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17799 Alan Hayward <alan.hayward@arm.com>
17800 David Sherwood <david.sherwood@arm.com>
17802 * target.def (preferred_simd_mode): Take a scalar_mode
17803 instead of a machine_mode.
17804 * targhooks.h (default_preferred_simd_mode): Likewise.
17805 * targhooks.c (default_preferred_simd_mode): Likewise.
17806 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17807 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17808 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17809 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17810 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17811 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17812 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17813 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17815 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17816 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17817 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17818 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17819 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17820 * doc/tm.texi: Regenerate.
17821 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17825 Alan Hayward <alan.hayward@arm.com>
17826 David Sherwood <david.sherwood@arm.com>
17828 * target.def (scalar_mode_supported_p): Take a scalar_mode
17829 instead of a machine_mode.
17830 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17831 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17832 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17833 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17834 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17835 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17836 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17837 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17838 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17839 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17840 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17841 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17842 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17843 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17845 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17846 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17847 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17848 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17849 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17851 * doc/tm.texi: Regenerate.
17853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17854 Alan Hayward <alan.hayward@arm.com>
17855 David Sherwood <david.sherwood@arm.com>
17857 * coretypes.h (opt_scalar_mode): New typedef.
17858 * gdbhooks.py (build_pretty_printers): Handle it.
17859 * machmode.h (mode_iterator::get_2xwider): Add overload for
17861 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17863 * expr.c (convert_mode_scalar): Likewise.
17864 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17865 * optabs.c (expand_float): Likewise.
17866 (expand_fix): Likewise.
17867 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17870 Alan Hayward <alan.hayward@arm.com>
17871 David Sherwood <david.sherwood@arm.com>
17873 * optabs.c (expand_float): Explicitly check for scalars before
17874 using a branching expansion.
17875 (expand_fix): Likewise.
17877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17878 Alan Hayward <alan.hayward@arm.com>
17879 David Sherwood <david.sherwood@arm.com>
17881 * expr.c (convert_mode): Split scalar handling out into...
17882 (convert_mode_scalar): ...this new function. Treat the modes
17885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17886 Alan Hayward <alan.hayward@arm.com>
17887 David Sherwood <david.sherwood@arm.com>
17889 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17891 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17894 Alan Hayward <alan.hayward@arm.com>
17895 David Sherwood <david.sherwood@arm.com>
17897 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17898 rather than a machine_mode.
17899 (fixed_from_string): Likewise.
17900 (fixed_convert): Likewise.
17901 (fixed_convert_from_int): Likewise.
17902 (fixed_convert_from_real): Likewise.
17903 (real_convert_from_fixed): Likewise.
17904 * fixed-value.c (fixed_from_double_int): Likewise.
17905 (fixed_from_string): Likewise.
17906 (fixed_convert): Likewise.
17907 (fixed_convert_from_int): Likewise.
17908 (fixed_convert_from_real): Likewise.
17909 (real_convert_from_fixed): Likewise.
17910 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17913 Alan Hayward <alan.hayward@arm.com>
17914 David Sherwood <david.sherwood@arm.com>
17916 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17917 of separate mode class checks. Do not allow vector modes here.
17918 (immed_wide_int_const): Use as_a <scalar_mode>.
17919 * explow.c (trunc_int_for_mode): Likewise.
17920 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17921 (wi::shwi): Likewise.
17922 (wi::min_value): Likewise.
17923 (wi::max_value): Likewise.
17924 * dwarf2out.c (loc_descriptor): Likewise.
17925 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17926 for CONST_WIDE_INT.
17928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17929 Alan Hayward <alan.hayward@arm.com>
17930 David Sherwood <david.sherwood@arm.com>
17932 * tree.h (SCALAR_TYPE_MODE): New macro.
17933 * expr.c (expand_expr_addr_expr_1): Use it.
17934 (expand_expr_real_2): Likewise.
17935 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17936 (fold_convert_const_fixed_from_int): Likewise.
17937 (fold_convert_const_fixed_from_real): Likewise.
17938 (native_encode_fixed): Likewise
17939 (native_encode_complex): Likewise
17940 (native_encode_vector): Likewise.
17941 (native_interpret_fixed): Likewise.
17942 (native_interpret_real): Likewise.
17943 (native_interpret_complex): Likewise.
17944 (native_interpret_vector): Likewise.
17945 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17946 (simd_clone_adjust_argument_types): Likewise.
17947 (simd_clone_init_simd_arrays): Likewise.
17948 (simd_clone_adjust): Likewise.
17949 * stor-layout.c (layout_type): Likewise.
17950 * tree.c (build_minus_one_cst): Likewise.
17951 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17952 * tree-inline.c (estimate_move_cost): Likewise.
17953 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17954 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17955 (vectorizable_reduction): Likewise.
17956 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17957 (vect_recog_mixed_size_cond_pattern): Likewise.
17958 (check_bool_pattern): Likewise.
17959 (adjust_bool_pattern): Likewise.
17960 (search_type_for_mask_1): Likewise.
17961 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17962 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17963 (vectorizable_load): Likewise.
17964 (vectorizable_store): Likewise.
17965 * ubsan.c (ubsan_encode_value): Likewise.
17966 * varasm.c (output_constant): Likewise.
17968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17969 Alan Hayward <alan.hayward@arm.com>
17970 David Sherwood <david.sherwood@arm.com>
17972 * coretypes.h (scalar_mode): New class.
17973 * machmode.h (scalar_mode): Likewise.
17974 (scalar_mode::includes_p): New function.
17975 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17976 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17977 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17978 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17979 * expmed.c (store_bit_field_1): Likewise.
17980 (extract_bit_field_1): Likewise.
17981 * expr.c (write_complex_part): Likewise.
17982 (read_complex_part): Likewise.
17983 (emit_move_complex_push): Likewise.
17984 (expand_expr_real_2): Likewise.
17985 * function.c (assign_parm_setup_reg): Likewise.
17986 (assign_parms_unsplit_complex): Likewise.
17987 * optabs.c (expand_binop): Likewise.
17988 * rtlanal.c (subreg_get_info): Likewise.
17989 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17990 * varasm.c (output_constant_pool_2): Likewise.
17992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17993 Alan Hayward <alan.hayward@arm.com>
17994 David Sherwood <david.sherwood@arm.com>
17996 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17998 (expmed_mult_highpart_optab): Likewise.
17999 (expmed_mult_highpart): 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 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18006 instead of a machine_mode.
18007 (builtin_memset_read_str): Likewise.
18008 * builtins.c (c_readstr): Likewise.
18009 (builtin_memcpy_read_str): Likewise.
18010 (builtin_strncpy_read_str): Likewise.
18011 (builtin_memset_read_str): Likewise.
18012 (builtin_memset_gen_str): Likewise.
18013 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18014 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18015 instead of a machine_mode.
18016 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18018 (make_extraction): Likewise.
18019 (try_widen_shift_mode): Take and return scalar_int_modes instead
18021 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18022 a scalar_int_mode instead of a machine_mode.
18023 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18024 (avr_addr_space_pointer_mode): Likewise.
18025 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18026 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18027 (msp430_unwind_word_mode): Likewise.
18028 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18029 (spu_addr_space_pointer_mode): Likewise.
18030 (spu_addr_space_address_mode): Likewise.
18031 (spu_libgcc_cmp_return_mode): Likewise.
18032 (spu_libgcc_shift_count_mode): Likewise.
18033 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18034 (rl78_addr_space_pointer_mode): Likewise.
18035 (fl78_unwind_word_mode): Likewise.
18036 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18038 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18039 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18040 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18041 (mips_valid_pointer_mode): Likewise.
18042 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18043 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18044 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18046 (ft32_addr_space_address_mode): Likewise.
18047 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18048 scalar_int_mode instead of a machine_mode.
18049 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18051 (m32c_addr_space_address_mode): Likewise.
18052 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18053 (rs6000_eh_return_filter_mode): Likewise.
18054 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18055 (rs6000_eh_return_filter_mode): Likewise.
18056 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18057 (s390_libgcc_shift_count_mode): Likewise.
18058 (s390_unwind_word_mode): Likewise.
18059 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18061 * target.def (mode_rep_extended): Likewise.
18062 (valid_pointer_mode): Likewise.
18063 (addr_space.valid_pointer_mode): Likewise.
18064 (eh_return_filter_mode): Return a scalar_int_mode rather than
18066 (libgcc_cmp_return_mode): Likewise.
18067 (libgcc_shift_count_mode): Likewise.
18068 (unwind_word_mode): Likewise.
18069 (addr_space.pointer_mode): Likewise.
18070 (addr_space.address_mode): Likewise.
18071 * doc/tm.texi: Regenerate.
18072 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18074 (do_jump): Use scalar_int_mode for local variables.
18075 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18076 rather than a machine_mode.
18077 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18078 (scompare_loc_descriptor_wide): Likewise.
18079 (scompare_loc_descriptor_narrow): Likewise.
18080 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18082 * except.c (sjlj_emit_dispatch_table): Likewise.
18083 (expand_builtin_eh_copy_values): Likewise.
18084 * explow.c (convert_memory_address_addr_space_1): Likewise.
18085 Take a scalar_int_mode rather than a machine_mode.
18086 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18087 than a machine_mode.
18088 (memory_address_addr_space): Use scalar_int_mode for local variables.
18089 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18090 rather than a machine_mode.
18091 * expmed.c (mask_rtx): Likewise.
18092 (init_expmed_one_conv): Likewise.
18093 (expand_mult_highpart_adjust): Likewise.
18094 (extract_high_half): Likewise.
18095 (expmed_mult_highpart_optab): Likewise.
18096 (expmed_mult_highpart): Likewise.
18097 (expand_smod_pow2): Likewise.
18098 (expand_sdiv_pow2): Likewise.
18099 (emit_store_flag_int): Likewise.
18100 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18102 (extract_low_bits): Likewise.
18103 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18105 * expr.c (pieces_addr::adjust): Likewise.
18106 (can_store_by_pieces): Likewise.
18107 (store_by_pieces): Likewise.
18108 (clear_by_pieces_1): Likewise.
18109 (expand_expr_addr_expr_1): Likewise.
18110 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18111 (expand_expr_real_1): Likewise.
18112 (try_casesi): Likewise.
18113 * final.c (shorten_branches): Likewise.
18114 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18115 type of "mode" to machine_mode.
18116 * internal-fn.c (expand_arith_overflow_result_store): Take a
18117 scalar_int_mode rather than a machine_mode.
18118 (expand_mul_overflow): Use scalar_int_mode for local variables.
18119 * loop-doloop.c (doloop_modify): Likewise.
18120 (doloop_optimize): Likewise.
18121 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18122 than a machine_mode.
18123 (expand_doubleword_shift_condmove): Likewise.
18124 (expand_doubleword_shift): Likewise.
18125 (expand_doubleword_clz): Likewise.
18126 (expand_doubleword_popcount): Likewise.
18127 (expand_doubleword_parity): Likewise.
18128 (expand_absneg_bit): Use scalar_int_mode for local variables.
18129 (prepare_float_lib_cmp): Likewise.
18130 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18131 rather than a machine_mode.
18132 (convert_memory_address_addr_space): Likewise.
18133 (get_mode_bounds): Likewise.
18134 (get_address_mode): Return a scalar_int_mode rather than a
18136 * rtlanal.c (get_address_mode): Likewise.
18137 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18138 than a machine_mode.
18139 * targhooks.c (default_mode_rep_extended): Likewise.
18140 (default_valid_pointer_mode): Likewise.
18141 (default_addr_space_valid_pointer_mode): Likewise.
18142 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18143 than a machine_mode.
18144 (default_libgcc_cmp_return_mode): Likewise.
18145 (default_libgcc_shift_count_mode): Likewise.
18146 (default_unwind_word_mode): Likewise.
18147 (default_addr_space_pointer_mode): Likewise.
18148 (default_addr_space_address_mode): Likewise.
18149 * targhooks.h (default_eh_return_filter_mode): Likewise.
18150 (default_libgcc_cmp_return_mode): Likewise.
18151 (default_libgcc_shift_count_mode): Likewise.
18152 (default_unwind_word_mode): Likewise.
18153 (default_addr_space_pointer_mode): Likewise.
18154 (default_addr_space_address_mode): Likewise.
18155 (default_mode_rep_extended): Take a scalar_int_mode rather than
18157 (default_valid_pointer_mode): Likewise.
18158 (default_addr_space_valid_pointer_mode): Likewise.
18159 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18161 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18162 rather than a machine_mode.
18163 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18164 for local variables.
18165 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18166 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18167 than a machine_mode.
18169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18170 Alan Hayward <alan.hayward@arm.com>
18171 David Sherwood <david.sherwood@arm.com>
18173 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18174 the mode argument to scalar_int_mode.
18175 (do_jump_by_parts_zero_rtx): Likewise.
18176 (do_jump_by_parts_equality_rtx): Likewise.
18177 (do_jump_by_parts_greater): Take a mode argument.
18178 (do_jump_by_parts_equality): Likewise.
18179 (do_jump_1): Update calls accordingly.
18181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18182 Alan Hayward <alan.hayward@arm.com>
18183 David Sherwood <david.sherwood@arm.com>
18185 * is-a.h (safe_dyn_cast): New function.
18186 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18187 (jump_table_for_label): Likewise.
18188 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18189 instead of an rtx_insn *.
18190 (shorten_branches): Use dyn_cast instead of LABEL_P and
18191 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18192 rtx_jump_table_data::get_data_mode.
18193 (final_scan_insn): Likewise.
18195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18196 Alan Hayward <alan.hayward@arm.com>
18197 David Sherwood <david.sherwood@arm.com>
18199 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18200 trying to combine a full-register integer set with a subreg
18203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18204 Alan Hayward <alan.hayward@arm.com>
18205 David Sherwood <david.sherwood@arm.com>
18207 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18208 that is always either address_mode or pointer_mode.
18210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18211 Alan Hayward <alan.hayward@arm.com>
18212 David Sherwood <david.sherwood@arm.com>
18214 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18215 when the two are known to be equal.
18217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18218 Alan Hayward <alan.hayward@arm.com>
18219 David Sherwood <david.sherwood@arm.com>
18221 * simplify-rtx.c (simplify_const_unary_operation): Use
18222 is_a <scalar_int_mode> instead of checking for a nonzero
18223 precision. Forcibly convert op_mode to a scalar_int_mode
18224 in that case. More clearly differentiate the operand and
18225 result modes and use the former when deciding what the value
18226 of a count-bits operation should be. Use is_int_mode instead
18227 of checking for a MODE_INT. Remove redundant check for whether
18228 this mode has a zero precision.
18230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18231 Alan Hayward <alan.hayward@arm.com>
18232 David Sherwood <david.sherwood@arm.com>
18234 * optabs.c (widen_leading): Change the type of the mode argument
18235 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18236 (widen_bswap): Likewise.
18237 (expand_parity): Likewise.
18238 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18239 (expand_ffs): Likewise.
18240 (epand_unop): Check for scalar integer modes before calling the
18243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18244 Alan Hayward <alan.hayward@arm.com>
18245 David Sherwood <david.sherwood@arm.com>
18247 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18249 (expand_expr_real_1): Update call accordingly.
18251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18252 Alan Hayward <alan.hayward@arm.com>
18253 David Sherwood <david.sherwood@arm.com>
18255 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18256 value_mode arguments. Use scalar_int_mode internally.
18257 (store_bit_field_1): Rename the new integer mode from imode
18258 to op0_mode and use it instead of GET_MODE (op0). Update calls
18259 to store_split_bit_field, store_bit_field_using_insv and
18260 store_fixed_bit_field.
18261 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18262 Use scalar_int_mode internally. Use a bit count rather than a mode
18263 when calculating the largest bit size for get_best_mode.
18264 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18265 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18266 Remove assertion that OP0 has a scalar integer mode.
18267 (store_split_bit_field): Add op0_mode and value_mode arguments.
18268 Update calls to extract_fixed_bit_field.
18269 (extract_bit_field_using_extv): Add an op0_mode argument.
18270 Use scalar_int_mode internally.
18271 (extract_bit_field_1): Rename the new integer mode from imode to
18272 op0_mode and use it instead of GET_MODE (op0). Update calls to
18273 extract_split_bit_field, extract_bit_field_using_extv and
18274 extract_fixed_bit_field.
18275 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18276 to extract_split_bit_field and extract_fixed_bit_field_1.
18277 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18278 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18279 on the target mode.
18280 (extract_split_bit_field): Add an op0_mode argument. Update call
18281 to extract_fixed_bit_field.
18283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18284 Alan Hayward <alan.hayward@arm.com>
18285 David Sherwood <david.sherwood@arm.com>
18287 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18288 * explow.c (hard_function_value): Likewise.
18289 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18290 convert_to_mode call outside the loop.
18291 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18292 for the mode iterator. Require the mode specified by max_pieces
18294 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18296 (copy_blkmode_to_reg): Likewise.
18297 (set_storage_via_setmem): Likewise.
18298 * optabs.c (prepare_cmp_insn): Likewise.
18299 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18300 * stor-layout.c (finish_bitfield_representative): Likewise.
18302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18303 Alan Hayward <alan.hayward@arm.com>
18304 David Sherwood <david.sherwood@arm.com>
18306 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18307 * expr.c (convert_move): Use them.
18308 (convert_modes): Likewise.
18309 (store_expr_with_bounds): 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 (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18316 parameter for the mode of "x". Remove the "known_x", "known_mode"
18317 and "known_ret" arguments. Change the type of the mode argument
18318 to scalar_int_mode.
18319 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18320 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18321 (reg_num_sign_bit_copies_for_combine): Likewise.
18322 * rtlanal.c (nonzero_bits1): Likewise.
18323 (num_sign_bit_copies1): Likewise.
18324 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18325 (reg_num_sign_bit_copies_general): Likewise.
18326 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18327 (reg_nonzero_bits_general): Likewise.
18329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18330 Alan Hayward <alan.hayward@arm.com>
18331 David Sherwood <david.sherwood@arm.com>
18333 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18334 than in subroutines. Return 1 for non-integer modes.
18335 (cached_num_sign_bit_copies): Change the type of the mode parameter
18336 to scalar_int_mode.
18337 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18338 classes. Handle CONST_INT_P first and then check whether X also
18339 has a scalar integer mode. Check the same thing for inner registers
18340 of a SUBREG and for values that are being extended or truncated.
18342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18343 Alan Hayward <alan.hayward@arm.com>
18344 David Sherwood <david.sherwood@arm.com>
18346 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18347 in subroutines. Return the mode mask for non-integer modes.
18348 (cached_nonzero_bits): Change the type of the mode parameter
18349 to scalar_int_mode.
18350 (nonzero_bits1): Likewise. Remove early exit for other mode
18351 classes. Handle CONST_INT_P first and then check whether X
18352 also has a scalar integer mode.
18354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18355 Alan Hayward <alan.hayward@arm.com>
18356 David Sherwood <david.sherwood@arm.com>
18358 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18359 Return a scalar_int_mode and assert that the size is greater than
18361 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18362 (op_by_pieces_d::op_by_pieces_d): Likewise.
18363 (op_by_pieces_d::run): Likewise.
18364 (can_store_by_pieces): Likewise.
18366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18367 Alan Hayward <alan.hayward@arm.com>
18368 David Sherwood <david.sherwood@arm.com>
18370 * combine.c (extract_left_shift): Add a mode argument and update
18372 (make_compound_operation_int): Change the type of the mode parameter
18373 to scalar_int_mode and update the call to extract_left_shift.
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 (simplify_and_const_int): Change the type of the mode
18380 parameter to scalar_int_mode.
18381 (simplify_and_const_int_1): Likewise. Update recursive call.
18383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18384 Alan Hayward <alan.hayward@arm.com>
18385 David Sherwood <david.sherwood@arm.com>
18387 * combine.c (simplify_compare_const): Check that the mode is a
18388 scalar_int_mode (rather than VOIDmode) before testing its
18390 (simplify_comparison): Move COMPARISON_P handling out of the
18391 loop and restrict the latter part of the loop to scalar_int_modes.
18392 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18393 and when considering SUBREG_REGs. Use is_int_mode instead of
18394 checking GET_MODE_CLASS against MODE_INT.
18396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18397 Alan Hayward <alan.hayward@arm.com>
18398 David Sherwood <david.sherwood@arm.com>
18400 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18401 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18402 shift_unit_mode and for modes involved in scalar shifts.
18404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18405 Alan Hayward <alan.hayward@arm.com>
18406 David Sherwood <david.sherwood@arm.com>
18408 * combine.c (force_int_to_mode): New function, split out from...
18409 (force_to_mode): ...here. Keep xmode up-to-date and use it
18410 instead of GET_MODE (x).
18412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18413 Alan Hayward <alan.hayward@arm.com>
18414 David Sherwood <david.sherwood@arm.com>
18416 * optabs-query.h (extraction_insn::struct_mode): Change type to
18417 opt_scalar_int_mode and update comment.
18418 (extraction_insn::field_mode): Change type to scalar_int_mode.
18419 (extraction_insn::pos_mode): Likewise.
18420 * combine.c (make_extraction): Update accordingly.
18421 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18422 (get_optab_extraction_insn): Likewise.
18423 * recog.c (simplify_while_replacing): Likewise.
18424 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18425 parameter to opt_scalar_int_mode.
18427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18428 Alan Hayward <alan.hayward@arm.com>
18429 David Sherwood <david.sherwood@arm.com>
18431 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18432 to a scalar_int_mode instead of a machine_mode.
18433 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18434 (get_best_mode): Return a boolean and use a pointer argument to store
18435 the selected mode. Replace the limit mode parameter with a bit limit.
18436 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18437 for the values returned by bit_field_mode_iterator::next_mode.
18438 (store_bit_field): Update call to get_best_mode.
18439 (store_fixed_bit_field): Likewise.
18440 (extract_fixed_bit_field): Likewise.
18441 * expr.c (optimize_bitfield_assignment_op): Likewise.
18442 * fold-const.c (optimize_bit_field_compare): Likewise.
18443 (fold_truth_andor_1): Likewise.
18444 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18445 Update for new type of m_mode.
18446 (get_best_mode): As above.
18448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18449 Alan Hayward <alan.hayward@arm.com>
18450 David Sherwood <david.sherwood@arm.com>
18452 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18453 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18454 (store_bit_field): Check is_a <scalar_int_mode> before calling
18455 strict_volatile_bitfield_p.
18456 (extract_bit_field): Likewise.
18458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18459 Alan Hayward <alan.hayward@arm.com>
18460 David Sherwood <david.sherwood@arm.com>
18462 * target.def (cstore_mode): Return a scalar_int_mode.
18463 * doc/tm.texi: Regenerate.
18464 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18465 * targhooks.h (default_cstore_mode): Likewise.
18466 * targhooks.c (default_cstore_mode): Likewise, using a forced
18468 * expmed.c (emit_cstore): Expect the target of the cstore to be
18471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18472 Alan Hayward <alan.hayward@arm.com>
18473 David Sherwood <david.sherwood@arm.com>
18475 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18477 (niter_desc): Likewise mode.
18478 (iv_analyze): Add a mode parameter.
18480 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18481 and change its type to scalar_int_mode.
18482 * loop-iv.c: Update commentary at head of file.
18483 (iv_constant): Pass the mode paraeter before the rtx it describes
18484 and change its type to scalar_int_mode. Remove VOIDmode handling.
18485 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18486 (iv_extend): Likewise.
18487 (shorten_into_mode): Likewise.
18488 (iv_add): Use scalar_int_mode.
18489 (iv_mult): Likewise.
18490 (iv_shift): Likewise.
18491 (canonicalize_iv_subregs): Likewise.
18492 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18493 it describes and change its mode to scalar_int_mode. Also change
18494 the type of the returned inner_mode to scalar_int_mode.
18495 (get_biv_step): Likewise, turning outer_mode from a pointer
18496 into a direct parameter. Update call to get_biv_step_1.
18497 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18498 iv_constant and get_biv_step.
18499 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18500 and change its type to scalar_int_mode. Don't initialise iv->mode
18501 to VOIDmode and remove later checks for its still being VOIDmode.
18502 Update calls to iv_analyze_op and iv_analyze_expr. Check
18503 is_a <scalar_int_mode> when changing the mode under consideration.
18504 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18505 Update call to iv_analyze_expr.
18506 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18507 inner register is not also a scalar_int_mode. Update call to
18509 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18510 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18511 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18512 separate mode class checks. Update calls to iv_analyze. Remove
18513 fix-up of VOIDmodes after iv_analyze_biv.
18514 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18515 don't have a scalar_int_mode. Update call to biv_p.
18517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18518 Alan Hayward <alan.hayward@arm.com>
18519 David Sherwood <david.sherwood@arm.com>
18521 * cfgexpand.c (convert_debug_memory_address): Use
18522 as_a <scalar_int_mode>.
18523 * combine.c (expand_compound_operation): Likewise.
18524 (make_extraction): Likewise.
18525 (change_zero_ext): Likewise.
18526 (simplify_comparison): Likewise.
18527 * cse.c (cse_insn): Likewise.
18528 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18529 (mem_loc_descriptor): Likewise.
18530 (loc_descriptor): Likewise.
18531 * expmed.c (init_expmed_one_mode): Likewise.
18532 (synth_mult): Likewise.
18533 (emit_store_flag_1): Likewise.
18534 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18535 of a comparison with size.
18536 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18537 (reduce_to_bit_field_precision): Likewise.
18538 * function.c (expand_function_end): Likewise.
18539 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18540 * loop-doloop.c (doloop_modify): Likewise.
18541 * optabs.c (expand_binop): Likewise.
18542 (expand_unop): Likewise.
18543 (expand_copysign_absneg): Likewise.
18544 (prepare_cmp_insn): Likewise.
18545 (maybe_legitimize_operand): Likewise.
18546 * recog.c (const_scalar_int_operand): Likewise.
18547 * rtlanal.c (get_address_mode): Likewise.
18548 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18549 (simplify_cond_clz_ctz): Likewise.
18550 * tree-nested.c (get_nl_goto_field): Likewise.
18551 * tree.c (build_vector_type_for_mode): Likewise.
18552 * var-tracking.c (use_narrower_mode): Likewise.
18554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18555 Alan Hayward <alan.hayward@arm.com>
18556 David Sherwood <david.sherwood@arm.com>
18558 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18559 * builtins.c (expand_builtin_signbit): Use it.
18560 * cfgexpand.c (expand_debug_expr): Likewise.
18561 * dojump.c (do_jump): Likewise.
18562 (do_compare_and_jump): Likewise.
18563 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18564 * expmed.c (make_tree): Likewise.
18565 * expr.c (expand_expr_real_2): Likewise.
18566 (expand_expr_real_1): Likewise.
18567 (try_casesi): Likewise.
18568 * fold-const-call.c (fold_const_call_ss): Likewise.
18569 * fold-const.c (unextend): Likewise.
18570 (extract_muldiv_1): Likewise.
18571 (fold_single_bit_test): Likewise.
18572 (native_encode_int): Likewise.
18573 (native_encode_string): Likewise.
18574 (native_interpret_int): Likewise.
18575 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18576 * internal-fn.c (expand_addsub_overflow): Likewise.
18577 (expand_neg_overflow): Likewise.
18578 (expand_mul_overflow): Likewise.
18579 (expand_arith_overflow): Likewise.
18580 * match.pd: Likewise.
18581 * stor-layout.c (layout_type): Likewise.
18582 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18583 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18584 * tree-ssanames.c (get_range_info): Likewise.
18585 * tree-switch-conversion.c (array_value_type) Likewise.
18586 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18587 (vect_recog_divmod_pattern): Likewise.
18588 (vect_recog_mixed_size_cond_pattern): Likewise.
18589 * tree-vrp.c (extract_range_basic): Likewise.
18590 (simplify_float_conversion_using_ranges): Likewise.
18591 * tree.c (int_fits_type_p): Likewise.
18592 * ubsan.c (instrument_bool_enum_load): Likewise.
18593 * varasm.c (mergeable_string_section): Likewise.
18594 (narrowing_initializer_constant_valid_p): Likewise.
18595 (output_constant): Likewise.
18597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18598 Alan Hayward <alan.hayward@arm.com>
18599 David Sherwood <david.sherwood@arm.com>
18601 * machmode.h (NARROWEST_INT_MODE): New macro.
18602 * expr.c (alignment_for_piecewise_move): Use it instead of
18603 GET_CLASS_NARROWEST_MODE (MODE_INT).
18604 (push_block): Likewise.
18605 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18607 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18610 Alan Hayward <alan.hayward@arm.com>
18611 David Sherwood <david.sherwood@arm.com>
18613 * postreload.c (move2add_valid_value_p): Change the type of the
18614 mode parameter to scalar_int_mode.
18615 (move2add_use_add2_insn): Add a mode parameter and use it instead
18617 (move2add_use_add3_insn): Likewise.
18618 (reload_cse_move2add): Update accordingly.
18620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18621 Alan Hayward <alan.hayward@arm.com>
18622 David Sherwood <david.sherwood@arm.com>
18624 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18626 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18627 * optabs.c (expand_unop): Likewise.
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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18634 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18635 (popcount_loc_descriptor): Likewise.
18636 (bswap_loc_descriptor): Likewise.
18637 (rotate_loc_descriptor): Likewise.
18638 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18639 calling the functions above.
18641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18642 Alan Hayward <alan.hayward@arm.com>
18643 David Sherwood <david.sherwood@arm.com>
18645 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18647 (try_combine): Likewise.
18648 (simplify_if_then_else): Likewise.
18649 * cse.c (cse_insn): Likewise.
18650 * dwarf2out.c (mem_loc_descriptor): Likewise.
18651 * emit-rtl.c (gen_lowpart_common): Likewise.
18652 * simplify-rtx.c (simplify_truncation): Likewise.
18653 (simplify_binary_operation_1): Likewise.
18654 (simplify_const_relational_operation): Likewise.
18655 (simplify_ternary_operation): Likewise.
18656 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18659 Alan Hayward <alan.hayward@arm.com>
18660 David Sherwood <david.sherwood@arm.com>
18662 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18663 * reload.c (push_reload): Likewise.
18664 (find_reloads): Likewise.
18666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18667 Alan Hayward <alan.hayward@arm.com>
18668 David Sherwood <david.sherwood@arm.com>
18670 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18671 (make_compound_operation_int): Likewise.
18672 (change_zero_ext): Likewise.
18673 * expr.c (convert_move): Likewise.
18674 (convert_modes): Likewise.
18675 * fwprop.c (forward_propagate_subreg): Likewise.
18676 * loop-iv.c (get_biv_step_1): Likewise.
18677 * optabs.c (widen_operand): Likewise.
18678 * postreload.c (move2add_valid_value_p): Likewise.
18679 * recog.c (simplify_while_replacing): Likewise.
18680 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18681 (simplify_binary_operation_1): Likewise. Remove redundant
18682 mode equality check.
18684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18685 Alan Hayward <alan.hayward@arm.com>
18686 David Sherwood <david.sherwood@arm.com>
18688 * combine.c (combine_simplify_rtx): Add checks for
18689 is_a <scalar_int_mode>.
18690 (simplify_if_then_else): Likewise.
18691 (make_field_assignment): Likewise.
18692 (simplify_comparison): Likewise.
18693 * ifcvt.c (noce_try_bitop): Likewise.
18694 * loop-invariant.c (canonicalize_address_mult): Likewise.
18695 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18698 Alan Hayward <alan.hayward@arm.com>
18699 David Sherwood <david.sherwood@arm.com>
18701 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18702 is_a <scalar_int_mode> instead of != BLKmode.
18704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18705 Alan Hayward <alan.hayward@arm.com>
18706 David Sherwood <david.sherwood@arm.com>
18708 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18709 instead of != VOIDmode.
18710 * combine.c (if_then_else_cond): Likewise.
18711 (change_zero_ext): Likewise.
18712 * dwarf2out.c (mem_loc_descriptor): Likewise.
18713 (loc_descriptor): Likewise.
18714 * rtlanal.c (canonicalize_condition): Likewise.
18715 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18718 Alan Hayward <alan.hayward@arm.com>
18719 David Sherwood <david.sherwood@arm.com>
18721 * simplify-rtx.c (simplify_binary_operation_1): Use
18722 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18725 Alan Hayward <alan.hayward@arm.com>
18726 David Sherwood <david.sherwood@arm.com>
18728 * wide-int.h (int_traits<unsigned char>) New class.
18729 (int_traits<unsigned short>) Likewise.
18730 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18731 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18733 * combine.c (set_nonzero_bits_and_sign_copies): Use
18734 is_a <scalar_int_mode>.
18735 (find_split_point): Likewise.
18736 (combine_simplify_rtx): Likewise.
18737 (simplify_logical): Likewise.
18738 (expand_compound_operation): Likewise.
18739 (expand_field_assignment): Likewise.
18740 (make_compound_operation): Likewise.
18741 (extended_count): Likewise.
18742 (change_zero_ext): Likewise.
18743 (simplify_comparison): Likewise.
18744 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18745 (ucompare_loc_descriptor): Likewise.
18746 (minmax_loc_descriptor): Likewise.
18747 (mem_loc_descriptor): Likewise.
18748 (loc_descriptor): Likewise.
18749 * expmed.c (init_expmed_one_mode): Likewise.
18750 * lra-constraints.c (lra_constraint_offset): Likewise.
18751 * optabs.c (prepare_libcall_arg): Likewise.
18752 * postreload.c (move2add_note_store): Likewise.
18753 * reload.c (operands_match_p): Likewise.
18754 * rtl.h (load_extend_op): Likewise.
18755 * rtlhooks.c (gen_lowpart_general): Likewise.
18756 * simplify-rtx.c (simplify_truncation): Likewise.
18757 (simplify_unary_operation_1): Likewise.
18758 (simplify_binary_operation_1): Likewise.
18759 (simplify_const_binary_operation): Likewise.
18760 (simplify_const_relational_operation): Likewise.
18761 (simplify_subreg): Likewise.
18762 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18763 * var-tracking.c (adjust_mems): Likewise.
18764 (prepare_call_arguments): Likewise.
18766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18767 Alan Hayward <alan.hayward@arm.com>
18768 David Sherwood <david.sherwood@arm.com>
18770 * machmode.h (is_int_mode): New fuction.
18771 * combine.c (find_split_point): Use it.
18772 (combine_simplify_rtx): Likewise.
18773 (simplify_if_then_else): Likewise.
18774 (simplify_set): Likewise.
18775 (simplify_shift_const_1): Likewise.
18776 (simplify_comparison): Likewise.
18777 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18778 * cse.c (notreg_cost): Likewise.
18779 (cse_insn): Likewise.
18780 * cselib.c (cselib_lookup_1): Likewise.
18781 * dojump.c (do_jump_1): Likewise.
18782 (do_compare_rtx_and_jump): Likewise.
18783 * dse.c (get_call_args): Likewise.
18784 * dwarf2out.c (rtl_for_decl_init): Likewise.
18785 (native_encode_initializer): Likewise.
18786 * expmed.c (emit_store_flag_1): Likewise.
18787 (emit_store_flag): Likewise.
18788 * expr.c (convert_modes): Likewise.
18789 (store_field): Likewise.
18790 (expand_expr_real_1): Likewise.
18791 * fold-const.c (fold_read_from_constant_string): Likewise.
18792 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18793 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18794 * optabs.c (expand_binop): Likewise.
18795 (expand_unop): Likewise.
18796 (expand_abs_nojump): Likewise.
18797 (expand_one_cmpl_abs_nojump): Likewise.
18798 * simplify-rtx.c (mode_signbit_p): Likewise.
18799 (val_signbit_p): Likewise.
18800 (val_signbit_known_set_p): Likewise.
18801 (val_signbit_known_clear_p): Likewise.
18802 (simplify_relational_operation_1): Likewise.
18803 * tree.c (vector_type_mode): Likewise.
18805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18806 Alan Hayward <alan.hayward@arm.com>
18807 David Sherwood <david.sherwood@arm.com>
18809 * machmode.h (smallest_mode_for_size): Fix formatting.
18810 (smallest_int_mode_for_size): New function.
18811 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18812 instead of smallest_mode_for_size.
18813 * combine.c (make_extraction): Likewise.
18814 * config/arc/arc.c (arc_expand_movmem): Likewise.
18815 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18816 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18817 * config/s390/s390.c (s390_expand_insv): Likewise.
18818 * config/sparc/sparc.c (assign_int_registers): Likewise.
18819 * config/spu/spu.c (spu_function_value): Likewise.
18820 (spu_function_arg): Likewise.
18821 * coverage.c (get_gcov_type): Likewise.
18822 (get_gcov_unsigned_t): Likewise.
18823 * dse.c (find_shift_sequence): Likewise.
18824 * expmed.c (store_bit_field_1): Likewise.
18825 * expr.c (convert_move): Likewise.
18826 (store_field): Likewise.
18827 * internal-fn.c (expand_arith_overflow): Likewise.
18828 * optabs-query.c (get_best_extraction_insn): Likewise.
18829 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18830 * stor-layout.c (layout_type): Likewise.
18831 (initialize_sizetypes): Likewise.
18832 * targhooks.c (default_get_mask_mode): Likewise.
18833 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18836 Alan Hayward <alan.hayward@arm.com>
18837 David Sherwood <david.sherwood@arm.com>
18839 * machmode.h (opt_mode::else_blk): New function.
18840 (int_mode_for_mode): Declare.
18841 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18842 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18844 * cfgexpand.c (expand_debug_expr): Likewise.
18845 * combine.c (gen_lowpart_or_truncate): Likewise.
18846 (gen_lowpart_for_combine): Likewise.
18847 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18848 * config/avr/avr.c (avr_to_int_mode): Likewise.
18849 (avr_out_plus_1): Likewise.
18850 (avr_out_plus): Likewise.
18851 (avr_out_round): Likewise.
18852 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18853 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18854 (s390_expand_vcond): Likewise.
18855 * config/spu/spu.c (spu_split_immediate): Likewise.
18856 (spu_expand_mov): Likewise.
18857 * dse.c (get_stored_val): Likewise.
18858 * expmed.c (store_bit_field_1): Likewise.
18859 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18861 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18862 (extract_low_bits): Likewise.
18863 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18864 handling rather than repeating the check.
18865 (emit_group_store): Likewise.
18866 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18867 * optabs.c (expand_absneg_bit): Likewise.
18868 (expand_copysign_absneg): Likewise.
18869 (expand_copysign_bit): Likewise.
18870 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18871 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18872 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18873 * var-tracking.c (prepare_call_arguments): Likewise.
18874 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18875 int_mode_for_mode instead of mode_for_size.
18876 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18879 Alan Hayward <alan.hayward@arm.com>
18880 David Sherwood <david.sherwood@arm.com>
18882 * machmode.h (int_mode_for_size): New function.
18883 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18884 instead of mode_for_size.
18885 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18887 * combine.c (expand_field_assignment): Use int_mode_for_size
18888 instead of mode_for_size.
18889 (make_extraction): Likewise.
18890 (simplify_shift_const_1): Likewise.
18891 (simplify_comparison): Likewise.
18892 * dojump.c (do_jump): Likewise.
18893 * dwarf2out.c (mem_loc_descriptor): Likewise.
18894 * emit-rtl.c (init_derived_machine_modes): Likewise.
18895 * expmed.c (flip_storage_order): Likewise.
18896 (convert_extracted_bit_field): Likewise.
18897 * expr.c (copy_blkmode_from_reg): Likewise.
18898 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18899 * internal-fn.c (expand_mul_overflow): Likewise.
18900 * lower-subreg.c (simple_move): Likewise.
18901 * optabs-libfuncs.c (init_optabs): Likewise.
18902 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18903 * tree.c (vector_type_mode): Likewise.
18904 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18905 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18906 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18907 * tree-vect-stmts.c (vectorizable_load): Likewise.
18908 (vectorizable_store): Likewise.
18910 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18911 Alan Hayward <alan.hayward@arm.com>
18912 David Sherwood <david.sherwood@arm.com>
18914 * coretypes.h (pod_mode): New type.
18915 (scalar_int_mode_pod): New typedef.
18916 * machmode.h (pod_mode): New class.
18917 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18918 * genmodes.c (emit_mode_int_n): Update accordingly.
18919 * lower-subreg.h (target_lower_subreg): Change type to
18920 scalar_int_mode_pod.
18921 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18922 scalar_int_mode_pod.
18924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18925 Alan Hayward <alan.hayward@arm.com>
18926 David Sherwood <david.sherwood@arm.com>
18928 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18929 machine_mode to scalar_int_mode.
18930 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18931 (rs6000_option_override_internal): Remove cast to int.
18932 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18933 machine_mode to scalar_int_mode.
18934 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18935 (rs6000_option_override_internal): Remove cast to int.
18936 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18937 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18939 * config/s390/s390.c (s390_expand_builtin): Likewise.
18940 * coretypes.h (scalar_int_mode): New type.
18941 (opt_scalar_int_mode): New typedef.
18942 * machmode.h (scalar_int_mode): New class.
18943 (scalar_int_mode::includes_p): New function.
18944 (byte_mode): Change type to scalar_int_mode.
18945 (word_mode): Likewise.
18946 (ptr_mode): Likewise.
18947 * emit-rtl.c (byte_mode): Likewise.
18948 (word_mode): Likewise.
18949 (ptr_mode): Likewise.
18950 (init_derived_machine_modes): Update accordingly.
18951 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18952 and MODE_PARTIAL_INT.
18953 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18954 opt_scalar_int_mode.
18956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18957 Alan Hayward <alan.hayward@arm.com>
18958 David Sherwood <david.sherwood@arm.com>
18960 * target.def (libgcc_floating_mode_supported_p): Take a
18962 * doc/tm.texi: Regenerate.
18963 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18965 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18966 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18970 Alan Hayward <alan.hayward@arm.com>
18971 David Sherwood <david.sherwood@arm.com>
18973 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18974 * doc/tm.texi: Regenerate.
18975 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18976 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18977 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18978 * targhooks.h (default_floatn_mode): Likewise.
18979 * targhooks.c (default_floatn_mode): Likewise.
18980 * tree.c (build_common_tree_nodes): Update accordingly.
18982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18983 Alan Hayward <alan.hayward@arm.com>
18984 David Sherwood <david.sherwood@arm.com>
18986 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18987 (mode_iterator::iterate_p): Likewise.
18988 (mode_iterator::get_wider): Likewise.
18989 * expr.c (init_expr_target): Use opt_scalar_float_mode.
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 * coretypes.h (opt_scalar_float_mode): New typedef.
18996 * machmode.h (float_mode_for_size): New function.
18997 * emit-rtl.c (double_mode): Delete.
18998 (init_emit_once): Use float_mode_for_size.
18999 * stor-layout.c (layout_type): Likewise.
19000 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19003 Alan Hayward <alan.hayward@arm.com>
19004 David Sherwood <david.sherwood@arm.com>
19006 * output.h (assemble_real): Take a scalar_float_mode.
19007 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19008 * config/arm/arm.md (consttable_4): Likewise.
19009 (consttable_8): Likewise.
19010 (consttable_16): Likewise.
19011 * config/mips/mips.md (consttable_float): Likewise.
19012 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19013 * varasm.c (assemble_real): Take a scalar_float_mode.
19014 (output_constant_pool_2): Update accordingly.
19015 (output_constant): Likewise.
19017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19018 Alan Hayward <alan.hayward@arm.com>
19019 David Sherwood <david.sherwood@arm.com>
19021 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19022 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19023 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19024 (native_encode_real): Likewise.
19025 (native_interpret_real): Likewise.
19026 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19027 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19030 Alan Hayward <alan.hayward@arm.com>
19031 David Sherwood <david.sherwood@arm.com>
19033 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19034 <scalar_float_mode>. Simplify.
19035 (gen_extend_conv_libfunc): Likewise.
19037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19038 Alan Hayward <alan.hayward@arm.com>
19039 David Sherwood <david.sherwood@arm.com>
19041 * coretypes.h (scalar_float_mode): New type.
19042 * machmode.h (mode_traits::from_int): Use machine_mode if
19043 USE_ENUM_MODES is defined.
19044 (is_a): New function.
19046 (dyn_cast): Likewise.
19047 (scalar_float_mode): New class.
19048 (scalar_float_mode::includes_p): New function.
19049 (is_float_mode): Likewise.
19050 * gdbhooks.py (MachineModePrinter): New class.
19051 (build_pretty_printer): Use it for scalar_float_mode.
19052 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19053 (format_helper::format_helper): Turn into a template.
19054 * genmodes.c (get_mode_class): New function.
19055 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19056 or machine_mode if none.
19057 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19058 as_a <scalar_float_mode>.
19059 * dwarf2out.c (mem_loc_descriptor): Likewise.
19060 (insert_float): Likewise.
19061 (add_const_value_attribute): Likewise.
19062 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19063 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19064 (expand_unop): Update accordingly.
19065 (expand_abs_nojump): Likewise.
19066 (expand_copysign_absneg): Take a scalar_float_mode.
19067 (expand_copysign_bit): Likewise.
19068 (expand_copysign): Update accordingly.
19070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19071 Alan Hayward <alan.hayward@arm.com>
19072 David Sherwood <david.sherwood@arm.com>
19074 * coretypes.h (opt_mode): New class.
19075 * machmode.h (opt_mode): Likewise.
19076 (opt_mode::else_void): New function.
19077 (opt_mode::require): Likewise.
19078 (opt_mode::exists): Likewise.
19079 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19080 (GET_MODE_2XWIDER_MODE): Likewise.
19081 (mode_iterator::get_wider): Update accordingly.
19082 (mode_iterator::get_2xwider): Likewise.
19083 (mode_iterator::get_known_wider): Likewise, turning into a template.
19084 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19085 forcing a wider mode to exist.
19086 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19087 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19088 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19089 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19090 * lower-subreg.c (init_lower_subreg): Likewise.
19091 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19092 on the final iteration.
19093 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19094 a wider mode exists before asking for a move pattern.
19095 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19096 forcing a wider mode to exist.
19097 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19098 returning false if no such mode exists.
19099 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19100 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19101 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19102 Avoid checking for a MODE_INT if we already know the mode is not a
19104 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19105 forcing a wider mode to exist.
19106 (expmed_mult_highpart_optab): Likewise.
19107 (expmed_mult_highpart): Likewise.
19108 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19110 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19111 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19112 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19113 * internal-fn.c (expand_mul_overflow): Update use of
19114 GET_MODE_2XWIDER_MODE.
19115 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19116 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19117 GET_MODE_WIDER_MODE.
19118 (convert_plusminus_to_widen): Likewise.
19119 * tree-switch-conversion.c (array_value_type): Likewise.
19120 * var-tracking.c (emit_note_insn_var_location): Likewise.
19121 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19122 Return false inside rather than outside the loop if no wider mode
19124 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19125 and GET_MODE_2XWIDER_MODE
19126 (can_compare_p): Use else_void.
19127 * gdbhooks.py (OptMachineModePrinter): New class.
19128 (build_pretty_printer): Use it for opt_mode.
19130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19131 Alan Hayward <alan.hayward@arm.com>
19132 David Sherwood <david.sherwood@arm.com>
19134 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19135 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19138 Alan Hayward <alan.hayward@arm.com>
19139 David Sherwood <david.sherwood@arm.com>
19141 * machmode.h (mode_traits): New structure.
19142 (get_narrowest_mode): New function.
19143 (mode_iterator::start): Likewise.
19144 (mode_iterator::iterate_p): Likewise.
19145 (mode_iterator::get_wider): Likewise.
19146 (mode_iterator::get_known_wider): Likewise.
19147 (mode_iterator::get_2xwider): Likewise.
19148 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19149 (FOR_EACH_MODE): Likewise.
19150 (FOR_EACH_MODE_FROM): Likewise.
19151 (FOR_EACH_MODE_UNTIL): Likewise.
19152 (FOR_EACH_WIDER_MODE): Likewise.
19153 (FOR_EACH_2XWIDER_MODE): Likewise.
19154 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19155 * combine.c (simplify_comparison): Likewise
19156 * config/i386/i386.c (type_natural_mode): Likewise.
19157 * cse.c (cse_insn): Likewise.
19158 * dse.c (find_shift_sequence): Likewise.
19159 * emit-rtl.c (init_derived_machine_modes): Likewise.
19160 (init_emit_once): Likewise.
19161 * explow.c (hard_function_value): Likewise.
19162 * expmed.c (extract_fixed_bit_field_1): Likewise.
19163 (extract_bit_field_1): Likewise.
19164 (expand_divmod): Likewise.
19165 (emit_store_flag_1): Likewise.
19166 * expr.c (init_expr_target): Likewise.
19167 (convert_move): Likewise.
19168 (alignment_for_piecewise_move): Likewise.
19169 (widest_int_mode_for_size): Likewise.
19170 (emit_block_move_via_movmem): Likewise.
19171 (copy_blkmode_to_reg): Likewise.
19172 (set_storage_via_setmem): Likewise.
19173 (compress_float_constant): Likewise.
19174 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19175 * optabs-query.c (get_best_extraction_insn): Likewise.
19176 * optabs.c (expand_binop): Likewise.
19177 (expand_twoval_unop): Likewise.
19178 (expand_twoval_binop): Likewise.
19179 (widen_leading): Likewise.
19180 (widen_bswap): Likewise.
19181 (expand_parity): Likewise.
19182 (expand_unop): Likewise.
19183 (prepare_cmp_insn): Likewise.
19184 (prepare_float_lib_cmp): Likewise.
19185 (expand_float): Likewise.
19186 (expand_fix): Likewise.
19187 (expand_sfix_optab): Likewise.
19188 * postreload.c (move2add_use_add2_insn): Likewise.
19189 * reg-stack.c (reg_to_stack): Likewise.
19190 * reginfo.c (choose_hard_reg_mode): Likewise.
19191 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19192 * stor-layout.c (mode_for_size): Likewise.
19193 (smallest_mode_for_size): Likewise.
19194 (mode_for_vector): Likewise.
19195 (finish_bitfield_representative): Likewise.
19196 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19197 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19198 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19199 * var-tracking.c (prepare_call_arguments): Likewise.
19201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19202 Alan Hayward <alan.hayward@arm.com>
19203 David Sherwood <david.sherwood@arm.com>
19205 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19206 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19207 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19208 * machmode.h (mode_size): Move earlier in file.
19209 (mode_precision): Likewise.
19210 (mode_inner): Likewise.
19211 (mode_nunits): Likewise.
19212 (mode_unit_size): Likewise.
19213 (unit_unit_precision): Likewise.
19214 (mode_wider): Likewise.
19215 (mode_2xwider): Likewise.
19216 (machine_mode): New class.
19217 (mode_to_bytes): New function.
19218 (mode_to_bits): Likewise.
19219 (mode_to_precision): Likewise.
19220 (mode_to_inner): Likewise.
19221 (mode_to_unit_size): Likewise.
19222 (mode_to_unit_precision): Likewise.
19223 (mode_to_nunits): Likewise.
19224 (GET_MODE_SIZE): Use mode_to_bytes.
19225 (GET_MODE_BITSIZE): Use mode_to_bits.
19226 (GET_MODE_PRECISION): Use mode_to_precision.
19227 (GET_MODE_INNER): Use mode_to_inner.
19228 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19229 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19230 (GET_MODE_NUNITS): Use mode_to_nunits.
19231 * system.h (ALWAYS_INLINE): New macro.
19232 * config/powerpcspe/powerpcspe-c.c
19233 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19234 int for arg1_mode and arg2_mode.
19236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19237 Alan Hayward <alan.hayward@arm.com>
19238 David Sherwood <david.sherwood@arm.com>
19240 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19241 Prefix mode names with E_ in case statements.
19242 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19243 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19244 (aarch64_split_simd_move): Likewise.
19245 (aarch64_gen_storewb_pair): Likewise.
19246 (aarch64_gen_loadwb_pair): Likewise.
19247 (aarch64_gen_store_pair): Likewise.
19248 (aarch64_gen_load_pair): Likewise.
19249 (aarch64_get_condition_code_1): Likewise.
19250 (aarch64_constant_pool_reload_icode): Likewise.
19251 (get_rsqrte_type): Likewise.
19252 (get_rsqrts_type): Likewise.
19253 (get_recpe_type): Likewise.
19254 (get_recps_type): Likewise.
19255 (aarch64_gimplify_va_arg_expr): Likewise.
19256 (aarch64_simd_container_mode): Likewise.
19257 (aarch64_emit_load_exclusive): Likewise.
19258 (aarch64_emit_store_exclusive): Likewise.
19259 (aarch64_expand_compare_and_swap): Likewise.
19260 (aarch64_gen_atomic_cas): Likewise.
19261 (aarch64_emit_bic): Likewise.
19262 (aarch64_emit_atomic_swap): Likewise.
19263 (aarch64_emit_atomic_load_op): Likewise.
19264 (aarch64_evpc_trn): Likewise.
19265 (aarch64_evpc_uzp): Likewise.
19266 (aarch64_evpc_zip): Likewise.
19267 (aarch64_evpc_ext): Likewise.
19268 (aarch64_evpc_rev): Likewise.
19269 (aarch64_evpc_dup): Likewise.
19270 (aarch64_gen_ccmp_first): Likewise.
19271 (aarch64_gen_ccmp_next): Likewise.
19272 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19273 (alpha_emit_xfloating_libcall): Likewise.
19274 (emit_insxl): Likewise.
19275 (alpha_arg_type): Likewise.
19276 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19277 (arc_preferred_simd_mode): Likewise.
19278 (arc_secondary_reload): Likewise.
19279 (get_arc_condition_code): Likewise.
19280 (arc_print_operand): Likewise.
19281 (arc_legitimate_constant_p): Likewise.
19282 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19283 * config/arc/arc.md (casesi_load): Likewise.
19284 (casesi_compact_jump): Likewise.
19285 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19286 (cc_use_register): Likewise.
19287 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19288 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19289 (arm_init_iwmmxt_builtins): Likewise.
19290 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19291 (neon_expand_vector_init): Likewise.
19292 (arm_attr_length_move_neon): Likewise.
19293 (maybe_get_arm_condition_code): Likewise.
19294 (arm_emit_vector_const): Likewise.
19295 (arm_preferred_simd_mode): Likewise.
19296 (arm_output_iwmmxt_tinsr): Likewise.
19297 (thumb1_output_casesi): Likewise.
19298 (thumb2_output_casesi): Likewise.
19299 (arm_emit_load_exclusive): Likewise.
19300 (arm_emit_store_exclusive): Likewise.
19301 (arm_expand_compare_and_swap): Likewise.
19302 (arm_evpc_neon_vuzp): Likewise.
19303 (arm_evpc_neon_vzip): Likewise.
19304 (arm_evpc_neon_vrev): Likewise.
19305 (arm_evpc_neon_vtrn): Likewise.
19306 (arm_evpc_neon_vext): Likewise.
19307 (arm_validize_comparison): Likewise.
19308 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19309 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19310 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19311 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19312 (c6x_preferred_simd_mode): Likewise.
19313 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19314 (epiphany_rtx_costs): Likewise.
19315 * config/epiphany/predicates.md (proper_comparison_operator):
19317 * config/frv/frv.c (condexec_memory_operand): Likewise.
19318 (frv_emit_move): Likewise.
19319 (output_move_single): Likewise.
19320 (output_condmove_single): Likewise.
19321 (frv_hard_regno_mode_ok): Likewise.
19322 (frv_matching_accg_mode): Likewise.
19323 * config/h8300/h8300.c (split_adds_subs): Likewise.
19324 (h8300_rtx_costs): Likewise.
19325 (h8300_print_operand): Likewise.
19326 (compute_mov_length): Likewise.
19327 (output_logical_op): Likewise.
19328 (compute_logical_op_length): Likewise.
19329 (compute_logical_op_cc): Likewise.
19330 (h8300_shift_needs_scratch_p): Likewise.
19331 (output_a_shift): Likewise.
19332 (compute_a_shift_length): Likewise.
19333 (compute_a_shift_cc): Likewise.
19334 (expand_a_rotate): Likewise.
19335 (output_a_rotate): Likewise.
19336 * config/i386/i386.c (classify_argument): Likewise.
19337 (function_arg_advance_32): Likewise.
19338 (function_arg_32): Likewise.
19339 (function_arg_64): Likewise.
19340 (function_value_64): Likewise.
19341 (ix86_gimplify_va_arg): Likewise.
19342 (ix86_legitimate_constant_p): Likewise.
19343 (put_condition_code): Likewise.
19344 (split_double_mode): Likewise.
19345 (ix86_avx256_split_vector_move_misalign): Likewise.
19346 (ix86_expand_vector_logical_operator): Likewise.
19347 (ix86_split_idivmod): Likewise.
19348 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19349 (ix86_build_const_vector): Likewise.
19350 (ix86_build_signbit_mask): Likewise.
19351 (ix86_match_ccmode): Likewise.
19352 (ix86_cc_modes_compatible): Likewise.
19353 (ix86_expand_branch): Likewise.
19354 (ix86_expand_sse_cmp): Likewise.
19355 (ix86_expand_sse_movcc): Likewise.
19356 (ix86_expand_int_sse_cmp): Likewise.
19357 (ix86_expand_vec_perm_vpermi2): Likewise.
19358 (ix86_expand_vec_perm): Likewise.
19359 (ix86_expand_sse_unpack): Likewise.
19360 (ix86_expand_int_addcc): Likewise.
19361 (ix86_split_to_parts): Likewise.
19362 (ix86_vectorize_builtin_gather): Likewise.
19363 (ix86_vectorize_builtin_scatter): Likewise.
19364 (avx_vpermilp_parallel): Likewise.
19365 (inline_memory_move_cost): Likewise.
19366 (ix86_tieable_integer_mode_p): Likewise.
19367 (x86_maybe_negate_const_int): Likewise.
19368 (ix86_expand_vector_init_duplicate): Likewise.
19369 (ix86_expand_vector_init_one_nonzero): Likewise.
19370 (ix86_expand_vector_init_one_var): Likewise.
19371 (ix86_expand_vector_init_concat): Likewise.
19372 (ix86_expand_vector_init_interleave): Likewise.
19373 (ix86_expand_vector_init_general): Likewise.
19374 (ix86_expand_vector_set): Likewise.
19375 (ix86_expand_vector_extract): Likewise.
19376 (emit_reduc_half): Likewise.
19377 (ix86_emit_i387_round): Likewise.
19378 (ix86_mangle_type): Likewise.
19379 (ix86_expand_round_sse4): Likewise.
19380 (expand_vec_perm_blend): Likewise.
19381 (canonicalize_vector_int_perm): Likewise.
19382 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19383 (expand_vec_perm_1): Likewise.
19384 (expand_vec_perm_interleave3): Likewise.
19385 (expand_vec_perm_even_odd_pack): Likewise.
19386 (expand_vec_perm_even_odd_1): Likewise.
19387 (expand_vec_perm_broadcast_1): Likewise.
19388 (ix86_vectorize_vec_perm_const_ok): Likewise.
19389 (ix86_expand_vecop_qihi): Likewise.
19390 (ix86_expand_mul_widen_hilo): Likewise.
19391 (ix86_expand_sse2_abs): Likewise.
19392 (ix86_expand_pextr): Likewise.
19393 (ix86_expand_pinsr): Likewise.
19394 (ix86_preferred_simd_mode): Likewise.
19395 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19396 * config/i386/sse.md (*andnot<mode>3): Likewise.
19397 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19398 (*<code><mode>3): Likewise.
19399 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19400 (ia64_expand_atomic_op): Likewise.
19401 (ia64_arg_type): Likewise.
19402 (ia64_mode_to_int): Likewise.
19403 (ia64_scalar_mode_supported_p): Likewise.
19404 (ia64_vector_mode_supported_p): Likewise.
19405 (expand_vec_perm_broadcast): Likewise.
19406 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19407 (iq2000_function_arg_advance): Likewise.
19408 (iq2000_function_arg): Likewise.
19409 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19410 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19411 (m68k_libcall_value): Likewise.
19412 (m68k_function_value): Likewise.
19413 (sched_attr_op_type): Likewise.
19414 * config/mcore/mcore.c (mcore_output_move): Likewise.
19415 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19417 (microblaze_function_arg): Likewise.
19418 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19419 (mips_print_operand): Likewise.
19420 (mips_mode_ok_for_mov_fmt_p): Likewise.
19421 (mips_vector_mode_supported_p): Likewise.
19422 (mips_preferred_simd_mode): Likewise.
19423 (mips_expand_vpc_loongson_even_odd): Likewise.
19424 (mips_expand_vec_unpack): Likewise.
19425 (mips_expand_vi_broadcast): Likewise.
19426 (mips_expand_vector_init): Likewise.
19427 (mips_expand_vec_reduc): Likewise.
19428 (mips_expand_msa_cmp): Likewise.
19429 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19430 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19431 (cc_flags_for_mode): Likewise.
19432 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19433 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19434 (nds32_output_casesi_pc_relative): Likewise.
19435 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19436 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19437 (nvptx_gen_unpack): Likewise.
19438 (nvptx_gen_pack): Likewise.
19439 (nvptx_gen_shuffle): Likewise.
19440 (nvptx_gen_wcast): Likewise.
19441 (nvptx_preferred_simd_mode): Likewise.
19442 * config/pa/pa.c (pa_secondary_reload): Likewise.
19443 * config/pa/predicates.md (base14_operand): Likewise.
19444 * config/powerpcspe/powerpcspe-c.c
19445 (altivec_resolve_overloaded_builtin): Likewise.
19446 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19448 (rs6000_preferred_simd_mode): Likewise.
19449 (output_vec_const_move): Likewise.
19450 (rs6000_expand_vector_extract): Likewise.
19451 (rs6000_split_vec_extract_var): Likewise.
19452 (reg_offset_addressing_ok_p): Likewise.
19453 (rs6000_legitimate_offset_address_p): Likewise.
19454 (rs6000_legitimize_address): Likewise.
19455 (rs6000_emit_set_const): Likewise.
19456 (rs6000_const_vec): Likewise.
19457 (rs6000_emit_move): Likewise.
19458 (spe_build_register_parallel): Likewise.
19459 (rs6000_darwin64_record_arg_recurse): Likewise.
19460 (swap_selector_for_mode): Likewise.
19461 (spe_init_builtins): Likewise.
19462 (paired_init_builtins): Likewise.
19463 (altivec_init_builtins): Likewise.
19464 (do_load_for_compare): Likewise.
19465 (rs6000_generate_compare): Likewise.
19466 (rs6000_expand_float128_convert): Likewise.
19467 (emit_load_locked): Likewise.
19468 (emit_store_conditional): Likewise.
19469 (rs6000_output_function_epilogue): Likewise.
19470 (rs6000_handle_altivec_attribute): Likewise.
19471 (rs6000_function_value): Likewise.
19472 (emit_fusion_gpr_load): Likewise.
19473 (emit_fusion_p9_load): Likewise.
19474 (emit_fusion_p9_store): Likewise.
19475 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19476 (fusion_gpr_mem_load): Likewise.
19477 (fusion_addis_mem_combo_load): Likewise.
19478 (fusion_addis_mem_combo_store): Likewise.
19479 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19480 (fusion_gpr_mem_load): Likewise.
19481 (fusion_addis_mem_combo_load): Likewise.
19482 (fusion_addis_mem_combo_store): Likewise.
19483 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19485 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19486 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19487 (rs6000_preferred_simd_mode): Likewise.
19488 (output_vec_const_move): Likewise.
19489 (rs6000_expand_vector_extract): Likewise.
19490 (rs6000_split_vec_extract_var): Likewise.
19491 (reg_offset_addressing_ok_p): Likewise.
19492 (rs6000_legitimate_offset_address_p): Likewise.
19493 (rs6000_legitimize_address): Likewise.
19494 (rs6000_emit_set_const): Likewise.
19495 (rs6000_const_vec): Likewise.
19496 (rs6000_emit_move): Likewise.
19497 (rs6000_darwin64_record_arg_recurse): Likewise.
19498 (swap_selector_for_mode): Likewise.
19499 (paired_init_builtins): Likewise.
19500 (altivec_init_builtins): Likewise.
19501 (rs6000_expand_float128_convert): Likewise.
19502 (emit_load_locked): Likewise.
19503 (emit_store_conditional): Likewise.
19504 (rs6000_output_function_epilogue): Likewise.
19505 (rs6000_handle_altivec_attribute): Likewise.
19506 (rs6000_function_value): Likewise.
19507 (emit_fusion_gpr_load): Likewise.
19508 (emit_fusion_p9_load): Likewise.
19509 (emit_fusion_p9_store): Likewise.
19510 * config/rx/rx.c (rx_gen_move_template): Likewise.
19511 (flags_from_mode): Likewise.
19512 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19513 (s390_slb_comparison): Likewise.
19514 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19515 (s390_vector_mode_supported_p): Likewise.
19516 (s390_cc_modes_compatible): Likewise.
19517 (s390_match_ccmode_set): Likewise.
19518 (s390_canonicalize_comparison): Likewise.
19519 (s390_emit_compare_and_swap): Likewise.
19520 (s390_branch_condition_mask): Likewise.
19521 (s390_rtx_costs): Likewise.
19522 (s390_secondary_reload): Likewise.
19523 (__SECONDARY_RELOAD_CASE): Likewise.
19524 (s390_expand_cs): Likewise.
19525 (s390_preferred_simd_mode): Likewise.
19526 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19527 * config/sh/sh.c (sh_print_operand): Likewise.
19528 (dump_table): Likewise.
19529 (sh_secondary_reload): Likewise.
19530 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19531 * config/sh/sh.md (casesi_worker_1): Likewise.
19532 (casesi_worker_2): Likewise.
19533 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19534 (fcc_comparison_operator): Likewise.
19535 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19536 (emit_soft_tfmode_cvt): Likewise.
19537 (sparc_preferred_simd_mode): Likewise.
19538 (output_cbranch): Likewise.
19539 (sparc_print_operand): Likewise.
19540 (sparc_expand_vec_perm_bmask): Likewise.
19541 (vector_init_bshuffle): Likewise.
19542 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19543 (spu_vector_mode_supported_p): Likewise.
19544 (spu_expand_insv): Likewise.
19545 (spu_emit_branch_or_set): Likewise.
19546 (spu_handle_vector_attribute): Likewise.
19547 (spu_builtin_splats): Likewise.
19548 (spu_builtin_extract): Likewise.
19549 (spu_builtin_promote): Likewise.
19550 (spu_expand_sign_extend): Likewise.
19551 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19552 (tilegx_simd_int): Likewise.
19553 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19554 (tilepro_simd_int): Likewise.
19555 * config/v850/v850.c (const_double_split): Likewise.
19556 (v850_print_operand): Likewise.
19557 (ep_memory_offset): Likewise.
19558 * config/vax/vax.c (vax_rtx_costs): Likewise.
19559 (vax_output_int_move): Likewise.
19560 (vax_output_int_add): Likewise.
19561 (vax_output_int_subtract): Likewise.
19562 * config/visium/predicates.md (visium_branch_operator): Likewise.
19563 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19564 (visium_print_operand_address): Likewise.
19565 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19566 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19567 (xtensa_expand_conditional_branch): Likewise.
19568 (xtensa_copy_incoming_a7): Likewise.
19569 (xtensa_output_literal): Likewise.
19570 * dfp.c (decimal_real_maxval): Likewise.
19571 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19574 Alan Hayward <alan.hayward@arm.com>
19575 David Sherwood <david.sherwood@arm.com>
19577 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19578 (mode_nunits_inline): Likewise.
19579 (mode_inner_inline): Likewise.
19580 (mode_unit_size_inline): Likewise.
19581 (mode_unit_precision_inline): Likewise.
19582 (emit_insn_modes_h): Likewise. Also emit a #define of the
19584 (emit_mode_wider): Add an E_ prefix to mode names.
19585 (emit_mode_complex): Likewise.
19586 (emit_mode_inner): Likewise.
19587 (emit_mode_adjustments): Likewise.
19588 (emit_mode_int_n): Likewise.
19589 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19590 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19591 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19592 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19593 (CRC32_BUILTIN, ENTRY): Likewise.
19594 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19595 (aarch64_pop_regs): Likewise.
19596 (aarch64_process_components): Likewise.
19597 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19598 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19599 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19600 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19601 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19602 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19604 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19605 (pcmpistr): Likewise.
19606 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19607 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19608 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19610 * config/rl78/rl78.c (mduc_regs): Likewise.
19611 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19612 (htm_expand_builtin): Likewise.
19613 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19614 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19615 * config/xtensa/xtensa.c (print_operand): Likewise.
19616 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19617 (NUM_MODE_VECTOR_INT): Likewise.
19618 * genoutput.c (null_operand): Likewise.
19619 (output_operand_data): Likewise.
19620 * genrecog.c (print_parameter_value): Likewise.
19621 * lra.c (debug_operand_data): Likewise.
19623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19625 * dwarf2out.c (scompare_loc_descriptor_wide)
19626 (scompare_loc_descriptor_narrow): New functions, split out from...
19627 (scompare_loc_descriptor): ...here.
19628 * expmed.c (emit_store_flag_int): New function, split out from...
19629 (emit_store_flag): ...here.
19631 2017-08-30 Richard Biener <rguenther@suse.de>
19633 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19634 (dwarf2out_early_finish): Move setting of AT_pubnames from
19635 early debug output to early finish.
19637 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19639 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19640 and -mdata-region to the assembler.
19642 2017-08-30 Richard Biener <rguenther@suse.de>
19644 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19646 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19647 (dwarf2out_early_global_decl): Only generate a DIE for the
19648 abstract origin if it doesn't already exist or is a declaration DIE.
19649 (resolve_addr): Do not add the linkage name twice when
19650 generating a stub DIE for the DW_TAG_GNU_call_site target.
19652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19654 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19655 Use machine_mode rather than int for arg1_mode.
19657 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19660 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19661 that the second argument of the built-in functions to unpack
19662 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19663 switch statement instead a lot of if statements.
19664 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19665 Allow 64-bit values to be in Altivec registers as well as
19666 traditional floating point registers.
19667 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19669 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19671 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19672 MAX_REGS_PER_ADDRESS == 1.
19674 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19676 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19677 * config/i386/i386.c (ix86_option_override_internal): Simplify
19678 setting of opts->x_flag_entry.
19680 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19681 Jakub Jelinek <jakub@redhat.com>
19682 Richard Biener <rguenther@suse.de>
19684 PR tree-optimization/81503
19685 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19686 folded constant fits in the target type; reorder tests for clarity.
19688 2017-08-29 Martin Liska <mliska@suse.cz>
19690 * passes.def: Include pass_lower_switch.
19691 * stmt.c (dump_case_nodes): Remove and move to
19692 tree-switch-conversion.
19693 (case_values_threshold): Likewise.
19694 (expand_switch_as_decision_tree_p): Likewise.
19695 (emit_case_decision_tree): Likewise.
19696 (expand_case): Likewise.
19697 (balance_case_nodes): Likewise.
19698 (node_has_low_bound): Likewise.
19699 (node_has_high_bound): Likewise.
19700 (node_is_bounded): Likewise.
19701 (emit_case_nodes): Likewise.
19702 (struct simple_case_node): New struct.
19703 (add_case_node): Remove.
19704 (emit_case_dispatch_table): Use vector instead of case_list.
19705 (reset_out_edges_aux): Remove.
19706 (compute_cases_per_edge): Likewise.
19707 (expand_case): Build list of simple_case_node.
19708 (expand_sjlj_dispatch_table): Use it.
19709 * tree-switch-conversion.c (struct case_node): Moved from
19710 stmt.c and adjusted.
19711 (emit_case_nodes): Likewise.
19712 (node_has_low_bound): Likewise.
19713 (node_has_high_bound): Likewise.
19714 (node_is_bounded): Likewise.
19715 (case_values_threshold): Likewise.
19716 (reset_out_edges_aux): Likewise.
19717 (compute_cases_per_edge): Likewise.
19718 (add_case_node): Likewise.
19719 (dump_case_nodes): Likewise.
19720 (balance_case_nodes): Likewise.
19721 (expand_switch_as_decision_tree_p): Likewise.
19722 (emit_jump): Likewise.
19723 (emit_case_decision_tree): Likewise.
19724 (try_switch_expansion): Likewise.
19725 (do_jump_if_equal): Likewise.
19726 (emit_cmp_and_jump_insns): Likewise.
19727 (fix_phi_operands_for_edge): New function.
19728 (record_phi_operand_mapping): Likewise.
19729 (class pass_lower_switch): New pass.
19730 (pass_lower_switch::execute): New function.
19731 (make_pass_lower_switch): Likewise.
19732 (conditional_probability):
19733 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19734 * tree-pass.h: Add make_pass_lower_switch.
19736 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19739 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19742 2017-08-29 Richard Biener <rguenther@suse.de>
19744 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19745 we do not add a DW_AT_inline attribute twice.
19746 (gen_subprogram_die): Remove code setting DW_AT_inline on
19747 DECL_ABSTRACT_P nodes.
19749 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19751 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19752 calls to internal functions.
19753 (gimplify_modify_expr): Likewise.
19754 * tree-call-cdce.c (use_internal_fn): Likewise.
19755 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19756 (convert_to_divmod): Set the nothrow flag.
19757 * tree-if-conv.c (predicate_mem_writes): Likewise.
19758 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19759 (vectorizable_call): Likewise.
19760 (vectorizable_store): Likewise.
19761 (vectorizable_load): Likewise.
19762 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19763 (vect_recog_mask_conversion_pattern): Likewise.
19765 2017-08-29 Martin Liska <mliska@suse.cz>
19768 * gcc.c (driver_handle_option): Add new argument.
19769 * opts-common.c (handle_option): Pass
19770 target_option_override_hook.
19771 * opts-global.c (lang_handle_option): Add new option.
19772 (set_default_handlers): Add new argument.
19773 (decode_options): Likewise.
19774 * opts.c (target_handle_option): Likewise.
19775 (common_handle_option): Call target_option_override_hook.
19776 * opts.h (struct cl_option_handler_func): Add hook for
19777 target option override.
19778 (struct cl_option_handlers): Likewise.
19779 (set_default_handlers): Add new argument.
19780 (decode_options): Likewise.
19781 (common_handle_option): Likewise.
19782 (target_handle_option): Likewise.
19783 * toplev.c (toplev::main): Pass targetm.target_option.override
19786 2017-08-29 Richard Biener <rguenther@suse.de>
19787 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19789 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19790 life to the active subtree.
19792 2017-08-28 Jeff Law <law@redhat.com>
19794 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19795 derive_equivalences.
19796 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19797 Code moved into....
19798 (edge_info::derive_equivalences): New private member function
19800 * tree-ssa-dom.c (class edge_info): Changed from a struct
19801 to a class. Add ctor/dtor, methods and data members.
19802 (edge_info::edge_info): Renamed from allocate_edge_info.
19803 Initialize additional members.
19804 (edge_info::~edge_info): New.
19805 (free_dom_edge_info): Delete the edge info.
19806 (record_edge_info): Use new class & associated member functions.
19807 Tighten forms for testing for edge equivalences.
19808 (record_temporary_equivalences): Iterate over the simple
19809 equivalences rather than assuming there's only one per edge.
19810 (cprop_into_successor_phis): Iterate over the simple
19811 equivalences rather than assuming there's only one per edge.
19812 (optimize_stmt): Use operand_equal_p rather than pointer
19813 equality for mini-DSE code.
19815 2017-08-28 Nathan Sidwell <nathan@acm.org>
19817 * gcc.c (execute): Fold SIGPIPE handling into switch
19818 statement. Adjust internal error message.
19820 2017-08-28 Richard Biener <rguenther@suse.de>
19823 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19824 Do nothing for removed DIEs.
19826 2017-08-28 Richard Biener <rguenther@suse.de>
19828 PR tree-optimization/81977
19829 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19832 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19835 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19836 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19837 using targetm.gen_mem_thread_fence.
19839 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19842 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19843 predicate to register_operand. Reorder operands.
19844 (*btr<mode>): Ditto.
19845 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19846 (*btr<mode>_mask): Ditto.
19848 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19850 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19851 * config/rs6000/xmmintrin.h: New file.
19852 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19854 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19857 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19858 parameter and_insn and return it.
19859 (recombine_lvx_pattern): Insert a copy to ensure availability of
19860 the base register of the copied masking operation at the point of
19861 the instruction replacement.
19862 (recombine_stvx_pattern): Likewise.
19864 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19866 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19867 undocumented switches.
19868 (-mpower9-dform-vector): Likewise.
19869 (-mpower9-dform): Likewise.
19870 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19871 comments to delete references to -mpower9-dform* switches.
19872 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19873 Delete reference to -mpower9-dform* switches, test for
19874 -mpower9-vector instead.
19875 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19876 (OTHER_P9_VECTOR_MASKS): Likewise.
19877 (POWERPC_MASKS): Likewise.
19878 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19879 tests against -mpower9-dform* to -mpower9-vector. Delete code
19880 that checked for -mpower9-dform* consistancy with other options.
19881 Add test for -mpower9-misc to enable other power9 switches.
19882 (rs6000_init_hard_regno_mode_ok): Likewise.
19883 (rs6000_option_override_internal): Likewise.
19884 (rs6000_emit_prologue): Likewise.
19885 (rs6000_emit_epilogue): Likewise.
19886 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19887 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19888 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19890 (emit_fusion_p9_store): Likewise.
19891 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19892 resetting these macros if the assembler does not support ISA 3.0
19894 (TARGET_P9_DFORM_VECTOR): Likewise.
19895 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19896 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19898 2017-08-25 Alan Modra <amodra@gmail.com>
19900 PR rtl-optimization/81747
19901 * cse.c (cse_extended_basic_block): Don't attempt to record
19902 equivalences for degenerate conditional branches that branch
19903 to their fall-through.
19905 2017-08-24 Martin Sebor <msebor@redhat.com>
19907 PR middle-end/81908
19908 * gimple-fold.c (size_must_be_zero_p): New function.
19909 (gimple_fold_builtin_memory_op): Call it.
19911 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19913 * config/rs6000/mm_malloc.h: New file.
19915 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19917 PR tree-optimization/81913
19918 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19919 analysis when either IVs in condition can wrap.
19921 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19923 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19924 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19926 2017-08-24 Richard Biener <rguenther@suse.de>
19929 * targhooks.c (default_target_can_inline_p): Properly
19930 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19931 is present and always compare.
19932 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19933 infer -mfpmath=sse from TARGET_SSE_P.
19934 (ix86_can_inline_p): Properly use target_option_default_node when
19935 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19937 2017-08-24 Richard Biener <rguenther@suse.de>
19940 * dwarf2out.c (output_die): Handle flag_generate_offload like
19942 (output_comp_unit): Likewise.
19943 (gen_array_type_die): Likewise.
19944 (dwarf2out_early_finish): Likewise.
19945 (note_variable_value_in_expr): Likewise.
19946 (dwarf2out_finish): Likewise. Adjust assert.
19947 * cgraphunit.c (symbol_table::compile): Move setting of
19948 flag_generate_offload earlier ...
19949 (symbol_table::finalize_compilation_unit): ... here, before
19950 early debug finalization.
19952 2017-08-24 Richard Biener <rguenther@suse.de>
19954 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19955 and ipa-fnsummary.h.
19956 (ix86_can_inline_p): When ix86_fpmath flags do not match
19957 check whether the callee uses FP math at all.
19959 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19961 PR middle-end/81931
19962 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19963 instead of TYPE_PRECISION.
19965 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19966 Alan Hayward <alan.hayward@arm.com>
19967 David Sherwood <david.sherwood@arm.com>
19969 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19971 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19973 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19975 (ix86_frame::stack_realign_allocate): New field.
19976 (struct machine_frame_state): Modify comments.
19977 (machine_frame_state::sp_realigned_fp_end): New field.
19978 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19979 layout calculation.
19980 (sp_valid_at): Add assertion to assure no attempt to access invalid
19981 offset of a realigned stack.
19982 (fp_valid_at): Likewise.
19983 (choose_baseaddr): Modify comments.
19984 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19985 ix86_expand_prologue.
19986 (ix86_expand_prologue): Modify stack realignment and allocation.
19987 (ix86_expand_epilogue): Modify comments.
19988 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19989 avx2_runtime, avx512f, and avx512f_runtime.
19991 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19993 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19994 (mstackrealign): Do not init to -1.
19995 * config/i386/i386.c (ix86_option_override_internal):
19996 Check opts_set, not opts when setting default value of
19997 opts->x_ix86_force_align_arg_pointer.
19999 2017-08-23 Richard Biener <rguenther@suse.de>
20001 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20002 lang_hooks.decl_printable_name.
20003 * print-rtl-function.c (print_rtx_function): Likewise.
20004 * tree-pretty-print.c (dump_function_header): Likewise.
20006 2017-08-23 Richard Biener <rguenther@suse.de>
20009 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20010 -g0 at compile-time.
20012 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20014 PR middle-end/19706
20015 * doc/sourcebuild.texi (Other hardware attributes):
20018 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20020 PR middle-end/19706
20021 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20022 Fix single-use check.
20024 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20026 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20028 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20030 * doc/install.texi: Modify to add more details on running selected
20033 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20035 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20036 is combined with -mabi=ms.
20037 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20040 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20042 PR tree-optimization/81488
20043 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20044 and cached_basis fields.
20045 (MAX_SPREAD): New constant.
20046 (alloc_cand_and_find_basis): Initialize new fields.
20047 (clear_visited): New function.
20048 (create_phi_basis_1): Rename from create_phi_basis, set visited
20049 and cached_basis fields.
20050 (create_phi_basis): New wrapper function.
20051 (phi_add_costs_1): Rename from phi_add_costs, add spread
20052 parameter, set visited field, short-circuit when limits reached.
20053 (phi_add_costs): New wrapper function.
20054 (record_phi_increments_1): Rename from record_phi_increments, set
20056 (record_phi_increments): New wrapper function.
20057 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20058 (phi_incr_cost): New wrapper function.
20059 (all_phi_incrs_profitable_1): Rename from
20060 all_phi_incrs_profitable, set visited field.
20061 (all_phi_incrs_profitable): New wrapper function.
20063 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20064 Alan Hayward <alan.hayward@arm.com>
20065 David Sherwood <david.sherwood@arm.com>
20067 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20068 that takes the outer and inner modes.
20069 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20070 comparison as the canonical test for a paradoxical subreg.
20071 * combine.c (simplify_set): Use paradoxical_subreg_p.
20072 (make_extraction): Likewise.
20073 (force_to_mode): Likewise.
20074 (rtx_equal_for_field_assignment_p): Likewise.
20075 (gen_lowpart_for_combine): Likewise.
20076 (simplify_comparison): Likewise.
20077 * cse.c (equiv_constant): Likewise.
20078 * expmed.c (store_bit_field_1): Likewise.
20079 * final.c (alter_subreg): Likewise.
20080 * fwprop.c (propagate_rtx): Likewise.
20081 (forward_propagate_subreg): Likewise.
20082 * ira-conflicts.c (ira_build_conflicts): Likewise.
20083 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20084 * lra-constraints.c (curr_insn_transform): Likewise.
20085 (split_reg): Likewise.
20086 * lra-eliminations.c (move_plus_up): Likewise.
20087 (lra_eliminate_regs_1): Likewise.
20088 * recog.c (general_operand): Likewise.
20089 * ree.c (combine_reaching_defs): Likewise.
20090 * reload.c (push_reload): Likewise.
20091 (find_reloads): Likewise.
20092 * reload1.c (elimination_effects): Likewise.
20093 (compute_reload_subreg_offset): Likewise.
20094 (choose_reload_regs): Likewise.
20095 * rtlanal.c (subreg_lsb_1): Likewise.
20096 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20097 (simplify_subreg): Likewise.
20098 * var-tracking.c (track_loc_p): Likewise.
20099 * emit-rtl.c (byte_lowpart_offset): Likewise.
20100 (paradoxical_subreg_p): Delete out-of-line definition.
20102 2017-08-22 Jeff Law <law@redhat.com>
20104 PR tree-optimization/81741
20105 PR tree-optimization/71947
20106 * tree-ssa-dom.c: Include tree-inline.h.
20107 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20108 equivalences if one is more expensive to compute than the other.
20109 * tree-ssa-scopedtables.h (class const_or_copies): Make
20110 record_const_or_copy_raw method private.
20111 (class avail_exprs_stack): New method simplify_binary_operation.
20112 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20113 avail_exprs_stack::simplify_binary_operation as needed.
20114 (avail_exprs_stack::simplify_binary_operation): New function.
20116 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20118 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20119 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20120 (DOT_SYMBOLS): Likewise.
20121 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20122 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20123 (RS6000_ABI_NAME): Likewise.
20124 (TARGET_CMODEL): Likewise.
20125 (TOC_SECTION_ASM_OP): Likewise.
20126 (SET_CMODEL): New macro.
20127 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20129 2017-08-22 Richard Biener <rguenther@suse.de>
20131 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20132 to allow for free-lang-data replacements similar to verify_type_variant.
20134 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20135 Alan Hayward <alan.hayward@arm.com>
20136 David Sherwood <david.sherwood@arm.com>
20138 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20139 VOIDmode for the LABEL_REF.
20141 2017-08-22 Richard Biener <rguenther@suse.de>
20143 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20145 2017-08-22 Richard Biener <rguenther@suse.de>
20147 * common.opt (feliminate-dwarf2-dups): Ignore.
20148 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20149 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20150 same_die_p_wrap, compute_section_prefix,
20151 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20152 (comdat_symbol_id, comdat_symbol_number): Likewise.
20153 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20155 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20156 (output_die): Mark unreachable path unreachable.
20157 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20158 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20159 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20160 (dwarf2out_early_finish): Likewise.
20162 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20164 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20166 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20169 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20170 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20171 (avr_attribute_table) <io, io_low, address>: Initialize
20172 .decl_required with true.
20174 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20176 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20177 undocumented debugging options.
20178 (-mvsx-scalar-double): Likewise.
20179 (-mallow-df-permute): Likewise.
20180 (-mvectorize-builtins): Likewise.
20181 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20182 (rs6000_builtin_vectorized_function): Likewise.
20183 (rs6000_builtin_md_vectorized_function): Likewise.
20184 (rs6000_opt_vars): Likewise.
20186 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20189 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20190 (*btrq_imm): Rename from *btrq.
20191 (*btcq_imm): Rename from *btcq.
20192 (btsc): New code attribute.
20193 (*<btsc><mode>): New insn pattern.
20194 (*btr<mode>): Ditto.
20195 (*<btsc><mode>_mask): New insn_and_split pattern.
20196 (*btr<mode>_mask): Ditto.
20198 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20199 Alan Hayward <alan.hayward@arm.com>
20200 David Sherwood <david.sherwood@arm.com>
20202 * function.c (pad_below): Simplify padding calculation.
20204 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20205 Alan Hayward <alan.hayward@arm.com>
20206 David Sherwood <david.sherwood@arm.com>
20208 * target.def (function_prologue): Remove frame size argument.
20209 (function_epilogue): Likewise.
20210 * doc/tm.texi: Regenerate.
20211 * final.c (final_start_function): Update call to function_prologue.
20212 (final_end_function): Update call to function_epilogue.
20213 (default_function_pro_epilogue): Remove frame size argument.
20214 * output.h (default_function_pro_epilogue): Likewise.
20215 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20216 (arm_output_function_prologue): Likewise.
20217 * config/frv/frv.c (frv_function_prologue): Likewise.
20218 (frv_function_epilogue): Likewise.
20219 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20220 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20221 (ia64_output_function_epilogue): Likewise.
20222 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20223 (m32r_output_function_epilogue): Likewise.
20224 * config/microblaze/microblaze.c (microblaze_function_prologue)
20225 (microblaze_function_epilogue): Likewise.
20226 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20227 (mips_output_function_epilogue): Likewise.
20228 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20229 (mmix_target_asm_function_epilogue): Likewise.
20230 * config/msp430/msp430.c (msp430_start_function): Likewise.
20231 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20232 (nds32_asm_function_epilogue): Likewise.
20233 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20234 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20235 (pa_output_function_epilogue): Likewise.
20236 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20237 (rs6000_output_function_epilogue): Likewise.
20238 * config/rl78/rl78.c (rl78_start_function): Likewise.
20239 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20240 (rs6000_output_function_epilogue): Likewise.
20241 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20242 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20243 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20244 (sparc_asm_function_epilogue): Likewise.
20246 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20248 * tree.h (type_has_mode_precision_p): New function.
20249 * convert.c (convert_to_integer_1): Use it.
20250 * expr.c (expand_expr_real_2): Likewise.
20251 (expand_expr_real_1): Likewise.
20252 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20253 * match.pd: Likewise.
20254 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20255 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20256 * tree-tailcall.c (process_assignment): Likewise.
20257 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20258 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20259 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20260 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20261 (vectorizable_assignment): Likewise.
20262 (vectorizable_shift): Likewise.
20263 (vectorizable_operation): Likewise.
20264 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20266 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20268 * match.pd: Add pow (C, x) simplification.
20270 2017-08-21 Richard Biener <rguenther@suse.de>
20272 PR tree-optimization/81900
20273 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20274 for blocks with abnormal predecessors.
20275 (compute_antic): Do not set visited flag prematurely.
20277 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20280 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20282 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20284 * stor-layout.h (vector_type_mode): Move to...
20285 * tree.h (vector_type_mode): ...here.
20286 * stor-layout.c (vector_type_mode): Move to...
20287 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20289 2017-08-21 Richard Biener <rguenther@suse.de>
20291 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20292 register_external_die hooks.
20293 (debug_false_tree_charstarstar_uhwistar): Declare.
20294 (debug_nothing_tree_charstar_uhwi): Likewise.
20295 * debug.c (do_nothing_debug_hooks): Adjust.
20296 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20297 (debug_nothing_tree_charstar_uhwi): Likewise.
20298 * dbxout.c (dbx_debug_hooks): Adjust.
20299 (xcoff_debug_hooks): Likewise.
20300 * sdbout.c (sdb_debug_hooks): Likewise.
20301 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20302 * dwarf2out.c (macinfo_label_base): New global.
20303 (dwarf2out_register_external_die): New function for the
20304 register_external_die hook.
20305 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20306 (dwarf2_debug_hooks): Use them.
20307 (dwarf2_lineno_debug_hooks): Adjust.
20308 (struct die_struct): Add with_offset flag.
20309 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20310 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20311 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20312 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20313 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20314 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20315 defining section names for the early LTO debug variants.
20316 (reset_indirect_string): New helper.
20317 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20318 (print_dw_val): Add support for offsetted symbol references.
20319 (get_ultimate_context): Split out from is_cxx.
20320 (is_cxx): Use get_ultimate_context.
20321 (is_fortran): Add decl overload.
20322 (compute_comp_unit_symbol): Split out worker from
20323 compute_section_prefix.
20324 (compute_section_prefix): Call compute_comp_unit_symbol and
20325 set comdat_type_p here.
20326 (output_die): Skip DIE symbol output for the LTO added one.
20327 Handle DIE symbol references with offset.
20328 (output_comp_unit): Guard section name mangling properly.
20329 For LTO debug sections emit a symbol at the section beginning
20330 which we use to refer to its DIEs.
20331 (add_abstract_origin_attribute): For DIEs registered via
20332 dwarf2out_register_external_die directly refer to the early
20333 DIE rather than indirectly through the shadow one we created.
20334 Remove obsolete call to dwarf2out_abstract_function for
20335 non-function/block origins.
20336 (gen_array_type_die): When generating early LTO debug do
20337 not emit DW_AT_string_length.
20338 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20339 late when in LTO. As suggested place a gcc_unreachable for
20340 the DECL_ABSTRACT_P case.
20341 (gen_subprogram_die): Avoid another specification DIE
20342 for early built declarations/definitions for the late LTO case.
20343 (gen_variable_die): Add type references for late duplicated VLA dies
20345 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20346 we have the abstract instance already.
20347 (process_scope_var): Adjust decl DIE contexts in LTO which
20348 first puts them in limbo.
20349 (gen_decl_die): Do not generate type DIEs late apart from
20350 types for VLAs or for decls we do not yet have a DIE. Do not
20351 call dwarf2out_abstract_function late.
20352 (dwarf2out_early_global_decl): Make sure to create DIEs
20353 for abstract instances of a decl first.
20354 (dwarf2out_late_global_decl): Adjust comment.
20355 (output_macinfo_op): With multiple macro sections use
20356 macinfo_label_base to distinguish labels.
20357 (output_macinfo): Likewise. Update macinfo_label_base.
20358 Pass in the line info label.
20359 (note_variable_value_in_expr): When generating LTO resolve
20360 all variable values here by generating DIEs as needed.
20361 (init_sections_and_labels): Add early LTO debug flag parameter
20362 and generate different sections and names if set. Add generation
20363 counter for the labels so we can have multiple of them.
20364 (reset_dies): Helper to allow DIEs to be output multiple times.
20365 (dwarf2out_finish): When outputting DIEs to the fat part of an
20366 LTO object first reset DIEs.
20367 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20368 (modified_type_die): Check for decl_ultimate_origin being self
20370 (gen_type_die_with_usage): Likewise.
20371 (gen_typedef_die): Allow decl_ultimate_origin being self.
20372 (set_decl_abstract_flags): Remove.
20373 (set_block_abstract_flags): Likewise.
20374 (dwarf2out_abstract_function): Treat the early generated DIEs
20375 as the abstract copy and only add DW_AT_inline and
20376 DW_AT_artificial here and call set_decl_origin_self.
20377 If the DIE has an abstract origin don't do anything.
20378 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20379 if we have none yet (Go fails to build one, PR78628).
20380 (variably_modified_type_p): Prevent endless recursion for Ada
20381 cyclic pointer types.
20382 * lto-streamer-in.c: Include debug.h.
20383 (dref_queue): New global.
20384 (lto_read_tree_1): Stream in DIE references.
20385 (lto_input_tree): Register DIE references.
20386 (input_function): Stream DECL_DEBUG_ARGS.
20387 * lto-streamer-out.c: Include debug.h.
20388 (lto_write_tree_1): Output DIE references.
20389 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20390 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20391 (output_function): Stream DECL_DEBUG_ARGS.
20392 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20393 Stream DECL_ABSTRACT_ORIGIN.
20394 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20395 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20396 DECL_CONTEXT for file-scope decls.
20397 * lto-streamer.h (struct dref_entry): Declare.
20398 (dref_queue): Likewise.
20399 * cfgexpand.c (pass_expand::execute): Do not call the
20400 outlining_inline_function hook here.
20401 * lto-wrapper.c (debug_obj): New global.
20402 (tool_cleanup): Unlink it if required.
20403 (debug_objcopy): New function.
20404 (run_gcc): Handle early debug sections in the IL files by
20405 extracting them to separate files, partially linkin them and
20406 feeding the result back as result to the linker.
20407 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20408 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20409 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20410 sections into a separate segment.
20411 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20413 (darwin_asm_dwarf_section): Likewise.
20414 (darwin_asm_output_dwarf_offset): Likewise.
20415 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20417 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20418 Alan Hayward <alan.hayward@arm.com>
20419 David Sherwood <david.sherwood@arm.com>
20421 * read-md.h (md_reader::record_potential_iterator_use): Replace
20422 pointer argument with an rtx and an index.
20423 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20424 (apply_mode_iterator): Likewise.
20425 (apply_code_iterator): Likewise.
20426 (apply_int_iterator): Likewise.
20427 (apply_subst_iterator): Likewise.
20428 (record_iterator_use): Likewise.
20429 (record_attribute_use): Likewise.
20430 (md_reader::record_potential_iterator_use): Likewise. Update calls
20431 to record_iterator_use and apply_iterator.
20432 (iterator_use): Replace ptr with x and index.
20433 (attribute_use): Likewise.
20434 (apply_attribute_uses): Update calls to apply_iterator.
20435 (apply_iterators): Likewise. Update initialization of iterator_use.
20436 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20437 and record_potential_iterator_use.
20438 (rtx_reader::read_rtx_operand): Likewise.
20440 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20441 Alan Hayward <alan.hayward@arm.com>
20442 David Sherwood <david.sherwood@arm.com>
20444 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20447 2017-08-21 Richard Biener <rguenther@suse.de>
20449 PR middle-end/81884
20450 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20451 at struct end conservatively when comparing common bases.
20453 2017-08-21 Richard Biener <rguenther@suse.de>
20455 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20456 (mem_ref_in_stmt): Remove.
20457 (determine_max_movement): Use ref index to get at the reference.
20458 (invariantness_dom_walker::before_dom_children): Deal with
20459 lim data already initialized.
20460 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20462 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20464 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20465 (TARGET_ISA_ROUND): Ditto.
20466 (TARGET_ROUND): Ditto.
20467 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20468 * config/i386/i386.md: Ditto.
20469 * config/i386/sse.md: Ditto.
20470 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20471 with OPTION_MASK_ISA_SSE4_1.
20473 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20476 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20477 __builtin_ia32_lzcnt_u16.
20479 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20482 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20483 (rs6000_set_current_function): Rewrite function to use it.
20485 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20488 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20489 and TYPE_WARN_IF_NOT_ALIGN.
20490 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20491 (handle_warn_if_not_align): New.
20492 (place_union_field): Call handle_warn_if_not_align.
20493 (place_field): Call handle_warn_if_not_align.
20494 Copy TYPE_WARN_IF_NOT_ALIGN.
20495 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20496 (layout_type): Likewise.
20497 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20499 (tree_decl_common): Add warn_if_not_align.
20500 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20501 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20502 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20503 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20504 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20505 * doc/extend.texi: Document warn_if_not_aligned attribute.
20506 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20508 2017-08-17 Martin Liska <mliska@suse.cz>
20511 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20512 (get_data_dependence): Use it as pointer type.
20513 (distribute_loop): Likewise.
20515 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20517 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20518 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20519 (p8_vmrgow_<mode>_direct): New define_insn.
20520 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20521 handle endianness for vmrgew and vmrgow permute patterns.
20523 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20525 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20526 * config/rs6000/rs6000-cpus.def: Remove comment.
20527 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20528 (POWERPC_MASKS): Likewise.
20529 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20530 use of TARGET_VSX_TIMODE.
20531 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20532 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20533 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20534 (rs6000_option_override_internal): Remove dead code.
20535 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20536 (rs6000_legitimize_reload_address): Likewise.
20537 (rs6000_legitimate_address_p): Likewise.
20538 (rs6000_opt_masks): Delete "vsx-timode".
20539 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20540 from function comment.
20541 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20542 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20543 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20545 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20546 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20547 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20548 (VSX_TI): Likewise.
20550 (define_peephole2): Likewise.
20552 2017-08-17 Martin Sebor <msebor@redhat.com>
20555 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20556 past the end of an array.
20557 (test_pp_format): Add test cases.
20559 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20561 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20562 missing ECF_NOTHROW flags.
20564 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20567 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20568 operands residing in integer registers.
20569 (*vsx_le_perm_load_<mode>): Likewise.
20570 (*vsx_le_perm_store_<mode>): Likewise.
20571 (define_peephole2): Add peepholes to optimize the above.
20573 2017-08-17 Marek Polacek <polacek@redhat.com>
20575 PR middle-end/81814
20576 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20577 to mimic what shorten_compare did. Change the return type to bool.
20578 (fold_cond_expr_with_comparison): Update call to
20579 operand_equal_for_comparison_p.
20580 (fold_ternary_loc): Likewise.
20582 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20584 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20586 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20588 2017-08-17 Richard Biener <rguenther@suse.de>
20590 * tree-ssa-structalias.c (solve_graph): When propagating
20591 to successors update the graphs succ edges and avoid duplicate work.
20593 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20596 * config/i386/i386.c (ix86_option_override_internal): Save target
20597 specific options after ix86_stack_protector_guard_reg was changed.
20599 2017-08-17 Richard Biener <rguenther@suse.de>
20601 PR tree-optimization/81827
20602 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20603 (new_var_info): Initialize it conservatively.
20604 (get_call_vi): Mark register vars.
20605 (new_scalar_tmp_constraint_exp): Likewise.
20606 (handle_rhs_call): Likewise.
20607 (handle_const_call): Likewise.
20608 (create_function_info_for): Likewise.
20609 (solve_constraints): Sort varinfos to separate register from
20610 non-register vars to pack points-to solution bitmaps during
20613 2017-08-17 Marek Polacek <polacek@redhat.com>
20615 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20617 2017-08-17 Richard Biener <rguenther@suse.de>
20619 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20620 to true when overflow is undefined and we saturated the result.
20622 2017-08-17 Alan Modra <amodra@gmail.com>
20625 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20626 Don't use store multiple if only one reg needs saving.
20627 (interesting_frame_related_regno): New function.
20628 (rs6000_frame_related): Don't emit frame info for regs that
20630 (rs6000_emit_epilogue): Likewise.
20632 2017-08-16 Nathan Sidwell <nathan@acm.org>
20634 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20635 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20636 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20637 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20638 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20639 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20640 (verify_type): Adjust for TYPE_BINFO move.
20641 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20642 process TYPE_BINFO directly.
20643 (hash_tree): Likewise.
20644 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20646 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20649 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20651 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20653 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20656 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20657 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20658 Add ix86_binary_operator_ok to insn constraint.
20659 (*iordi_1_bts): Ditto.
20660 (*xordi_1_btc): Ditto.
20661 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20662 Update corresponding peephole2 pattern.
20666 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20668 PR tree-optimization/81832
20669 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20670 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20672 2017-08-16 Marek Polacek <polacek@redhat.com>
20675 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20676 regarding size_zero_node.
20678 2017-08-16 Martin Liska <mliska@suse.cz>
20681 * config.gcc: Respect previously set extra_objs in case
20684 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20686 PR tree-optimization/81835
20687 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20688 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20689 not depend on the phi.
20691 2017-08-16 Alan Modra <amodra@gmail.com>
20693 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20696 2017-08-16 Alan Modra <amodra@gmail.com>
20698 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20699 (save_reg_p): ..into this. Update all callers.
20700 (first_reg_to_save): Simplify.
20702 2017-08-16 Alan Modra <amodra@gmail.com>
20704 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20707 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20711 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20712 constant count if that count is less than 32.
20714 2017-08-15 Nathan Sidwell <nathan@acm.org>
20716 * gcc.c (execute): Emit friendlier message if inferior is killed
20717 by an external cause.
20719 2017-08-15 Richard Biener <rguenther@suse.de>
20721 PR tree-optimization/81790
20722 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20723 CONSTRUCTORs from simplifying and VN.
20725 2017-08-14 Martin Sebor <msebor@redhat.com>
20727 * builtin-attrs.def: Add comments.
20729 2017-08-14 Martin Sebor <msebor@redhat.com>
20732 * doc/extend.texi (attribute nonstring): Document new attribute.
20734 2017-08-14 Martin Sebor <msebor@redhat.com>
20737 * tree-diagnostic.c (default_tree_printer): Handle %G.
20738 * gimple-pretty-print.h (percent_G_format): Declare new function.
20739 * gimple-pretty-print.c (percent_G_format): Define.
20740 * tree-pretty-print.c (percent_K_format): Add argument.
20742 2017-08-14 Martin Sebor <msebor@redhat.com>
20744 PR translation/79998
20745 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20746 Remove a stray space.
20748 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20751 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20752 (*iordi_1_bts): Ditto.
20753 (*xordi_1_btc): Ditto.
20755 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20758 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20759 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20761 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20762 quoted strings, and make more translator-friendly.
20763 (darwin_rs6000_override_options): Likewise.
20764 (rs6000_option_override_internal): Likewise.
20765 (rs6000_return_in_memory): Fix overlong line.
20766 (init_cmulative_args): Use quoted strings, and make more
20767 translator-friendly.
20768 (rs6000_pass_by_reference): Fix overlong line.
20769 (def_builtin): Use quoted strings.
20770 (altivec_expand_predicate_builtin): Use quoted strings, and make
20771 more translator-friendly.
20772 (htm_expand_builtin): Use quoted strings.
20773 (cpu_expand_builtin): Use quoted strings, and make more
20774 translator-friendly.
20775 (altivec_expand_builtin): Likewise.
20776 (paired_expand_predicate_builtin): Likewise.
20777 (rs6000_invalid_builtin): Likewise.
20778 (builtin_function_type): Use quoted strings.
20779 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20780 more translator-friendly.
20781 (rs6000_trampoline_init): Likewise.
20782 (rs6000_handle_altivec_attribute): Likewise.
20783 (rs6000_inner_target_options): Use quoted strings.
20784 (rs6000_disable_incompatible_switches): Likewise.
20785 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20786 strings, and make more translator-friendly.
20787 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20789 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20791 PR tree-optimization/81799
20792 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20793 cond_expr to simple gimple operand.
20795 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20797 PR middle-end/46932
20798 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20800 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20804 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20805 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20806 TARGET_GASISR_PROLOGUES.
20807 * config/avr/avr.c (avr_option_override): Same.
20808 (avr_pass_pre_proep::execute): Same.
20810 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20813 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20814 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20816 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20818 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20819 define_insn_and_split. Split to a memory load from 0 in
20820 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20821 using PTR mode iterator.
20822 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20823 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20824 (*add_tp_<mode>): Redefine as define_insn_and_split.
20825 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20826 address space. Merge with *add_tp_x32 using PTR mode iterator.
20827 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20828 Split to an add with a memory load from 0 in
20829 DEFAULT_TLS_SEG_REG address space.
20831 2017-08-12 Andrew Pinski <apinski@cavium.com>
20833 * config/aarch64/aarch64-option-extensions.def (rdma):
20834 Fix feature string to what Linux prints out in /proc/cpuinfo.
20836 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20839 * dwarf2out.c (modified_type_die): For C typedef types that have
20840 an ultimate origin, process the ultimate origin instead of the
20842 (gen_typedef_die): Assert that input DECLs have no ultimate
20844 (gen_type_die_with_usage): For typedef variants that have an
20845 ultimate origin, just call gen_decl_die on the original DECL.
20846 (process_scope_var): Avoid creating DIEs for local typedefs and
20847 concrete static variables.
20849 2017-08-12 Alan Modra <amodra@gmail.com>
20853 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20854 match gnu-user.h startfile.
20855 (ENDFILE_LINUX_SPEC): Similarly.
20857 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20860 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20862 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20864 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20865 * config/aarch64/aarch64.md (mov<mode>): Change.
20866 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20867 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20868 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20870 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20872 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20873 for storage order barriers.
20875 2017-08-11 Martin Liska <mliska@suse.cz>
20878 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20879 variables of void type.
20881 2017-08-11 Martin Liska <mliska@suse.cz>
20883 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20884 TARGET_SUPPORTS_ALIASES.
20885 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20886 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20887 (optimize_weakref): Likewise.
20888 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20889 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20890 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20892 2017-08-11 Martin Liska <mliska@suse.cz>
20895 * config/i386/i386.c (make_resolver_func): Do complete
20896 refactoring of the function.
20898 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20901 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20902 * config/i386/i386.c (ix86_stack_protect_guard): Use
20903 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20904 * doc/invoke.texi (x86 Options): Document
20905 -mstack-protector-guard-symbol= option.
20907 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20909 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20910 * config/i386/i386.c (ix86_split_stack_guard): New function.
20911 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20912 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20913 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20914 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20915 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20916 (split_stack_space_check): Call ix86_split_stack_guard.
20918 2017-08-10 Martin Sebor <msebor@redhat.com>
20920 * print-tree.c (print_node): Print location using the established
20922 Replace spaces with colons.
20923 (debug_raw, debug): Ditto.
20925 2017-08-10 Martin Sebor <msebor@redhat.com>
20928 * pretty-print.c (pp_format): Correct the handling of %s precision.
20930 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20933 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20935 (ix86_finalize_stack_frame_flags): This. Also clear
20936 frame_pointer_needed if -fno-omit-frame-pointer is used without
20938 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20939 with ix86_finalize_stack_frame_flags.
20940 (ix86_expand_epilogue): Likewise.
20941 (ix86_expand_split_stack_prologue): Likewise.
20942 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20944 2017-08-10 Martin Liska <mliska@suse.cz>
20947 * c-attribs.c (handle_target_attribute):
20948 Report warning for an empty string argument of target attribute.
20950 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20953 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20955 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20956 or DECL_NONLOCAL labels.
20957 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20958 or DECL_NONLOCAL labels here.
20960 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20962 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20963 to indicate when early gimple folding has been disabled.
20964 (rs6000_gimple_fold_builtin): Add debug content.
20965 (rs6000_invalid_builtin): Fix whitespace.
20966 (rs6000_expand_builtin): Fix whitespace.
20967 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20969 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20972 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20973 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20975 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20977 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20978 (qdf24xx): Likewise.
20979 * config/aarch64/aarch64.md: Include falkor.md.
20980 * config/aarch64/falkor.md: New.
20982 2017-08-09 Marek Polacek <polacek@redhat.com>
20985 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20986 * diagnostic.c (emit_diagnostic): Add a comment.
20987 (emit_diagnostic_valist): New function.
20989 2017-08-09 Marek Polacek <polacek@redhat.com>
20992 * input.c (make_location): New overload.
20993 * input.h (make_location): Declare.
20995 2017-08-08 Alan Modra <amodra@gmail.com>
20996 H.J. Lu <hongjiu.lu@intel.com>
20999 * gcc.c (NO_PIE_SPEC): Delete.
21000 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21002 (LINK_PIE_SPEC): ..to here.
21003 (LINK_COMMAND_SPEC): Support -no-pie.
21004 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21005 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21007 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21008 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21009 (ENDFILE_CRTEND_SPEC): Similarly.
21011 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21014 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21015 (mstack-protector-guard-offset=): Ditto.
21016 * config/i386/i386.c (ix86_option_override): Handle
21017 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21019 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21020 ix86_stack_protect_guard_offset variables.
21021 (TARGET_STACK_PROTECT_GUARD): Always define.
21022 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21023 and -mstack-protector-guard-offset= options.
21025 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21027 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21028 boundary case for the last candidate.
21030 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21032 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21033 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21035 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21037 PR middle-end/19706
21038 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21039 * config/aarch64/aarch64-builtins.c
21040 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21041 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21042 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21044 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21045 Andrew Pinski <pinskia@gmail.com>
21047 PR middle-end/19706
21048 * internal-fn.def (XORSIGN): New.
21049 * optabs.def (xorsign_optab): New.
21050 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21051 (convert_expand_mult_copysign): New.
21052 (pass_optimize_widening_mul::execute): Call
21053 convert_expand_mult_copysign.
21055 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21057 PR tree-optimization/81354
21058 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21059 Insert on edges rather than explicitly creating landing pads.
21060 (analyze_candidates_and_replace): Commit edge inserts.
21062 2017-08-08 Richard Biener <rguenther@suse.de>
21064 PR middle-end/81719
21065 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21066 (expand_simple_operations): Also look through ADDR_EXPRs with
21067 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21069 2017-08-08 Richard Biener <rguenther@suse.de>
21071 PR tree-optimization/81723
21072 * tree-vect-slp.c (struct bst_traits): New hash traits.
21073 (bst_fail): New global.
21074 (vect_build_slp_tree_2): New worker, split out from ...
21075 (vect_build_slp_tree): ... this now wrapping it with using
21076 bst_fail set to cache SLP tree build fails. Properly handle
21078 (vect_analyze_slp_instance): Allocate and free bst_fail.
21080 2017-08-08 Martin Liska <mliska@suse.cz>
21083 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21084 LABEL_DECLs that can be from a different function.
21086 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21088 PR tree-optimization/81744
21089 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21090 loop's number of iterations.
21092 2017-08-08 Martin Liska <mliska@suse.cz>
21094 * asan.c: Include header files.
21095 * attribs.c (build_decl_attribute_variant): New function moved
21097 (build_type_attribute_qual_variant): Likewise.
21098 (cmp_attrib_identifiers): Likewise.
21099 (simple_cst_list_equal): Likewise.
21100 (omp_declare_simd_clauses_equal): Likewise.
21101 (attribute_value_equal): Likewise.
21102 (comp_type_attributes): Likewise.
21103 (build_type_attribute_variant): Likewise.
21104 (lookup_ident_attribute): Likewise.
21105 (remove_attribute): Likewise.
21106 (merge_attributes): Likewise.
21107 (merge_type_attributes): Likewise.
21108 (merge_decl_attributes): Likewise.
21109 (merge_dllimport_decl_attributes): Likewise.
21110 (handle_dll_attribute): Likewise.
21111 (attribute_list_equal): Likewise.
21112 (attribute_list_contained): Likewise.
21113 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21114 (lookup_attribute_by_prefix): Likewise.
21115 * bb-reorder.c: Include header files.
21116 * builtins.c: Likewise.
21117 * calls.c: Likewise.
21118 * cfgexpand.c: Likewise.
21119 * cgraph.c: Likewise.
21120 * cgraphunit.c: Likewise.
21121 * convert.c: Likewise.
21122 * dwarf2out.c: Likewise.
21123 * final.c: Likewise.
21124 * fold-const.c: Likewise.
21125 * function.c: Likewise.
21126 * gimple-expr.c: Likewise.
21127 * gimple-fold.c: Likewise.
21128 * gimple-pretty-print.c: Likewise.
21129 * gimple.c: Likewise.
21130 * gimplify.c: Likewise.
21131 * hsa-common.c: Likewise.
21132 * hsa-gen.c: Likewise.
21133 * internal-fn.c: Likewise.
21134 * ipa-chkp.c: Likewise.
21135 * ipa-cp.c: Likewise.
21136 * ipa-devirt.c: Likewise.
21137 * ipa-fnsummary.c: Likewise.
21138 * ipa-inline.c: Likewise.
21139 * ipa-visibility.c: Likewise.
21141 * lto-cgraph.c: Likewise.
21142 * omp-expand.c: Likewise.
21143 * omp-general.c: Likewise.
21144 * omp-low.c: Likewise.
21145 * omp-offload.c: Likewise.
21146 * omp-simd-clone.c: Likewise.
21147 * opts-global.c: Likewise.
21148 * passes.c: Likewise.
21149 * predict.c: Likewise.
21150 * sancov.c: Likewise.
21151 * sanopt.c: Likewise.
21152 * symtab.c: Likewise.
21153 * toplev.c: Likewise.
21154 * trans-mem.c: Likewise.
21155 * tree-chkp.c: Likewise.
21156 * tree-eh.c: Likewise.
21157 * tree-into-ssa.c: Likewise.
21158 * tree-object-size.c: Likewise.
21159 * tree-parloops.c: Likewise.
21160 * tree-profile.c: Likewise.
21161 * tree-ssa-ccp.c: Likewise.
21162 * tree-ssa-live.c: Likewise.
21163 * tree-ssa-loop.c: Likewise.
21164 * tree-ssa-sccvn.c: Likewise.
21165 * tree-ssa-structalias.c: Likewise.
21166 * tree-ssa.c: Likewise.
21167 * tree-streamer-in.c: Likewise.
21168 * tree-vectorizer.c: Likewise.
21169 * tree-vrp.c: Likewise.
21170 * tsan.c: Likewise.
21171 * ubsan.c: Likewise.
21172 * varasm.c: Likewise.
21173 * varpool.c: Likewise.
21174 * tree.c: Remove functions moved to attribs.[ch].
21175 * tree.h: Likewise.
21176 * config/aarch64/aarch64.c: Add attrs.h header file.
21177 * config/alpha/alpha.c: Likewise.
21178 * config/arc/arc.c: Likewise.
21179 * config/arm/arm.c: Likewise.
21180 * config/avr/avr.c: Likewise.
21181 * config/bfin/bfin.c: Likewise.
21182 * config/c6x/c6x.c: Likewise.
21183 * config/cr16/cr16.c: Likewise.
21184 * config/cris/cris.c: Likewise.
21185 * config/darwin.c: Likewise.
21186 * config/epiphany/epiphany.c: Likewise.
21187 * config/fr30/fr30.c: Likewise.
21188 * config/frv/frv.c: Likewise.
21189 * config/ft32/ft32.c: Likewise.
21190 * config/h8300/h8300.c: Likewise.
21191 * config/i386/winnt.c: Likewise.
21192 * config/ia64/ia64.c: Likewise.
21193 * config/iq2000/iq2000.c: Likewise.
21194 * config/lm32/lm32.c: Likewise.
21195 * config/m32c/m32c.c: Likewise.
21196 * config/m32r/m32r.c: Likewise.
21197 * config/m68k/m68k.c: Likewise.
21198 * config/mcore/mcore.c: Likewise.
21199 * config/microblaze/microblaze.c: Likewise.
21200 * config/mips/mips.c: Likewise.
21201 * config/mmix/mmix.c: Likewise.
21202 * config/mn10300/mn10300.c: Likewise.
21203 * config/moxie/moxie.c: Likewise.
21204 * config/msp430/msp430.c: Likewise.
21205 * config/nds32/nds32-isr.c: Likewise.
21206 * config/nds32/nds32.c: Likewise.
21207 * config/nios2/nios2.c: Likewise.
21208 * config/nvptx/nvptx.c: Likewise.
21209 * config/pa/pa.c: Likewise.
21210 * config/pdp11/pdp11.c: Likewise.
21211 * config/powerpcspe/powerpcspe.c: Likewise.
21212 * config/riscv/riscv.c: Likewise.
21213 * config/rl78/rl78.c: Likewise.
21214 * config/rx/rx.c: Likewise.
21215 * config/s390/s390.c: Likewise.
21216 * config/sh/sh.c: Likewise.
21217 * config/sol2.c: Likewise.
21218 * config/sparc/sparc.c: Likewise.
21219 * config/spu/spu.c: Likewise.
21220 * config/stormy16/stormy16.c: Likewise.
21221 * config/tilegx/tilegx.c: Likewise.
21222 * config/tilepro/tilepro.c: Likewise.
21223 * config/v850/v850.c: Likewise.
21224 * config/vax/vax.c: Likewise.
21225 * config/visium/visium.c: Likewise.
21226 * config/xtensa/xtensa.c: Likewise.
21228 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21231 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21232 constraints since the -mupper-regs-* switches have been
21234 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21235 into a vector from a double word element that was extracted from
21236 another vector, and eliminate extra XXPERMDI instructions.
21237 (vsx_concat_<mode>_2): Likewise.
21238 (vsx_concat_<mode>_3): Likewise.
21239 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21240 concat to allow optimizing inserts from previous extracts.
21242 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21244 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21245 memory reference to a SSP offset in TLS address space.
21246 (ix86_print_operand) <case '@'>: Remove.
21247 (ix86_print_operand_punct_valid_p): Remove '@' code.
21248 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21249 UNSPEC_SP_TLS_TEST.
21250 (stack_tls_protect_set_<mode>): Remove.
21251 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21252 (stack_tls_protect_test_<mode>): Remove.
21253 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21255 2017-08-07 Olivier Hainque <hainque@adacore.com>
21258 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21260 2017-08-07 Douglas Rupp <rupp@adacore.com>
21262 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21263 variable was referenced as multidir in command.
21265 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21268 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21271 2017-08-07 Martin Liska <mliska@suse.cz>
21273 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21274 * config/rl78/rl78.c: Add include of attribs.h.
21275 * config/sh/sh.c: Likewise.
21276 * config/v850/v850.c: Likewise.
21278 2017-08-07 Tom de Vries <tom@codesourcery.com>
21280 PR middle-end/78266
21281 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21283 2017-08-07 Martin Liska <mliska@suse.cz>
21285 * config/mips/mips.c: Include attribs.h.
21287 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21290 * doc/invoke.texi: Document change in behvaior for -Ofast for
21293 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21295 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21297 (aarch64_pop_regs): Likewise.
21298 (aarch64_gen_load_pair): Likewise.
21299 (aarch64_save_callee_saves): Likewise.
21300 (aarch64_restore_callee_saves): Likewise.
21302 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21304 * config/i386/i386.c: Revert the last change.
21306 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21309 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21311 (ix86_finalize_stack_frame_flags): This. Also clear
21312 frame_pointer_needed if -fno-omit-frame-pointer is used without
21314 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21315 with ix86_finalize_stack_frame_flags.
21316 (ix86_expand_epilogue): Likewise.
21317 (ix86_expand_split_stack_prologue): Likewise.
21319 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21322 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21323 to P_AES for Westmere.
21325 2017-08-07 Jonathan Yong <10walls@gmail.com>
21327 * config/i386/mingw.opt (fset-stack-executable): Removed.
21328 * config/i386/cygming.opt (fset-stack-executable): Moved
21330 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21332 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21334 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21336 2017-08-07 Marek Polacek <polacek@redhat.com>
21338 PR middle-end/81737
21339 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21341 2017-08-07 Martin Liska <mliska@suse.cz>
21343 * attribs.h (canonicalize_attr_name): New function.
21344 (cmp_attribs): Move from c-format.c and adjusted.
21345 (is_attribute_p): Moved from tree.h.
21346 * tree-inline.c: Add new includes.
21347 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21348 (private_is_attribute_p): Remove.
21349 (private_lookup_attribute): Likewise.
21350 (private_lookup_attribute_by_prefix): Simplify.
21351 (remove_attribute): Use is_attribute_p.
21352 * tree.h: Remove removed declarations.
21354 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21356 PR middle-end/81698
21357 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21358 instead of computing it in the function. Formatting fix.
21359 (expand_case): Don't rely on default_edge being the first edge,
21360 clear it if removing it, pass default_edge to
21361 emit_case_dispatch_table.
21362 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21365 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21367 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21368 insn in the function, emit NOP after the insn.
21370 2017-08-06 Tom de Vries <tom@codesourcery.com>
21372 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21375 2017-08-06 Tom de Vries <tom@codesourcery.com>
21377 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21380 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21382 PR tree-optimization/57371
21383 * match.pd: New pattern.
21385 2017-08-04 Marek Polacek <polacek@redhat.com>
21387 PR middle-end/81695
21388 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21389 perform the computation in offset_int.
21391 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21393 PR tree-optimization/81136
21394 * tree-vectorizer.h: Include tree-hash-traits.h.
21395 (vec_base_alignments): New typedef.
21396 (vec_info): Add a base_alignments field.
21397 (vect_record_base_alignments): Declare.
21398 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21400 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21401 (create_data_ref): Add an is_conditional_in_stmt argument.
21402 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21403 the is_conditional_in_stmt field.
21404 (data_ref_loc): Add an is_conditional_in_stmt field.
21405 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21406 (find_data_references_in_stmt): Update call to create_data_ref.
21407 (graphite_find_data_references_in_stmt): Likewise.
21408 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21409 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21410 (vect_record_base_alignment): New function.
21411 (vect_record_base_alignments): Likewise.
21412 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21413 for nested statements even if we fail to compute a misalignment.
21414 Use pooled base alignments for unconditional references.
21415 (vect_find_same_alignment_drs): Compare base addresses instead
21417 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21418 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21420 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21422 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21423 Add an explicit name for the enum. Use auto_vec for slp_instances
21424 and grouped_stores.
21425 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21427 (_bb_vec_info): Add a constructor and destructor.
21428 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21429 (destroy_loop_vec_info): Delete.
21430 (vect_destroy_datarefs): Likewise.
21431 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21432 (vec_info::vec_info): New function.
21433 (vec_info::~vec_info): Likewise.
21434 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21435 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21436 destroy_loop_vec_info.
21437 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21438 (_loop_vec_info::_loop_vec_info): ...this.
21439 (destroy_loop_vec_info): Replace with...
21440 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21441 the stmt_vec_infos. Leave handling of vec_info information to its
21442 destructor. Remove explicit vector releases.
21443 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21444 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21445 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21446 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21447 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21448 (destroy_bb_vec_info): Replace with...
21449 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21450 information to its destructor.
21451 (vect_slp_analyze_bb_1): Use new and delete instead of
21452 new_bb_vec_info and destroy_bb_vec_info.
21453 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21456 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21458 * tree-data-ref.h (subscript): Add access_fn field.
21459 (data_dependence_relation): Add could_be_independent_p.
21460 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21461 (same_access_functions): Move to tree-data-ref.c.
21462 * tree-data-ref.c (ref_contains_union_access_p): New function.
21463 (access_fn_component_p): Likewise.
21464 (access_fn_components_comparable_p): Likewise.
21465 (dr_analyze_indices): Add a reference to access_fn_component_p.
21466 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21468 (constant_access_functions): Likewise.
21469 (add_other_self_distances): Likewise.
21470 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21471 (initialize_data_dependence_relation): Use XCNEW and remove
21472 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21473 of access functions that have the same type. Allow the
21474 subsequence to end with different bases in some circumstances.
21475 Record the chosen access functions in SUB_ACCESS_FN.
21476 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21477 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21478 (subscript_dependence_tester_1): Likewise dra and drb.
21479 (build_classic_dist_vector): Update calls accordingly.
21480 (subscript_dependence_tester): Likewise.
21481 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21482 DDR_COULD_BE_INDEPENDENT_P.
21483 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21484 comp_alias_ddrs instead of may_alias_ddrs.
21485 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21487 (vect_analyze_data_ref_dependence): Use it if
21488 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21489 distance vectors if that fails.
21490 (dependence_distance_ge_vf): New function.
21491 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21492 LOOP_VINFO_MAY_ALIAS_DDRS.
21494 2017-08-04 Richard Biener <rguenther@suse.de>
21496 PR middle-end/81705
21497 * fold-const.c (fold_binary_loc): Properly restrict
21498 minus_var0 && minus_var1 case when associating undefined overflow
21501 2017-08-04 Tom de Vries <tom@codesourcery.com>
21503 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21505 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21507 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21508 Don't start diagnostic messages with a capital letter.
21509 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21511 (rs6000_invalid_builtin): Likewise.
21512 (rs6000_trampoline_init): Likewise.
21514 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21517 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21518 after setting changeable df flags.
21520 2017-08-03 Richard Biener <rguenther@suse.de>
21522 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21523 up if the use is in USE - X.
21525 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21527 * toplev.c (dumpfile.h): New include.
21528 (internal_error_reentered): New static function. Use it...
21529 (internal_error_function): ...here to handle reentered internal_error.
21531 2017-08-03 Richard Biener <rguenther@suse.de>
21533 PR middle-end/81148
21534 * fold-const.c (split_tree): Add minus_var and minus_con
21535 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21536 here but always use minus_*.
21537 (associate_trees): Assert we never associate with MINUS_EXPR
21538 and NULL first operand. Do not recurse for PLUS_EXPR operands
21539 when associating as MINUS_EXPR either.
21540 (fold_binary_loc): Track minus_var and minus_con.
21542 2017-08-03 Tom de Vries <tom@codesourcery.com>
21545 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21546 ACCEL_COMPILER, apply finish_options on
21547 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21549 2017-08-03 Tom de Vries <tom@codesourcery.com>
21552 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21553 function_entry_patch_area_size > 0.
21555 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21558 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21559 instead of 10??LU, perform unit multiplication in wide_int,
21560 don't change alloc_object_size_limit if the limit is larger
21563 PR tree-optimization/81655
21564 PR tree-optimization/81588
21565 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21566 the case when ranges[i].low and high are 1 for unsigned type with
21569 PR middle-end/81052
21570 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21571 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21573 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21575 * tree-vrp.h: Add include guard.
21577 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21580 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21581 (ud2): New insn pattern.
21582 * config/i386/i386.c (ix86_expand_epilogue):
21583 For naked functions, generate ud2 instead of trap insn.
21585 2017-08-02 Marek Polacek <polacek@redhat.com>
21588 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21590 2017-08-02 Tom de Vries <tom@codesourcery.com>
21591 Cesar Philippidis <cesar@codesourcery.com>
21593 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21594 Add missing edge probabilities.
21596 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21598 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21599 Correct endianness.
21601 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21603 PR middle-end/79499
21604 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21605 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21606 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21607 prologue_seq sequences - if any.
21609 2017-08-02 Richard Biener <rguenther@suse.de>
21611 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21612 via vectors if supported, integer extracts via punning if supported
21613 or otherwise vector extracts.
21615 2017-08-02 Richard Biener <rguenther@suse.de>
21617 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21619 (bitmap_insert_into_set): ... this.
21621 2017-08-02 Richard Biener <rguenther@suse.de>
21623 PR tree-optimization/81633
21625 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21627 PR tree-optimization/71752
21628 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21630 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21632 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21633 (machine_function::call_ms2sysv_pad_out): Remove field.
21634 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21635 (ix86_compute_frame_layout): Likewise.
21637 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21640 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21641 attribute with interrupt attribute.
21643 2017-08-01 Andrew Pinski <apinski@cavium.com>
21645 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21646 BIT_INSERT_EXPR's operand 1
21647 to see if the types precision matches.
21649 2017-08-01 Martin Liska <mliska@suse.cz>
21651 PR middle-end/70140
21652 * builtins.c (expand_builtin_memcpy_args): Remove.
21653 (expand_builtin_memcpy): Call newly added function
21654 expand_builtin_memory_copy_args.
21655 (expand_builtin_memcpy_with_bounds): Likewise.
21656 (expand_builtin_mempcpy): Remove last argument.
21657 (expand_builtin_mempcpy_with_bounds): Likewise.
21658 (expand_builtin_memory_copy_args): New function created from
21659 expand_builtin_mempcpy_args with small modifications.
21660 (expand_builtin_mempcpy_args): Remove.
21661 (expand_builtin_stpcpy): Remove unused argument.
21662 (expand_builtin): Likewise.
21663 (expand_builtin_with_bounds): Likewise.
21665 2017-08-01 Martin Liska <mliska@suse.cz>
21668 Make mempcpy more optimal (PR middle-end/70140).
21670 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21673 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21674 __builtin_vec_cmpne verify both arguments are compatible vectors
21675 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21676 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21677 move computation of aligned to after checking the argument types.
21681 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21682 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21685 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21686 Doug Rupp <rupp@adacore.com>
21687 Olivier Hainque <hainque@adacore.com>
21689 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21690 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21692 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21693 for TARGET_OS_CPP_BUILTIN.
21694 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21695 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21697 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21698 passing required abi options to the assembler for EABI configurations.
21699 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21700 of .text.hot and .text.unlikely sections for kernel modules when
21701 using ARM style exceptions.
21702 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21703 options. Add EXTRA_CC1_SPEC.
21704 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21706 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21708 (ARM_TARGET2_DWARF_FORMAT): Define.
21709 * config/arm/t-vxworks: Adjust multilib control to removal of the
21710 Diab command line options.
21712 2017-08-01 Martin Liska <mliska@suse.cz>
21714 PR gcov-profile/81561
21715 * gcov.c (unblock): Make unblocking safe as we need to preserve
21716 index correspondence of blocks and block_lists.
21718 2017-08-01 Richard Biener <rguenther@suse.de>
21720 PR tree-optimization/81181
21721 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21722 (compute_antic): ... end of iteration here.
21724 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21726 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21727 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21728 (ftree-slp-vectorize): Likewise.
21729 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21730 can no longer be set independent of flag_tree_loop_vectorize.
21731 * omp-general.c (emp_max_vf): Likewise.
21732 * opts.c (enable_fdo_optimizations): Remove references to
21733 flag_tree_vectorize, these are now implicit.
21734 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21735 and leave it for the options machinery.
21737 2017-08-01 Martin Liska <mliska@suse.cz>
21739 PR middle-end/70140
21740 * builtins.c (expand_builtin_memcpy_args): Remove.
21741 (expand_builtin_memcpy): Call newly added function
21742 expand_builtin_memory_copy_args.
21743 (expand_builtin_memcpy_with_bounds): Likewise.
21744 (expand_builtin_mempcpy): Remove last argument.
21745 (expand_builtin_mempcpy_with_bounds): Likewise.
21746 (expand_builtin_memory_copy_args): New function created from
21747 expand_builtin_mempcpy_args with small modifications.
21748 (expand_builtin_mempcpy_args): Remove.
21749 (expand_builtin_stpcpy): Remove unused argument.
21750 (expand_builtin): Likewise.
21751 (expand_builtin_with_bounds): Likewise.
21753 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21756 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21757 print "ds:" only for immediates in generic address space.
21759 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21762 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21763 (ix86_function_ok_for_sibcall): Return false for naked functions.
21765 2017-08-01 Richard Biener <rguenther@suse.de>
21767 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21768 (compute_antic): Seed worklist with exit block predecessors.
21769 * cfganal.c (dfs_find_deadend): For a cycle return the source
21770 of the edge closing it.
21772 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21774 * config/aarch64/aarch64.c
21775 (aarch64_can_const_movi_rtx_p): Move 0 check.
21777 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21779 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21780 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21782 * match.pd: Ditto in address comparison pattern.
21784 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21786 PR tree-optimization/81627
21787 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21788 closed ssa form for store-store chain.
21790 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21792 PR tree-optimization/81620
21793 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21794 for store-store chain.
21796 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21798 PR tree-optimization/81588
21799 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21800 ranges[i].in_p, invert comparison code ccode. For >/>=,
21801 swap rhs1 and rhs2 and comparison code unconditionally,
21802 for </<= don't do that. Don't swap rhs1/rhs2 again if
21803 ranges[i].in_p, instead invert comparison code ccode if
21804 opcode or oe->rank is BIT_IOR_EXPR.
21807 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21808 a direct optab to conversion optab.
21809 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21810 with GET_MODE_INNER as last argument instead of optab_handler.
21811 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21812 vector extraction if possible and optab is available.
21813 * expr.c (store_constructor): Use convert_optab_handler instead
21814 of optab_handler. Use vector initialization from smaller
21815 vectors if possible and optab is available.
21816 * tree-vect-stmts.c (vectorizable_load): Likewise.
21817 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21818 now have two modes.
21819 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21820 of vec_init from half-sized vectors with the same element mode.
21821 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21822 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21823 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21824 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21825 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21826 after mode in gen_vec_extract* calls.
21827 (vec_extract<mode>): Renamed to ...
21828 (vec_extract<mode><ssescalarmodelower>): ... this.
21829 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21830 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21831 element mode after mode in gen_vec_init* calls.
21832 (VEC_INIT_HALF_MODE): New mode iterator.
21833 (vec_init<mode>): Renamed to ...
21834 (vec_init<mode><ssescalarmodelower>): ... this.
21835 (vec_init<mode><ssehalfvecmodelower>): New expander.
21836 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21837 (vec_extractv2sfsf): ... this.
21838 (vec_initv2sf): Renamed to ...
21839 (vec_initv2sfsf): ... this.
21840 (vec_extractv2si): Renamed to ...
21841 (vec_extractv2sisi): ... this.
21842 (vec_initv2si): Renamed to ...
21843 (vec_initv2sisi): ... this.
21844 (vec_extractv4hi): Renamed to ...
21845 (vec_extractv4hihi): ... this.
21846 (vec_initv4hi): Renamed to ...
21847 (vec_initv4hihi): ... this.
21848 (vec_extractv8qi): Renamed to ...
21849 (vec_extractv8qiqi): ... this.
21850 (vec_initv8qi): Renamed to ...
21851 (vec_initv8qiqi): ... this.
21852 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21853 (vec_init<mode>): Renamed to ...
21854 (vec_init<mode><VEC_base_l>): ... this.
21855 (vec_extract<mode>): Renamed to ...
21856 (vec_extract<mode><VEC_base_l>): ... this.
21857 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21858 (vec_initv2sfsf): ... this.
21859 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21860 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21861 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21862 element mode after mode in gen_vec_init* calls.
21863 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21864 (vec_init<mode><Vel>): ... this.
21865 (vec_extract<mode>): Renamed to ...
21866 (vec_extract<mode><Vel>): ... this.
21867 * config/aarch64/iterators.md (Vel): New mode attribute.
21868 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21869 Add element mode after mode in gen_vec_extract* calls.
21870 * config/s390/vector.md (non_vec_l): New mode attribute.
21871 (vec_extract<mode>): Renamed to ...
21872 (vec_extract<mode><non_vec_l>): ... this.
21873 (vec_init<mode>): Renamed to ...
21874 (vec_init<mode><non_vec_l>): ... this.
21875 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21876 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21878 * config/arm/iterators.md (V_elem_l): New mode attribute.
21879 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21880 (vec_extract<mode><V_elem_l>): ... this.
21881 (vec_extractv2di): Renamed to ...
21882 (vec_extractv2didi): ... this.
21883 (vec_init<mode>): Renamed to ...
21884 (vec_init<mode><V_elem_l>): ... this.
21885 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21886 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21887 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21888 Add element mode after gen_vec_extract* calls.
21889 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21890 (vec_init<mode><unitmode>): ... this.
21891 (vec_extract<mode>): Renamed to ...
21892 (vec_extract<mode><unitmode>): ... this.
21893 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21894 (vec_init<mode><unitmode>): ... this.
21895 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21896 (vec_initv2sfsf): ... this.
21897 (vec_extractv2sf): Renamed to ...
21898 (vec_extractv2sfsf): ... this.
21899 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21900 Add element mode after gen_vec_extract* calls.
21901 * config/mips/mips.md (unitmode): New mode iterator.
21902 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21903 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21904 * config/spu/spu.md (inner_l): New mode attribute.
21905 (vec_init<mode>): Renamed to ...
21906 (vec_init<mode><inner_l>): ... this.
21907 (vec_extract<mode>): Renamed to ...
21908 (vec_extract<mode><inner_l>): ... this.
21909 * config/sparc/sparc.md (veltmode): New mode iterator.
21910 (vec_init<VMALL:mode>): Renamed to ...
21911 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21912 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21913 (vec_initv2sisi): ... this.
21914 (vec_initv2sf): Renamed to ...
21915 (vec_initv2sfsf): ... this.
21916 (vec_extractv2sf): Renamed to ...
21917 (vec_extractv2sfsf): ... this.
21918 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21919 (vec_init<mode>): Renamed to ...
21920 (vec_init<mode><VEC_base_l>): ... this.
21921 (vec_extract<mode>): Renamed to ...
21922 (vec_extract<mode><VEC_base_l>): ... this.
21923 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21924 (vec_initv2sfsf): ... this.
21925 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21926 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21927 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21928 gen_vec_init* calls.
21930 2017-08-01 Richard Biener <rguenther@suse.de>
21932 PR tree-optimization/81297
21933 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21934 TREE_OVERFLOW from INTEGER_CSTs.
21936 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21938 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21940 2017-07-31 Carl Love <cel@us.ibm.com>
21942 * config/rs6000/rs6000-c: Add support for built-in functions
21943 vector signed char vec_xl_be (signed long long, signed char *);
21944 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21945 vector signed int vec_xl_be (signed long long, signed int *);
21946 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21947 vector signed long long vec_xl_be (signed long long, signed long long *);
21948 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21949 vector signed short vec_xl_be (signed long long, signed short *);
21950 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21951 vector double vec_xl_be (signed long long, double *);
21952 vector float vec_xl_be (signed long long, float *);
21953 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21954 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21955 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21957 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21958 (altivec_expand_builtin): Add switch statement to call
21959 altivec_expand_xl_be for each builtin.
21960 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21961 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21962 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21963 __builtin_vsx_le_be_v16qi.
21964 * doc/extend.texi: Update the built-in documentation file for the
21965 new built-in functions.
21967 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21970 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21972 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21974 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21976 * config.gcc: Add z14.
21977 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21978 CPU model numbers for z13s and z14.
21979 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21981 * config/s390/s390-opts.h (enum processor_type): Rename
21982 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21983 * config/s390/s390.c (processor_table): Add field for CPU name to
21984 be passed to Binutils.
21985 (s390_asm_output_machine_for_arch): Use the new field in
21986 processor_table for Binutils.
21987 (s390_expand_builtin): Replace arch12 with z14.
21988 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21989 (s390_get_sched_attrmask): Likewise.
21990 (s390_get_unit_mask): Likewise.
21991 * config/s390/s390.opt: Add z14 to processor_type enum.
21993 2017-07-31 Martin Jambor <mjambor@suse.cz>
21996 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21997 regardless of optimization level.
21999 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22000 Martin Liska <mliska@suse.cz>
22002 * predict.def: Remove old comment and adjust probability.
22003 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22004 PREDICT statements.
22006 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22009 * config/i386/i386.c (ix86_function_naked): New function.
22010 (ix86_can_use_return_insn_p): Return false for naked functions.
22011 (ix86_expand_prologue): Skip prologue for naked functions.
22012 (ix86_expand_epilogue): Skip epilogue for naked functions
22013 and emit trap instruction.
22014 (ix86_warn_func_return): New function.
22015 (ix86_attribute_table): Add "naked" attribute specification.
22016 (TARGET_WARN_FUNC_RETURN): Define.
22017 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22019 2017-07-31 Martin Liska <mliska@suse.cz>
22021 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22022 (dump_gimple_bb_header): Always dump BB info.
22023 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22025 2017-07-31 Martin Liska <mliska@suse.cz>
22028 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22029 also with current_function_decl non-null equality.
22031 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22034 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22035 change type to the element type, instead add eltype variable and
22036 use it where we are interested in the element type.
22038 PR tree-optimization/81603
22039 * ipa-polymorphic-call.c
22040 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22041 offset arithmetic in offset_int, bail out if the resulting bit offset
22042 doesn't fit into shwi.
22044 2017-07-31 Martin Liska <mliska@suse.cz>
22046 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22047 (gimplify_save_expr): Fix comment.
22049 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22052 * config/i386/i386.c (ix86_function_arg): Update arguments for
22054 (ix86_compute_frame_layout): Set the initial stack offset to
22055 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22056 INCOMING_FRAME_SP_OFFSET.
22057 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22058 stack before exception handler returns.
22059 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22060 the 'ERROR_CODE' for exception handler.
22062 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22064 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22065 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22066 (ASM_OUTPUT_REG_POP): Ditto.
22067 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22068 instead of asm_fprintf to output pure string.
22070 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22072 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22073 to imported_module_or_decl hook.
22074 (debug_nothing_tree_tree_tree_bool): Remove.
22075 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22076 * debug.c (do_nothing_debug_hooks): Use
22077 debug_nothing_tree_tree_tree_bool_bool instead of
22078 debug_nothing_tree_tree_tree_bool.
22079 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22080 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22081 * sdbout.c (sdb_debug_hooks): Likewise.
22082 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22083 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22085 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22086 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22087 attribute, don't add anything.
22089 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22091 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22092 (fold_build2_stat_loc): Likewise.
22093 (fold_build3_stat_loc): Likewise.
22094 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22095 (fold_build1_loc): Remove macro.
22096 (fold_build2_loc): Likewise.
22097 (fold_build3_loc): Likewise.
22099 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22101 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22102 (gimple_build_debug_bind_source_stat): Likewise.
22103 * gimple.h (gimple_build_debug_bind): Remove macro.
22104 (gimple_build_debug_bind_source): Likewise.
22106 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22108 * bitmap.c (bitmap_alloc): Adjust.
22109 (bitmap_gc_alloc): Likewise.
22110 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22112 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22114 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22115 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22116 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22117 (bitmap_gc_alloc_stat): Likewise.
22118 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22120 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22122 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22123 * rtl.h (shallow_copy_rtx): Remove macro.
22125 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22127 * emit-rtl.c (gen_raw_REG): Adjust.
22128 * gengenrtl.c (gendef): Likewise.
22129 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22130 * rtl.h (rtx_alloc): Remove macro.
22132 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22134 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22135 (build_tree_list_stat): Likewise.
22136 * tree.h (build_tree_list): Remove macro.
22137 (build_tree_list_vec): Likewise.
22139 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22141 * tree.c (make_vector_stat): Remove _stat from name.
22142 (build_vector_stat): Likewise.
22143 * tree.h (make_vector_stat): Remove macro.
22144 (build_vector_stat): Likewise.
22146 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22148 * tree.h (build_var_debug_value): Remove prototype.
22150 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22152 * tree.c (tree_cons_stat): Remove _stat from name.
22153 * tree.h (tree_cons): Remove macro.
22155 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22157 * tree.c (build_vl_exp_stat): Remove _stat from name.
22158 * tree.h (build_vl_exp): Remove macro.
22160 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22162 * tree.c (build_decl_stat): Remove _stat from name.
22163 * tree.h (build_decl): Remove macro.
22165 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22167 * gimple.c (gimple_build_with_ops_stat): Adjust.
22168 (gimple_alloc_stat): Remove _stat from name.
22169 * gimple.h (gimple_alloc): Remove macro.
22171 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22173 * tree.c (make_tree_vec_stat): Remove _stat from name.
22174 (grow_tree_vec_stat): Likewise.
22175 * tree.h (make_tree_vec_stat): Adjust prototype.
22176 (grow_tree_vec_stat): Likewise.
22177 (make_tree_vec): Remove macro.
22178 (grow_tree_vec): Likewise.
22180 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22182 * fold-const.c (fold_build1_stat_loc): Adjust.
22183 (fold_build2_stat_loc): Likewise.
22184 (fold_build3_stat_loc): Likewise.
22185 * tree.c (build0_stat): Remove _stat from name.
22186 (build1_stat): Likewise.
22187 (build2_stat): Likewise.
22188 (build3_stat): Likewise.
22189 (build4_stat): Likewise.
22190 (build5_stat): Likewise.
22191 * tree.h (build1_loc): Remove macro, and rename _stat function
22193 (build2_loc): Likewise.
22194 (build3_loc): Likewise.
22195 (build4_loc): Likewise.
22196 (build5_loc): Likewise.
22198 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22200 * tree.c (make_int_cst_stat): Remove _stat from name.
22201 * tree.h (make_int_cst_stat): Adjust prototype.
22202 (make_int_cst): Remove macro.
22204 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22206 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22207 * tree.h (make_tree_binfo_stat): Adjust prototype.
22208 (make_tree_binfo): Remove.
22210 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22212 * tree.c (copy_node_stat): Rename to copy_node.
22213 (build_distinct_type_copy): Adjust.
22214 * tree.h (copy_node_stat): Adjust prototype.
22215 (copy_node): Remove macro.
22217 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22219 * tree.c (make_node_stat): rename to make_node.
22220 (build_tree_list_stat): Adjust.
22221 (build0_stat): Likewise.
22222 (build2_stat): Likewise.
22223 (build3_stat): Likewise.
22224 (build4_stat): Likewise.
22225 (build5_stat): Likewise.
22226 (build_decl_stat): Likewise.
22227 * tree.h (make_node_stat): Adjust prototype.
22228 (make_node): remove macro.
22230 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22232 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22233 (PPC_FEATURE2_SCV): Likewise.
22234 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22236 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22238 * config/aarch64/aarch64.c
22239 (aarch64_internal_mov_immediate): Add new special pattern.
22240 * config/aarch64/aarch64.md (*movdi_aarch64):
22241 Add reg/32bit const mov case.
22243 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22244 Richard Sandiford <richard.sandiford@linaro.org>
22246 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22247 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22248 Add integer and movi cases.
22249 (movi-split-hf-df-sf split, fp16): New.
22250 (enabled): Added TARGET_FP_F16INST.
22251 * config/aarch64/iterators.md (GPF_HF): New.
22252 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22254 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22256 * config/aarch64/aarch64.c
22257 (aarch64_simd_container_mode): Add prototype.
22258 (aarch64_expand_mov_immediate): Add HI support.
22259 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22260 (aarch64_can_const_movi_rtx_p): New.
22261 (aarch64_preferred_reload_class):
22262 Remove restrictions of using FP registers for certain SIMD operations.
22263 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22264 (aarch64_valid_floating_const): Add integer move validation.
22265 (aarch64_simd_imm_scalar_p): Remove.
22266 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22267 (aarch64_legitimate_constant_p): Expand list of supported cases.
22268 * config/aarch64/aarch64-protos.h
22269 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22270 (aarch64_reinterpret_float_as_int): New.
22271 (aarch64_simd_imm_scalar_p): Remove.
22272 * config/aarch64/constraints.md (Uvi): New.
22273 (Dd): Split into Ds and new Dd.
22274 * config/aarch64/aarch64.md (*movsi_aarch64):
22276 (*movdi_aarch64): Add SIMD mov case.
22278 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22280 * tree-predcom.c: (struct chain): Handle store-store chain in which
22281 stores for elimination only store loop invariant values.
22282 (execute_pred_commoning_chain): Ditto.
22283 (prepare_initializers_chain_store_elim): Ditto.
22284 (prepare_finalizers): Ditto.
22285 (is_inv_store_elimination_chain): New function.
22286 (initialize_root_vars_store_elim_1): New function.
22288 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22290 * tree-predcom.c: Revise general description of the pass.
22291 (enum chain_type): New enum type for store elimination.
22292 (struct chain): New field supporting store elimination.
22293 (struct component): Ditto.
22294 (dump_chain): Dump store-stores chain.
22295 (release_chain): Release resources.
22296 (split_data_refs_to_components): Compute and create component
22297 contains only stores for elimination.
22298 (get_chain_last_ref_at): New function.
22299 (make_invariant_chain): Initialization.
22300 (make_rooted_chain): Specify chain type in parameter and record it.
22301 (add_looparound_copies): Skip for store-stores chain.
22302 (determine_roots_comp): Compute type of chain and pass it to
22304 (initialize_root_vars_store_elim_2): New function.
22305 (finalize_eliminated_stores): New function.
22306 (remove_stmt): Handle store for elimination.
22307 (execute_pred_commoning_chain): Execute predictive commoning on
22308 store-store chains.
22309 (determine_unroll_factor): Skip unroll for store-stores chain.
22310 (prepare_initializers_chain_store_elim): New function.
22311 (prepare_initializers_chain): Hanlde store-store chain.
22312 (prepare_finalizers_chain, prepare_finalizers): New function.
22313 (tree_predictive_commoning_loop): Return integer value indicating
22314 if loop is unrolled or lcssa form is corrupted.
22315 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22317 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22319 * tree-predcom.c (initialize_root): Delete.
22320 (execute_pred_commoning_chain): Initialize root vars and replace
22321 reference of non-combined chain directly, rather than call above
22324 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22326 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22327 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22329 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22331 * tree-predcom.c (struct chain): New field init_seq.
22332 (release_chain): Release init_seq.
22333 (prepare_initializers_chain): Record intialization stmts in above
22335 (insert_init_seqs): New function.
22336 (tree_predictive_commoning_loop): Call insert_init_seqs.
22338 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22340 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22342 2017-07-28 Richard Biener <rguenther@suse.de>
22344 * match.pd: Remove superfluous :c.
22345 * genmatch.c (simplify::id): Add member.
22346 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22348 (current_id): New global.
22349 (dt_node::parent): Move from ...
22350 (dt_operand::parent): ... here. Add for_id member.
22351 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22352 (decision_tree::find_node): Relax order requirement when
22353 merging DT_TRUE nodes to ones inbetween the current simplify
22354 and the one we try to merge with. Add diagnostic whenever
22355 we need to enforce pattern order by not merging.
22356 (decision_tree::insert): Set current_id.
22357 (decision_tree::print_node): Dump parent node and for_id.
22358 (parser::last_id): Add member.
22359 (parser::push_simplify): Assign unique id.
22360 (parser::parser): Initialize last_id.
22362 2017-07-28 Martin Liska <mliska@suse.cz>
22365 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22366 gimple_build_debug_bind.
22368 2017-07-28 Richard Biener <rguenther@suse.de>
22370 PR tree-optimization/81502
22371 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22373 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22375 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22376 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22377 for BIT_FIELD_REF args.
22378 * fold-const.c (make_bit_field_ref): Likewise.
22379 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22381 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22384 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22385 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22386 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22387 Or it into SANITIZER_UNDEFINED.
22388 * ubsan.c: Include gimple-fold.h and varasm.h.
22389 (ubsan_expand_ptr_ifn): New function.
22390 (instrument_pointer_overflow): New function.
22391 (maybe_instrument_pointer_overflow): New function.
22392 (instrument_object_size): Formatting fix.
22393 (pass_ubsan::execute): Call instrument_pointer_overflow
22394 and maybe_instrument_pointer_overflow.
22395 * internal-fn.c (expand_UBSAN_PTR): New function.
22396 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22397 * sanitizer.def (__ubsan_handle_pointer_overflow,
22398 __ubsan_handle_pointer_overflow_abort): New builtins.
22399 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22400 * internal-fn.def (UBSAN_PTR): New internal function.
22401 * opts.c (sanitizer_opts): Add pointer-overflow.
22402 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22403 * fold-const.c (build_range_check): Compute pointer range check in
22404 integral type if pointer arithmetics would be needed. Formatting
22407 2017-07-28 Martin Liska <mliska@suse.cz>
22410 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22411 parameters that are of a variable-length.
22413 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22415 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22417 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22418 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22419 (CRT_CALL_STATIC_FUNCTION): Likewise.
22420 (ASM_DEFAULT_SPEC): New define.
22421 (ASM_SPEC32): Likewise.
22422 (ASM_SPEC64): Likewise.
22423 (ASM_SPEC_COMMON): Likewise.
22424 (ASM_SPEC): Likewise.
22425 (INVALID_64BIT): Likewise.
22426 (LINK_OS_DEFAULT_SPEC): Likewise.
22427 (LINK_OS_SPEC32): Likewise.
22428 (LINK_OS_SPEC64): Likewise.
22429 (POWERPC_LINUX): Likewise.
22430 (PTRDIFF_TYPE): Likewise.
22431 (RESTORE_FP_PREFIX): Likewise.
22432 (RESTORE_FP_SUFFIX): Likewise.
22433 (SAVE_FP_PREFIX): Likewise.
22434 (SAVE_FP_SUFFIX): Likewise.
22435 (SIZE_TYPE): Likewise.
22436 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22437 (TARGET_64BIT): Likewise.
22438 (TARGET_64BIT): Likewise.
22439 (TARGET_AIX): Likewise.
22440 (WCHAR_TYPE_SIZE): Likewise.
22441 (WCHAR_TYPE): Undefine.
22442 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22443 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22444 (CPP_OS_RTEMS_SPEC): Delete.
22445 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22446 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22448 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22450 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22452 PR tree-optimization/81578
22453 * tree-parloops.c (build_new_reduction): Bail out if
22454 reduction_code isn't one of the standard OpenMP reductions.
22455 Move the details printing after that decision.
22457 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22459 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22460 related to reload_in_progress.
22461 (splat_input_operand): Likewise.
22462 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22464 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22466 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22467 (TARGET_INSTANTIATE_DECLS): Likewise.
22468 (legitimate_indexed_address_p): Delete reload_in_progress code.
22469 (rs6000_debug_legitimate_address_p): Likewise.
22470 (rs6000_eliminate_indexed_memrefs): Likewise.
22471 (rs6000_emit_le_vsx_store): Likewise.
22472 (rs6000_emit_move_si_sf_subreg): Likewise.
22473 (rs6000_emit_move): Likewise.
22474 (register_to_reg_type): Likewise.
22475 (rs6000_pre_atomic_barrier): Likewise.
22476 (rs6000_machopic_legitimize_pic_address): Likewise.
22477 (rs6000_allocate_stack_temp): Likewise.
22478 (rs6000_address_for_fpconvert): Likewise.
22479 (rs6000_address_for_altivec): Likewise.
22480 (rs6000_secondary_memory_needed_rtx): Delete function.
22481 (rs6000_check_sdmode): Likewise.
22482 (rs6000_alloc_sdmode_stack_slot): Likewise.
22483 (rs6000_instantiate_decls): Likewise.
22484 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22485 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22486 Delete reload_in_progress.
22487 (*vec_reload_and_plus_<mptrsize>): Likewise.
22488 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22489 (vsx_div_v2di): Likewise.
22490 (vsx_udiv_v2di): Likewise.
22492 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22494 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22495 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22496 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22497 (rs6000_debug_reg_global): Delete print of LRA status.
22498 (rs6000_option_override_internal): Delete dead LRA related code.
22499 (rs6000_lra_p): Delete function.
22500 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22502 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22504 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22505 * config/riscv/rtems.h: New file.
22507 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22508 Sudakshina Das <sudi.das@arm.com>
22510 * config/aarch64/aarch64.md
22511 (define_split for and<mode>3nr_compare): Move
22512 non aarch64_logical_operand to a register.
22513 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22514 register immediate operand to a register.
22515 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22517 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22519 PR middle-end/81564
22520 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22522 2017-07-27 Richard Biener <rguenther@suse.de>
22524 PR tree-optimization/81573
22525 PR tree-optimization/81494
22526 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22527 multi defuse cycle case.
22529 2017-07-27 Richard Biener <rguenther@suse.de>
22531 PR tree-optimization/81571
22532 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22535 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22537 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22538 earlier and only if MASK_FPU is set. Adjust formatting.
22540 2017-07-27 Martin Liska <mliska@suse.cz>
22542 * opt-functions.awk: Add validation of value of Init.
22543 * optc-gen.awk: Pass new argument.
22545 2017-07-27 Martin Liska <mliska@suse.cz>
22547 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22548 Fix wrong condition.
22550 2017-07-27 Martin Liska <mliska@suse.cz>
22552 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22553 BBs and edges seen by autoFDO.
22555 2017-07-27 Richard Biener <rguenther@suse.de>
22557 PR tree-optimization/81502
22558 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22559 with incompatible but same sized type.
22560 (execute_update_addresses_taken): Likewise.
22562 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22564 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22565 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22567 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22570 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22571 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22572 Change s_operand to memory_operand.
22574 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22576 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22577 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22578 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22579 Emit instructions rather than returning an expression. Handle TFmode
22580 and KFmode by casting to TImode.
22581 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22582 (rs6000_emit_le_vsx_store): Likewise.
22583 * config/rs6000/vsx.md (VSX_TI): New iterator.
22584 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22585 (*vsx_le_undo_permute_<mode>): Likewise.
22586 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22587 emit the split sequence.
22588 (*vsx_le_perm_store_<mode>): Likewise.
22590 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22592 PR tree-optimization/81555
22593 PR tree-optimization/81556
22594 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22595 if true, force CHANGED for the recursive invocation.
22596 (reassociate_bb): Remember original length of ops array, pass
22597 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22599 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22600 attributes for noipa attribute. For naked attribute use
22601 lookup_attribute first before lookup_attribute_spec.
22602 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22604 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22605 for functions with noipa attribute.
22606 (cgraph_externally_visible_p): Return true for functions with noipa
22608 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22609 for functions with noipa attribute.
22610 * doc/extend.texi: Document noipa function attribute.
22611 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22612 also for functions with noipa attribute.
22613 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22615 2017-07-26 Andrew Pinski <apinski@cavium.com>
22617 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22618 vec_unalign_load_cost and vec_unalign_store_cost.
22620 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22622 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22623 -mvsx-small-integer option.
22624 (ISA_3_0_MASKS_IEEE): Likewise.
22625 (OTHER_VSX_VECTOR_MASKS): Likewise.
22626 (POWERPC_MASKS): Likewise.
22627 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22628 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22629 code, only testing for DImode being allowed in non-VSX floating
22631 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22632 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22634 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22635 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22636 TARGET_P8_VECTOR test.
22637 (rs6000_secondary_reload_simple_move): Likewise.
22638 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22639 since TARGET_P9_VECTOR was already tested.
22640 (rs6000_opt_masks): Remove -mvsx-small-integer.
22641 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22642 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22644 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22645 test for TARGET_VEXTRACTUB was used, and that uses
22647 (p9 extract splitter): Likewise.
22648 (vsx_extract_<mode>_di_p9): Likewise.
22649 (vsx_extract_<mode>_store_p9): Likewise.
22650 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22651 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22652 the elimination of TARGET_VSX_SMALL_INTEGER.
22653 (vsx_extract_<mode>_p8): Likewise.
22654 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22655 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22656 (vsx_set_<mode>_p9): Likewise.
22657 (vsx_set_v4sf_p9): Likewise.
22658 (vsx_set_v4sf_p9_zero): Likewise.
22659 (vsx_insert_extract_v4sf_p9): Likewise.
22660 (vsx_insert_extract_v4sf_p9_2): Likewise.
22661 * config/rs6000/rs6000.md (sign extend splitter): Change
22662 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22663 (floatsi<mode>2_lfiwax_mem): Likewise.
22664 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22665 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22666 since a test for TARGET_P9_VECTOR was used.
22667 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22668 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22669 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22670 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22671 TARGET_P8_VECTOR test.
22672 (fix_trunc<mode>si2_stfiwx): Likewise.
22673 (fix_trunc<mode>si2_internal): Likewise.
22674 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22675 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22677 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22678 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22679 TARGET_P8_VECTOR test.
22680 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22681 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22682 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22684 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22685 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22686 since a test for TARGET_P9_VECTOR was used.
22687 (splitter for loading small constants): Likewise.
22689 2017-07-26 Andrew Pinski <apinski@cavium.com>
22691 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22694 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22697 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22698 (fp_valid_at): Likewise.
22700 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22702 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22703 (qdf24xx_addrcost_table): Likewise.
22704 (cortexa57_tunings): Update to use generic_branch_cost.
22705 (cortexa72_tunings): Likewise.
22706 (cortexa73_tunings): Likewise.
22707 (qdf24xx_tunings): Likewise.
22709 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22711 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22712 (thunderx2t99_branch_cost): Likewise.
22713 (cortexa35_tunings): Update to use generic_branch_cost.
22714 (cortexa53_tunings): Likewise.
22715 (cortexa57_tunings): Likewise.
22716 (cortexa72_tunings): Likewise.
22717 (cortexa73_tunings): Likewise.
22718 (thunderx2t99_tunings): Likewise.
22720 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22722 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22723 (sparc_option_override): Honour MASK_FSMULD.
22724 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22725 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22726 * config/sparc/sparc.opt (mfsmuld): New option.
22727 * doc/invoke.texi (mfsmuld): Document option.
22729 2017-07-26 Marek Polacek <polacek@redhat.com>
22731 PR middle-end/70992
22732 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22734 2017-07-26 Richard Biener <rguenther@suse.de>
22736 * gimple-match-head.c (do_valueize): Return OP if valueize
22738 (get_def): New helper to get at the def stmt of a SSA name
22739 if valueize allows.
22740 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22741 do_valueize to get at the def stmt.
22742 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22744 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22746 PR middle-end/46932
22747 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22749 2017-07-26 Martin Liska <mliska@suse.cz>
22752 * function.c (expand_function_start): Make expansion of
22753 nonlocal_goto_save_area after parm_birth_insn.
22755 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22757 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22758 from all CPU target flags enable members.
22760 2017-07-26 Richard Biener <rguenther@suse.de>
22762 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22763 (decision_tree::gen): Make 'type' const.
22764 (write_predicate): Likewise.
22766 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22768 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22769 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22770 (rs6000_option_override_internal): Likewise.
22771 (rs6000_expand_vector_set): Likewise.
22772 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22773 (TARGET_UPPER_REGS_SF): Likewise.
22774 (TARGET_UPPER_REGS_DI): Likewise.
22775 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22776 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22777 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22778 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22779 (Splitters for DI constants in Altivec registers): Likewise.
22780 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22781 (vsx_set_v4sf_p9): Likewise.
22782 (vsx_set_v4sf_p9_zero): Likewise.
22783 (vsx_insert_extract_v4sf_p9): Likewise.
22784 (vsx_insert_extract_v4sf_p9_2): Likewise.
22786 2017-07-25 Carl Love <cel@us.ibm.com>
22788 * doc/extend.texi: Update the built-in documentation file for the
22789 existing built-in functions
22790 vector signed char vec_cnttz (vector signed char);
22791 vector unsigned char vec_cnttz (vector unsigned char);
22792 vector signed short vec_cnttz (vector signed short);
22793 vector unsigned short vec_cnttz (vector unsigned short);
22794 vector signed int vec_cnttz (vector signed int);
22795 vector unsigned int vec_cnttz (vector unsigned int);
22796 vector signed long long vec_cnttz (vector signed long long);
22797 vector unsigned long long vec_cnttz (vector unsigned long long);
22799 2017-07-25 Andrew Pinski <apinski@cavium.com>
22801 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22802 accesses where the use is for the first operand of a BIT_INSERT.
22804 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22807 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22808 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22810 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22812 * config/i386/gstabs.h: Delete.
22813 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22815 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22817 * config/i386/i386.c (ix86_decompose_address): Do not check for
22818 register RTX when looking at index_reg or base_reg.
22819 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22821 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22823 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22824 to update EH info here.
22826 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22828 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22830 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22832 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22834 2017-07-25 Torsten Duwe <duwe@suse.de>
22836 * common.opt: Introduce -fpatchable-function-entry
22837 command line option, and its variables function_entry_patch_area_size
22838 and function_entry_patch_area_start.
22839 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22840 including a two-value parser.
22841 * target.def (print_patchable_function_entry): New target hook.
22842 * targhooks.h (default_print_patchable_function_entry): New function.
22843 * targhooks.c (default_print_patchable_function_entry): Likewise.
22844 * toplev.c (process_options): Switch off IPA-RA if
22845 patchable function entries are being generated.
22846 * varasm.c (assemble_start_function): Look at the
22847 patchable-function-entry command line switch and current
22848 function attributes and maybe generate NOP instructions by
22849 calling the print_patchable_function_entry hook.
22850 * doc/extend.texi: Document patchable_function_entry attribute.
22851 * doc/invoke.texi: Document -fpatchable_function_entry
22852 command line option.
22853 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22855 * doc/tm.texi: Re-generate.
22857 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22860 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22861 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22863 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22865 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22867 2017-07-25 Richard Biener <rguenther@suse.de>
22869 PR tree-optimization/81455
22870 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22871 not walk in cycles when looking for guards.
22873 2017-07-25 Richard Biener <rguenther@suse.de>
22875 PR tree-optimization/81529
22876 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22877 when optimizing backedge uses.
22879 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22881 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22883 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22884 to dl_section_ref. On AIX, append an expression to subtract
22885 the size of the section length to dl_section_ref.
22887 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22889 * configure.ac: If any of the config.* scripts fail, exit 1.
22890 * configure: Regenerate.
22892 2017-07-25 Richard Biener <rguenther@suse.de>
22894 PR middle-end/81546
22895 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22896 of immediate uses, be more verbose on errors.
22898 2017-07-25 Richard Biener <rguenther@suse.de>
22900 PR tree-optimization/81510
22901 * tree-vect-loop.c (vect_is_simple_reduction): When the
22902 reduction stmt is not inside the loop bail out.
22904 2017-07-25 Richard Biener <rguenther@suse.de>
22906 PR tree-optimization/81303
22907 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22908 profitability check against LOOP_VINFO_NITERSM1.
22910 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22912 * domwalk.c (cmp_bb_postorder): Simplify.
22913 (sort_bbs_postorder): New function. Use it...
22914 (dom_walker::walk): ...here to optimize common cases.
22916 2017-07-25 Martin Liska <mliska@suse.cz>
22919 * ipa-visibility.c (function_and_variable_visibility): Make the
22920 redirection just on target that supports aliasing.
22921 Fix GNU coding style.
22923 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22926 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22928 * config/arm/rtems.h: Likewise.
22929 * config/bfin/rtems.h: Likewise.
22930 * config/i386/rtemself.h: Likewise.
22931 * config/lm32/rtems.h: Likewise.
22932 * config/m32c/rtems.h: Likewise.
22933 * config/m68k/rtemself.h: Likewise.
22934 * config/microblaze/rtems.h: Likewise.
22935 * config/mips/rtems.h: Likewise.
22936 * config/moxie/rtems.h: Likewise.
22937 * config/nios2/rtems.h: Likewise.
22938 * config/powerpcspe/rtems.h: Likewise.
22939 * config/rs6000/rtems.h: Likewise.
22940 * config/rtems.h: Likewise.
22941 * config/sh/rtems.h: Likewise.
22942 * config/sh/rtemself.h: Likewise.
22943 * config/sparc/rtemself.h: Likewise.
22945 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22948 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22949 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22950 * tree-ssa-structalias.c (alias_get_name): Same.
22952 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22955 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22956 instructions if no du chain is found.
22958 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22960 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22962 2017-07-25 Richard Biener <rguenther@suse.de>
22964 PR middle-end/81505
22965 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22968 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22970 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22971 upper-regs options.
22972 (ISA_2_7_MASKS_SERVER): Likewise.
22973 (ISA_3_0_MASKS_IEEE): Likewise.
22974 (OTHER_P8_VECTOR_MASKS): Likewise.
22975 (OTHER_VSX_VECTOR_MASKS): Likewise.
22976 (POWERPC_MASKS): Likewise.
22977 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22978 duplicate list of options.
22979 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22980 explicit -mupper-regs options.
22981 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22982 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22983 alias for -mupper-regs-df.
22984 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22985 (rs6000_init_hard_regno_mode_ok): Likewise.
22986 (rs6000_option_override_internal): Likewise.
22987 (rs6000_opt_masks): Likewise.
22988 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22989 options in terms of whether -mvsx or -mpower8-vector was used.
22990 (TARGET_UPPER_REGS_DI): Likewise.
22991 (TARGET_UPPER_REGS_SF): Likewise.
22992 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22993 -mupper-regs-* options.
22995 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22997 * passes.c (emergency_dump_function): Print some empty lines and a
22998 header before the RTL dump.
23000 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23002 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23004 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23007 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23008 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23010 2017-07-24 Carl Love <cel@us.ibm.com>
23012 * config/rs6000/rs6000-c.c: Add support for built-in functions
23013 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23014 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23015 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23016 vec_extract_fp_from_shortl): Add defines for the two builtins.
23017 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23018 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23020 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23021 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23022 * doc/extend.texi: Update the built-in documentation file for the
23023 new built-in function.
23025 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23028 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23030 * doc/generic.texi: Likewise.
23031 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23032 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23034 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23036 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23037 (aarch64_mls_elt_merge<mode>): Likewise.
23039 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23041 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23042 having __cxa_atexit.
23044 2017-07-23 Michael Collison <michael.collison@arm.com>
23046 * config/arm/arm.c (arm_option_override): Deprecate
23047 use of -mstructure-size-boundary.
23048 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23049 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23051 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23054 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23055 Reduce cost estimate for direct moves.
23057 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23060 * config/i386/i386.c (ix86_option_override_internal): Disable
23061 BMI, BMI2 and TBM instructions for -m16.
23063 2017-07-21 Carl Love <cel@us.ibm.com>
23065 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23066 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23067 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23068 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23069 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23070 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23071 VMULOSW): New enum "unspec" values.
23072 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23073 altivec_vmulosw): New patterns.
23074 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23075 VMULOSW): Add definitions.
23077 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23079 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23080 (qdf24xx): Likewise.
23081 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23082 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23083 (AARCH64_FL_V8_1): Renumber.
23084 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23085 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23086 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23087 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23088 rdma to feature modifiers list.
23090 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23092 PR middle-end/56727
23093 * ipa-visibility (function_and_variable_visibility): Convert
23094 recursive PLT call to direct call if appropriate.
23096 2017-07-21 Andrew Pinski <apinski@cavium.com>
23098 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23099 operand 1 to see if the types precision matches.
23100 * fold-const.c (operand_equal_p): Likewise.
23102 2017-07-21 Richard Biener <rguenther@suse.de>
23104 PR tree-optimization/81303
23105 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23106 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23107 (vect_peeling_hash_get_lowest_cost): Adjust.
23108 (vect_enhance_data_refs_alignment): Likewise. Use
23109 vect_get_peeling_costs_all_drs to compute the penalty for no
23110 peeling to match up costs.
23112 2017-07-21 Richard Biener <rguenther@suse.de>
23114 PR tree-optimization/81500
23115 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23116 we didn't identify a reduction path.
23118 2017-07-21 Tom de Vries <tom@codesourcery.com>
23119 Cesar Philippidis <cesar@codesourcery.com>
23121 PR gcov-profile/81442
23122 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23125 2017-07-21 Tom de Vries <tom@codesourcery.com>
23128 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23130 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23131 nvptx_override_options_after_change.
23133 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23135 * dwarf2out.c (output_file_names): Avoid double testing for
23136 dwarf_version >= 5.
23138 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23140 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23142 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23144 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23146 (try_crossjump_to_edge): Do not punt on partitioned functions.
23148 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23150 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23151 Put all BBs reachable only via paths crossing cold region to cold
23153 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23155 2016-07-21 Richard Biener <rguenther@suse.de>
23157 PR tree-optimization/81303
23158 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23159 into account prologue and epilogue iterations when raising
23160 min_profitable_iters to sth at least covering one vector iteration.
23162 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23164 * config/arm/arm.c (arm_test_cpu_arch_dat):
23167 2017-07-20 Nathan Sidwell <nathan@acm.org>
23169 Remove TYPE_METHODS.
23170 * tree.h (TYPE_METHODS): Delete.
23171 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23172 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23173 (dbxout_type_methods): Scan TYPE_FIELDS.
23174 (dbxout_type): Don't check TYPE_METHODS here.
23175 * function.c (use_register_for_decl): Always ignore register for
23176 class types when not optimizing.
23177 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23178 * tree.c (free_lang_data_in_type): Stitch out member functions and
23179 templates from TYPE_FIELDS.
23180 (build_distinct_type_copy, verify_type_variant,
23181 verify_type): Member fns are on TYPE_FIELDS.
23182 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23183 * tree-pretty-print.c (dump_generic_node): Likewise.
23185 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23188 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23189 V2TImode and V4TImode.
23190 (ix86_expand_vector_extract): Likewise.
23191 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23192 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23193 (ssescalarmode): Handle V4TImode and V2TImode.
23194 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23195 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23196 (VEXTRACTI128_MODE): New mode iterator.
23197 (splitter for *vec_extractv?ti first element): New.
23198 (VEC_INIT_MODE): New mode iterator.
23199 (vec_init<mode>): Consolidate 3 expanders into one using
23200 VEC_INIT_MODE mode iterator.
23202 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23204 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23205 non_spilled_static_chain_regno_p.
23207 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23209 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23211 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23213 * bb-reorder.c (connect_traces): Allow copying of blocks within
23216 2017-07-20 Richard Biener <rguenther@suse.de>
23218 * gimple.h (gimple_phi_result): Add gphi * overload.
23219 (gimple_phi_result_ptr): Likewise.
23220 (gimple_phi_arg): Likewise. Adjust index assert to only
23221 allow actual argument accesses rather than all slots available
23223 (gimple_phi_arg_def): Add gphi * overload.
23224 * tree-phinodes.c (make_phi_node): Initialize only actual
23226 (resize_phi_node): Clear memory not covered by old node,
23227 do not initialize excess argument slots.
23228 (reserve_phi_args_for_new_edge): Initialize new argument slot
23231 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23233 PR tree-optimization/81388
23235 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23237 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23238 by removing computation of may_be_zero.
23240 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23241 Tom de Vries <tom@codesourcery.com>
23243 PR middle-end/81030
23244 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23245 when gimple level profile disagrees with what RTL expander did.
23247 2017-07-20 Richard Biener <rguenther@suse.de>
23249 PR tree-optimization/61171
23250 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23251 (vect_analyze_stmt): Add slp instance parameter.
23252 (vectorizable_reduction): Likewise.
23253 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23254 (vect_is_simple_reduction): Deal with chains not detected
23255 as SLP reduction chain, specifically not properly associated
23256 chains containing a mix of plus/minus.
23257 (get_reduction_op): Remove.
23258 (get_initial_defs_for_reduction): Simplify, pass in whether
23259 this is a reduction chain, pass in the SLP node for the PHIs.
23260 (vect_create_epilog_for_reduction): Get the SLP instance as
23262 (vectorizable_reduction): Get the SLP instance as arg.
23263 During analysis remember the SLP node with the PHIs in the
23264 instance. Simplify getting at the vectorized reduction PHIs.
23265 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23266 through SLP instance.
23267 (vect_slp_analyze_operations): Likewise.
23268 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23269 (vect_transform_stmt): Likewise.
23271 2017-07-20 Tom de Vries <tom@codesourcery.com>
23273 PR tree-optimization/81489
23274 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23275 read of phi arg location to before loop that modifies phi.
23277 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23279 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23282 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23284 PR middle-end/81331
23285 * except.c (execute): Fix ordering issue.
23287 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23289 PR rtl-optimization/81423
23290 * combine.c (make_compound_operation_int): Don't try to optimize
23291 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23293 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23295 PR rtl-optimization/81423
23296 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23297 with a constant that is -1 in the truncated to mode.
23299 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23301 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23302 (determine_unlikely_bbs): ... here.
23303 * predict.h (propagate_unlikely_bbs_forward): Declare.
23304 * cfgexpand.c (pass_expand::execute): Use it.
23305 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23307 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23308 propagate_unlikely_bbs_forward.
23310 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23312 PR middle-end/81331
23313 * except.c (maybe_add_nop_after_section_switch): New function.
23316 2017-07-19 Tom de Vries <tom@codesourcery.com>
23318 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23320 2017-07-19 Tom de Vries <tom@codesourcery.com>
23322 * gimple.h (gimple_phi_arg): Make assert more strict.
23324 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23326 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23327 * config/rs6000/mmintrin.h: New file.
23328 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23330 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23332 PR tree-optimization/81346
23333 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23335 2017-07-19 Tom de Vries <tom@codesourcery.com>
23337 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23339 2017-07-19 Tom de Vries <tom@codesourcery.com>
23341 * config/nvptx/nvptx-modes.def: Add V2DImode.
23342 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23343 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23344 (nvptx_output_mov_insn): Handle lack of mov.b128.
23345 (nvptx_print_operand): Handle 'H' and 'L' codes.
23346 (nvptx_vector_mode_supported): Allow V2DImode.
23347 (nvptx_preferred_simd_mode): New function.
23348 (nvptx_data_alignment): New function.
23349 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23350 nvptx_preferred_simd_mode.
23351 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23353 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23355 2017-07-19 Tom de Vries <tom@codesourcery.com>
23357 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23358 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23359 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23360 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23361 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23362 (mov<VECIM>_insn): New define_insn.
23363 (define_expand "mov<VECIM>): New define_expand.
23365 2017-07-19 Tom de Vries <tom@codesourcery.com>
23367 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23369 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23371 PR tree-optimization/81346
23372 * fold-const.h (fold_div_compare, range_check_type): Declare.
23373 * fold-const.c (range_check_type): New function.
23374 (build_range_check): Use range_check_type.
23375 (fold_div_compare): No longer static, rewritten into
23376 a match.pd helper function.
23377 (fold_comparison): Don't call fold_div_compare here.
23378 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23379 as helper function.
23381 2017-07-19 Nathan Sidwell <nathan@acm.org>
23383 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23384 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23385 * tree.c (find_decls_types_r, verify_type): Use
23386 TYPE_{MIN,MAX}_VALUE_RAW.
23387 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23388 (hash_tree): Likewise.
23389 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23391 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23394 2017-07-18 Tom de Vries <tom@codesourcery.com>
23396 PR middle-end/81464
23397 * omp-expand.c (expand_omp_for_static_chunk): Handle
23398 equal-argument loop exit phi.
23400 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23403 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23404 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23405 operand 2 predicate.
23406 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23407 operand 2 predicate.
23408 (ror,rol -> rorx splitters): Use const_int_operand as
23409 operand 2 predicate.
23411 2017-06-18 Richard Biener <rguenther@suse.de>
23413 PR tree-optimization/81410
23414 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23415 the gap in the ! slp_perm SLP case after each group.
23417 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23419 PR middle-end/81463
23420 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23423 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23425 PR middle-end/81462
23426 * predict.c (set_even_probabilities): Cleanup; do not affect
23427 probabilities that are already known.
23428 (combine_predictions_for_bb): Call even when count is set.
23430 2017-07-18 Nathan Sidwell <nathan@acm.org>
23432 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23435 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23438 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23439 optimization for loop niter analysis.
23441 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23444 * config/avr/avr.c (avr_optimize_casesi): Don't use
23445 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23447 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23449 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23450 body_cost_vec from _vect_peel_extended_info.
23451 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23452 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23455 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23457 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23459 2017-07-18 Richard Biener <rguenther@suse.de>
23461 PR tree-optimization/80620
23462 PR tree-optimization/81403
23463 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23464 info when re-using a VN table entry.
23466 2017-07-18 Richard Biener <rguenther@suse.de>
23468 PR tree-optimization/81418
23469 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23470 vectype_in. Verify that with lane-reducing reduction operations
23471 we have a single def-use cycle.
23473 2017-07-17 Carl Love <cel@us.ibm.com>
23475 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23477 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23478 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23479 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23480 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23481 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23482 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23483 VMULOSW): New enum "unspec" values.
23484 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23485 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23486 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23487 altivec_vmulosw): New patterns.
23488 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23489 VMULOSW): Add definitions.
23491 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23493 * config/alpha/alpha.c: Include predict.h.
23495 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23497 * tree-vrp.c (compare_assert_loc): Fix comparison function
23498 to return predictable results.
23500 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23502 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23504 (subdi3): Likewise.
23505 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23506 * doc/invoke.texi (mexpand-adddi): Update text.
23508 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23510 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23511 that also clobbers the CC register. The old expand code is moved
23513 (*arc_clzsi2): ... here.
23514 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23515 the CC register. The old expand code is moved to ...
23516 (arc_ctzsi2): ... here.
23518 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23520 * config/arc/arc.opt (mindexed-loads): Use initial value
23521 TARGET_INDEXED_LOADS_DEFAULT.
23522 (mauto-modify-reg): Use initial value
23523 TARGET_AUTO_MODIFY_REG_DEFAULT.
23524 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23525 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23526 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23527 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23529 2017-07-17 Martin Liska <mliska@suse.cz>
23532 * opts.c (finish_options): Do not allow -fgnu-tm
23533 w/ -fsanitize={kernel-,}address. Say sorry.
23535 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23538 * tree-loop-distribution.c (classify_partition): Only assert on
23539 numer of iterations.
23540 (merge_dep_scc_partitions): Delete prameter. Update function call.
23541 (distribute_loop): Remove code handling loop with unknown niters.
23542 (pass_loop_distribution::execute): Skip loop with unknown niters.
23544 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23547 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23548 function sort_partitions_by_post_order.
23550 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23552 PR tree-optimization/81374
23553 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23554 the max index of basic blocks, rather than number of basic blocks.
23556 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23558 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23560 (arc_legitimate_pic_operand_p): Likewise.
23561 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23563 (arc_needs_pcl_p): Likewise.
23564 (arc_legitimate_pc_offset_p): Likewise.
23565 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23566 function is also used in constrains.md.
23567 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23568 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23569 PLUS. Only return true/false in known cases, otherwise assert.
23570 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23571 is already called in arc_legitimate_constant_p.
23572 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23574 (LEGITIMATE_PIC_OPERAND_P): Use
23575 arc_raw_symbolic_reference_mentioned_p function.
23576 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23581 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23582 Andrew Burgess <andrew.burgess@embecosm.com>
23584 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23585 (arc_return_address_register): New function.
23586 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23587 (arc_handle_fndecl_attribute): Add naked attribute.
23588 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23589 (TARGET_WARN_FUNC_RETURN): Likewise.
23590 (arc_allocate_stack_slots_for_args): New function.
23591 (arc_warn_func_return): Likewise.
23592 (machine_function): Change type fn_type.
23593 (arc_compute_function_type): Consider new naked function type,
23594 change function return type.
23595 (arc_must_save_register): Adapt to handle new
23596 arc_compute_function_type's return type.
23597 (arc_expand_prologue): Likewise.
23598 (arc_expand_epilogue): Likewise.
23599 (arc_return_address_regs): Delete.
23600 (arc_return_address_register): New function.
23601 (arc_epilogue_uses): Use above function.
23602 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23603 (arc_function_type): Change encoding, add naked type.
23604 (ARC_INTERRUPT_P): Change to handle the new encoding.
23605 (ARC_FAST_INTERRUPT_P): Likewise.
23606 (ARC_NORMAL_P): Define.
23607 (ARC_NAKED_P): Likewise.
23608 (arc_compute_function_type): Delete prototype.
23609 * config/arc/arc.md (in_ret_delay_slot): Use
23610 arc_return_address_register function.
23611 (simple_return): Likewise.
23612 (p_return_i): Likewise.
23614 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23616 PR tree-optimization/81428
23617 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23618 can't be built for those types.
23620 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23622 Remove stuff dead since r239246.
23624 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23625 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23626 (avr_inform_devices): Remove dead stuff.
23628 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23630 * config/arm/arm_neon.h: Fix softp typo.
23632 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23634 PR tree-optimization/81365
23635 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23636 aggregate moves onto bb predecessor edges, make sure there are no
23637 loads that could alias the lhs in between the start of bb and the
23640 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23643 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23644 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23645 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23647 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23649 PR tree-optimization/81396
23650 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23651 (init_symbolic_number): Initialize it to 1.
23652 (perform_symbolic_merge): Add n_ops from both operands into the new
23654 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23655 without base_addr as useless if they need more than one operation.
23656 (bswap_replace): Handle !bswap case for NULL base_addr.
23658 2017-07-17 Tom de Vries <tom@codesourcery.com>
23661 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23664 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23666 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23667 conditional builtin define __FIX_LEON3FT_B2BST.
23669 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23671 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23672 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23675 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23677 PR rtl-optimization/81424
23678 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23679 to remove potential trapping from operands if -fnon-call-exceptions.
23681 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23683 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23684 profile_proability for scalling.
23685 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23687 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23689 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23691 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23693 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23694 fixpoint arithmetics.
23696 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23698 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23699 fixpoint arithmetics.
23701 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23703 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23704 fixpoint arithmetics.
23706 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23708 * profile-count.h (profile_probability::from_reg_br_prob_note,
23709 profile_probability::to_reg_br_prob_note): New functions.
23710 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23711 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23712 * predict.c (probability_reliable_p): Update.
23713 (edge_probability_reliable_p): Update.
23714 (br_prob_note_reliable_p): Update.
23715 (invert_br_probabilities): Update.
23716 (add_reg_br_prob_note): New function.
23717 (combine_predictions_for_insn): Update.
23718 * asan.c (asan_clear_shadow): Update.
23719 * cfgbuild.c (compute_outgoing_frequencies): Update.
23720 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23721 (update_br_prob_note): Update.
23722 (rtl_verify_edges): Update.
23723 (purge_dead_edges): Update.
23724 (fixup_reorder_chain): Update.
23725 * emit-rtl.c (try_split): Update.
23726 * ifcvt.c (cond_exec_process_insns): Update.
23727 (cond_exec_process_if_block): Update.
23728 (dead_or_predicable): Update.
23729 * internal-fn.c (expand_addsub_overflow): Update.
23730 (expand_neg_overflow): Update.
23731 (expand_mul_overflow): Update.
23732 * loop-doloop.c (doloop_modify): Update.
23733 * loop-unroll.c (compare_and_jump_seq): Update.
23734 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23735 * predict.h: Update.
23736 * reorg.c (mostly_true_jump): Update.
23738 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23739 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23740 * config/arc/arc.c: (emit_unlikely_jump): Update.
23741 * config/arm/arm.c: (emit_unlikely_jump): Update.
23742 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23743 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23744 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23745 (ix86_print_operand): Update.
23746 (ix86_split_fp_branch): Update.
23747 (predict_jump): Update.
23748 * config/ia64/ia64.c (ia64_print_operand): Update.
23749 * config/mmix/mmix.c (mmix_print_operand): Update.
23750 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23751 (rs6000_expand_split_stack_prologue): Update.
23752 * config/rs6000/rs6000.c: Update.
23753 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23754 (s390_expand_vec_movstr): Update.
23755 (s390_expand_cs_tdsi): Update.
23756 (s390_expand_split_stack_prologue): Update.
23757 * config/sh/sh.c (sh_print_operand): Update.
23758 (expand_cbranchsi4): Update.
23759 (expand_cbranchdi4): Update.
23760 * config/sparc/sparc.c (output_v9branch): Update.
23761 * config/spu/spu.c (get_branch_target): Update.
23762 (ea_load_store_inline): Update.
23763 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23764 * config/tilepro/tilepro.c: Update.
23766 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23768 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23769 (gimplify_save_expr): Likewise.
23771 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23773 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23775 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23777 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23779 * timevar.def (TV_IPA_FNSUMMARY): Define.
23781 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23783 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23784 to back store errata sensitive sequence from being generated.
23785 (sqrtdf2_fix): Likewise.
23787 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23789 * tree-ssa-threadupdate.c (compute_path_counts,
23790 update_joiner_offpath_counts): Use profile_probability.
23792 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23795 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23797 * config/arm/arm-c.c (arm_cpu_builtins): Define
23798 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23800 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23802 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23803 array entries to represent __ieee128 versions of the
23804 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23805 scalar_extract_sig, and scalar_insert_exp built-in functions.
23806 (altivec_resolve_overloaded_builtin): Add special case handling
23807 for the __builtin_scalar_insert_exp function, as represented by
23808 the P9V_BUILTIN_VEC_VSIEDP constant.
23809 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23810 exponent support for __ieee128 argument.
23811 (VSESQP): Add scalar extract signature support for __ieee128
23813 (VSTDCNQP): Add scalar test negative support for __ieee128
23815 (VSIEQP): Add scalar insert exponent support for __int128 argument
23816 with __ieee128 result.
23817 (VSIEQPF): Add scalar insert exponent support for __ieee128
23818 argument with __ieee128 result.
23819 (VSTDCQP): Add scalar test data class support for __ieee128
23821 (VSTDCNQP): Add overload support for scalar test negative with
23822 __ieee128 argument.
23823 (VSTDCQP): Add overload support for scalar test data class
23824 __ieee128 argument.
23825 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23826 UNSPEC_VSX_SXSIGDP.
23827 (UNSPEC_VSX_SIEXPQP): New constant.
23828 (xsxexpqp): New insn for VSX scalar extract exponent quad
23830 (xsxsigqp): New insn for VSX scalar extract significand quad
23832 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23833 precision with floating point argument.
23834 (xststdcqp): New expand for VSX scalar test data class quad
23836 (xststdcnegqp): New expand for VSX scalar test negative quad
23838 (xststdcqp): New insn to match expansions for VSX scalar test data
23839 class quad precision and VSX scalar test negative quad precision.
23840 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23841 special case operand checking to enforce that second operand of
23842 VSX scalar test data class with quad precision argument is a 7-bit
23844 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23845 prototypes and descriptions of __ieee128 versions of
23846 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23847 scalar_test_data_class, and scalar_test_neg built-in functions.
23849 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23851 PR tree-optimization/81162
23852 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23853 replace a negate with an add.
23855 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23857 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23859 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23861 * config/arm/arm-c.c (arm_cpu_builtins): Define
23862 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23864 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23866 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23867 (armv8-r): Set ARM Cortex-R52 as default CPU.
23868 * config/arm/arm-tables.opt: Regenerate.
23869 * config/arm/arm-tune.md: Regenerate.
23870 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23872 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23873 extension for -mcpu=cortex-r52.
23875 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23877 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23878 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23879 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23880 (fp-armv8): Define it as FP_ARMv8 only.
23881 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23882 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23884 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23885 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23886 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23887 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23888 than TARGET_FPU_ARMV8.
23889 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23890 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23891 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23893 * config/arm/neon.md (neon_vrint): Likewise.
23894 (neon_vcvt): Likewise.
23895 (neon_<fmaxmin_op><mode>): Likewise.
23896 (<fmaxmin><mode>3): Likewise.
23897 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23898 * config/arm/predicates.md (arm_cond_move_operator): Check against
23899 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23901 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23903 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23904 to top of function.
23906 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23908 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23909 loop in comment with memset.
23911 2017-07-14 Martin Liska <mliska@suse.cz>
23913 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23914 * dwarf2out.c (is_java): Remove the function.
23915 (output_pubname): Remove usage of the function.
23916 (lower_bound_default): Remove usage of DW_LANG_Java.
23917 (gen_compile_unit_die): Likewise.
23918 * gcc.c: Remove compiler defaults for .java and .zip files.
23919 * gimple-expr.c (remove_suffix): Change as there's no longer
23920 extension than 4-letter one.
23921 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23922 (gimplify_save_expr): Likewise.
23923 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23924 as it's possible even for other languages than Java.
23925 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23926 * lto-opts.c (lto_write_options): Remove reference to Java.
23927 * opts.c (strip_off_ending): Update file extension handling.
23928 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23929 * tree-eh.c (lower_resx): Likewise.
23930 * tree.c (free_lang_data_in_type): Remove dead code.
23931 (find_decls_types_r): Likewise.
23932 (build_common_builtin_nodes): Remove Java from a comment.
23933 (verify_type): Remove dead code.
23934 * varasm.c (assemble_external): Remove Java from a comment.
23936 2017-07-14 Martin Liska <mliska@suse.cz>
23938 * opts.c (finish_options): Add quotes.
23939 (common_handle_option): Likewise.
23941 2017-07-14 Martin Liska <mliska@suse.cz>
23943 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23944 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23945 Remove N_SO_PASCAL.
23946 * dwarf2out.c (lower_bound_default): Do not handle
23948 (gen_compile_unit_die): Likewise.
23949 * gcc.c: Remove default extension binding for GNU Pascal.
23950 * stmt.c: Remove Pascal language from a comment.
23951 * xcoffout.c: Likewise.
23953 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23956 * diagnostic-show-locus.c (fixit_cmp): New function.
23957 (layout::layout): Sort m_fixit_hints.
23958 (column_range::column_range): Assert that the values are valid.
23959 (struct char_span): New struct.
23960 (correction::overwrite): New method.
23961 (struct source_line): New struct.
23962 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23963 calls in terms of classes source_line and char_span, and
23964 correction::overwrite.
23965 (selftest::test_overlapped_fixit_printing_2): New function.
23966 (selftest::diagnostic_show_locus_c_tests): Call it.
23968 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23970 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23971 early if there is no lhs.
23973 2017-07-13 Martin Liska <mliska@suse.cz>
23975 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23976 (gen_reference_type_die): Likewise.
23977 * stor-layout.c: Remove Pascal-related comment.
23979 2017-07-13 Martin Liska <mliska@suse.cz>
23981 * opts.c (finish_options): Add quotes to error messages.
23982 (parse_sanitizer_options): Likewise.
23984 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23986 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23988 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23990 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23992 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23994 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23996 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23998 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24001 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24002 provides the hardware capability bits, define the macro
24003 __BUILTIN_CPU_SUPPORTS__.
24004 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24005 if GLIBC does not provide the hardware capability bits. Add a
24006 gcc_unreachable call if the built-in cpu function is neither
24007 __builtin_cpu_is nor __builtin_cpu_supports.
24008 (rs6000_get_function_versions_dispatcher): Change the warning
24009 that an old GLIBC is used which does not export the capability
24010 bits to be an error.
24011 * doc/extend.texi (target_clones attribute): Document the
24012 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24013 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24014 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24015 the macros defined by GCC if the newer GLIBC is available.
24017 2017-07-12 Jeff Law <law@redhat.com>
24019 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24020 remaining includes slightly.
24021 * config/riscv/riscv-builtins.c: Include profile-count.h.
24023 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24026 * config/avr/avr.c (avr_set_current_function): In diagnostic
24027 messages: Quote keywords and (parts of) identifiers.
24028 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24031 2017-07-12 Carl Love <cel@us.ibm.com>
24033 * config/rs6000/rs6000-c.c: Add support for built-in functions
24034 vector bool char vec_revb (vector bool char);
24035 vector bool short vec_revb (vector short char);
24036 vector bool int vec_revb (vector bool int);
24037 vector bool long long vec_revb (vector bool long long);
24038 * doc/extend.texi: Update the built-in documentation file for the
24039 new built-in functions.
24041 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24043 * config/s390/s390.md: Remove movcc splitter.
24045 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24047 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24048 load/store on condition.
24050 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24053 * config/avr/avr.c (avr_encode_section_info)
24054 [progmem && !TREE_READONLY]: Error if progmem object needs
24057 2017-07-11 Michael Collison <michael.collison@arm.com>
24059 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24062 2017-07-11 Carl Love <cel@us.ibm.com>
24064 * config/rs6000/rs6000-c.c: Add support for builtins
24065 vector unsigned int vec_parity_lsbb (vector signed int);
24066 vector unsigned int vec_parity_lsbb (vector unsigned int);
24067 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24068 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24069 vector unsigned long long vec_parity_lsbb (vector signed long long);
24070 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24071 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24072 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24073 * doc/extend.texi: Update the built-in documentation file for the
24074 new built-in functions.
24076 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24078 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24079 (layout::m_primary_loc): New field.
24080 (layout::layout): Initialize new field. Move location filtering
24081 logic from here to...
24082 (layout::maybe_add_location_range): ...this new method. Add
24083 support for filtering to just the lines already specified by other
24085 (layout::will_show_line_p): New method.
24086 (gcc_rich_location::add_location_if_nearby): New method.
24087 (selftest::test_add_location_if_nearby): New test function.
24088 (selftest::diagnostic_show_locus_c_tests): Call it.
24089 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24092 2017-07-11 Tom de Vries <tom@codesourcery.com>
24094 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24095 (bb_first_real_insn): New function.
24096 (nvptx_single): Add extra initialization of broadcasted condition
24099 2017-07-11 Nathan Sidwell <nathan@acm.org>
24101 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24103 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24105 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24106 Binutils doc as TEXI will mess them up.
24107 * doc/invoke.texi (AVR Options): Same here.
24109 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24111 * config/sparc/sparc.opt (mfix-ut700): New option.
24112 (mfix-gr712rc): Likewise.
24113 (sparc_fix_b2bst): New variable.
24114 * doc/invoke.texi (SPARC options): Document them.
24115 (ARM options): Fix warnings.
24116 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24117 instructions to prevent sequences that can trigger the store-store
24118 errata for certain LEON3FT processors.
24119 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24120 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24121 * config/sparc/sparc.md (fix_b2bst): New attribute.
24122 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24124 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24127 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24129 (*rsqrtsf2_sse): Ditto.
24131 (div<mode>3): Macroize insn from divdf3 and divsf3
24132 using MODEF mode iterator.
24134 2017-07-10 Martin Sebor <msebor@redhat.com>
24136 PR tree-optimization/80397
24137 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24138 instead of testing for equality to INTEGER_TYPE.
24140 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24142 * config.gcc: Remove uclibc from arc target spec.
24144 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24146 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24148 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24151 * lto-wrapper.c (remove_option): New function.
24152 (merge_and_complain): Merge PIC/PIE options more realistically.
24154 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24156 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24160 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24161 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24162 * config.in: Regenerate.
24163 * configure: Regenerate.
24164 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24165 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24166 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24167 (TARGET_GASISR_PROLOGUES): ...target mask.
24168 * common/config/avr/avr-common.c
24169 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24170 Set -mgas-isr-prologues.
24171 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24172 INSERT_PASS_BEFORE for it.
24173 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24174 * config/avr/avr.c (avr_option_override)
24175 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24176 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24177 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24178 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24179 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24181 (make_avr_pass_pre_proep): New function.
24182 (emit_push_sfr) <treg>: Add argument to function and use it
24183 instead of TMP_REG.
24184 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24185 and set machine->gasisr.yes.
24186 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24187 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24188 __gcc_isr.n_pushed to .L__stack_usage.
24189 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24190 (avr_asm_final_postscan_insn): ...this new static function.
24191 * config/avr/avr.h (machine_function)
24192 <is_no_gccisr, use_L__stack_usage>: New fields.
24193 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24194 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24195 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24196 (gasisr, *gasisr): New expander and insn.
24197 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24198 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24199 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24201 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24203 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24206 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24208 Move jump-tables out of .text again.
24211 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24212 (ASM_OUTPUT_ADDR_VEC): New function.
24213 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24214 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24215 INSN_ADDRESSes as asm comment.
24216 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24217 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24218 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24219 * config/avr/avr.md (*tablejump): Adjust comment.
24220 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24221 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24223 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24224 (avr_output_addr_vec): New proto.
24225 (avr_log_t) <insn_addresses>: New field.
24227 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24230 * config/i386/i386.c (ix86_function_arg_advance): Set
24231 outgoing_args_on_stack to true if there are outgoing arguments
24233 (ix86_function_arg): Likewise.
24234 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24235 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24236 * config/i386/i386.h (machine_function): Add
24237 outgoing_args_on_stack.
24239 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24241 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24242 supporting pthreds.
24243 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24245 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24247 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24248 (REAL_H): Remove $(MACHMODE_H).
24249 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24251 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24252 $(MACHMODE_H) and double-int.h.
24253 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24255 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24258 2017-07-07 Andrew Pinski <apinski@cavium.com>
24260 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24261 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24263 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24265 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24266 Add warning if GCC was not configured to link against a GLIBC that
24267 exports the hardware capability bits.
24268 (make_resolver_func): Make resolver function private and not a
24269 COMDAT function. Create the name with clone_function_name instead
24270 of make_unique_name.
24273 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24274 correct operand in doing the split.
24276 2017-07-07 Carl Love <cel@us.ibm.com>
24278 * config/rs6000/rs6000-c: Add support for built-in function
24279 vector unsigned short vec_pack_to_short_fp32 (vector float,
24281 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24282 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24283 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24284 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24285 (convert_4f32_8i16): Add define_expand.
24286 * doc/extend.texi: Update the built-in documentation file for the
24287 new built-in function.
24289 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24291 * config/sparc/m8.md: New file.
24292 * config/sparc/sparc.md: Include m8.md.
24294 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24296 * config/sparc/sparc.opt: New option -mvis4b.
24297 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24298 (sparc_option_override): Handle VIS4B.
24299 (enum sparc_builtins): Define
24300 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24301 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24302 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24303 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24304 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24305 (check_constant_argument): New function.
24306 (sparc_vis_init_builtins): Define builtins
24307 __builtin_vis_dictunpack{8,16,32},
24308 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24309 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24310 __builtin_vis_fpcmpde{8,16,32}shl and
24311 __builtin_vis_fpcmpur{8,16,32}shl.
24312 (sparc_expand_builtin): Check that the constant operands to
24313 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24314 constant and in range.
24315 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24317 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24318 (SPARC_IMM5_P): Likewise.
24319 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24320 (enabled): Handle vis4b.
24321 (UNSPEC_DICTUNPACK): New unspec.
24322 (UNSPEC_FPCMPSHL): Likewise.
24323 (UNSPEC_FPUCMPSHL): Likewise.
24324 (UNSPEC_FPCMPDESHL): Likewise.
24325 (UNSPEC_FPCMPURSHL): Likewise.
24326 (cpu_feature): New CPU feature `vis4b'.
24327 (dictunpack{8,16,32}): New insns.
24328 (FPCSMODE): New mode iterator.
24329 (fpcscond): New code iterator.
24330 (fpcsucond): Likewise.
24331 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24332 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24333 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24334 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24335 * config/sparc/constraints.md: Define constraints `q' for unsigned
24336 2-bit integer constants and `t' for unsigned 5-bit integer
24338 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24340 (imm5_operand_dictunpack16): Likewise.
24341 (imm5_operand_dictunpack32): Likewise.
24342 (imm2_operand): Likewise.
24343 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24344 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24345 ditunpack* and fpcmp*shl builtins.
24347 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24349 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24350 * config.in: Add HAVE_AS_SPARC6 define.
24351 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24353 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24355 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24356 (CPP_CPU_SPEC): Handle m8.
24357 (ASM_CPU_SPEC): Likewise.
24358 * config/sparc/sparc-opts.h (enum processor_type): Add
24360 * config/sparc/sparc.c (m8_costs): New struct.
24361 (sparc_option_override): Handle TARGET_CPU_m8.
24362 (sparc32_initialize_trampoline): Likewise.
24363 (sparc64_initialize_trampoline): Likewise.
24364 (sparc_issue_rate): Likewise.
24365 (sparc_register_move_cost): Likewise.
24366 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24367 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24368 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24369 (CPP_CPU_SPEC): Handle M8.
24370 (ASM_CPU_SPEC): Likewise.
24371 (AS_M8_FLAG): Define.
24372 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24373 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24374 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24376 * configure: Regenerate.
24377 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24380 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24382 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24384 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24385 ("*movdi_insn_sp32"): Do not set v3pipe.
24386 ("*movsi_insn"): Likewise.
24387 ("*movdi_insn_sp64"): Likewise.
24388 ("*movsf_insn"): Likewise.
24389 ("*movdf_insn_sp32"): Likewise.
24390 ("*movdf_insn_sp64"): Likewise.
24391 ("*zero_extendsidi2_insn_sp64"): Likewise.
24392 ("*sign_extendsidi2_insn"): Likewise.
24393 ("*mov<VM32:mode>_insn"): Likewise.
24394 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24395 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24396 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24397 ("<vlop:code><VL:mode>3"): Likewise.
24398 ("*not_<vlop:code><VL:mode>3"): Likewise.
24399 ("*nand<VL:mode>_vis"): Likewise.
24400 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24401 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24402 ("one_cmpl<VL:mode>2"): Likewise.
24403 ("faligndata<VM64:mode>_vis"): Likewise.
24404 ("alignaddrsi_vis"): Likewise.
24405 ("alignaddrdi_vis"): Likweise.
24406 ("alignaddrlsi_vis"): Likewise.
24407 ("alignaddrldi_vis"): Likewise.
24408 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24409 ("bmaskdi_vis"): Likewise.
24410 ("bmasksi_vis"): Likewise.
24411 ("bshuffle<VM64:mode>_vis"): Likewise.
24412 ("cmask8<P:mode>_vis"): Likewise.
24413 ("cmask16<P:mode>_vis"): Likewise.
24414 ("cmask32<P:mode>_vis"): Likewise.
24415 ("pdistn<P:mode>_vis"): Likewise.
24416 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24418 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24420 * config/sparc/sparc.md ("subtype"): New insn attribute.
24421 ("*wrgsr_sp64"): Set insn subtype.
24422 ("*rdgsr_sp64"): Likewise.
24423 ("alignaddrsi_vis"): Likewise.
24424 ("alignaddrdi_vis"): Likewise.
24425 ("alignaddrlsi_vis"): Likewise.
24426 ("alignaddrldi_vis"): Likewise.
24427 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24428 ("fexpand_vis"): Likewise.
24429 ("fpmerge_vis"): Likewise.
24430 ("faligndata<VM64:mode>_vis"): Likewise.
24431 ("bshuffle<VM64:mode>_vis"): Likewise.
24432 ("cmask8<P:mode>_vis"): Likewise.
24433 ("cmask16<P:mode>_vis"): Likewise.
24434 ("cmask32<P:mode>_vis"): Likewise.
24435 ("fchksm16_vis"): Likewise.
24436 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24437 ("fmean16_vis"): Likewise.
24438 ("fp<plusminus_insn>64_vis"): Likewise.
24439 ("<plusminus_insn>v8qi3"): Likewise.
24440 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24441 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24442 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24443 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24444 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24445 ("*movqi_insn"): Likewise.
24446 ("*movhi_insn"): Likewise.
24447 ("*movsi_insn"): Likewise.
24448 ("movsi_pic_gotdata_op"): Likewise.
24449 ("*movdi_insn_sp32"): Likewise.
24450 ("*movdi_insn_sp64"): Likewise.
24451 ("movdi_pic_gotdata_op"): Likewise.
24452 ("*movsf_insn"): Likewise.
24453 ("*movdf_insn_sp32"): Likewise.
24454 ("*movdf_insn_sp64"): Likewise.
24455 ("*zero_extendhisi2_insn"): Likewise.
24456 ("*zero_extendqihi2_insn"): Likewise.
24457 ("*zero_extendqisi2_insn"): Likewise.
24458 ("*zero_extendqidi2_insn"): Likewise.
24459 ("*zero_extendhidi2_insn"): Likewise.
24460 ("*zero_extendsidi2_insn_sp64"): Likewise.
24461 ("ldfsr"): Likewise.
24462 ("prefetch_64"): Likewise.
24463 ("prefetch_32"): Likewise.
24464 ("tie_ld32"): Likewise.
24465 ("tie_ld64"): Likewise.
24466 ("*tldo_ldub_sp32"): Likewise.
24467 ("*tldo_ldub1_sp32"): Likewise.
24468 ("*tldo_ldub2_sp32"): Likewise.
24469 ("*tldo_ldub_sp64"): Likewise.
24470 ("*tldo_ldub1_sp64"): Likewise.
24471 ("*tldo_ldub2_sp64"): Likewise.
24472 ("*tldo_ldub3_sp64"): Likewise.
24473 ("*tldo_lduh_sp32"): Likewise.
24474 ("*tldo_lduh1_sp32"): Likewise.
24475 ("*tldo_lduh_sp64"): Likewise.
24476 ("*tldo_lduh1_sp64"): Likewise.
24477 ("*tldo_lduh2_sp64"): Likewise.
24478 ("*tldo_lduw_sp32"): Likewise.
24479 ("*tldo_lduw_sp64"): Likewise.
24480 ("*tldo_lduw1_sp64"): Likewise.
24481 ("*tldo_ldx_sp64"): Likewise.
24482 ("*mov<VM32:mode>_insn"): Likewise.
24483 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24484 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24486 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24488 * config/sparc/sparc.md ("type"): New insn type viscmp.
24489 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24491 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24492 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24493 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24494 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24496 ("n7_vis_logical_11cycle"): Likewise.
24497 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24498 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24499 * config/sparc/niagara.md ("niag_vis"): Likewise.
24500 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24501 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24503 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24505 * config/sparc/sparc.md: New instruction type `bmask'.
24506 (bmaskdi_vis): Use the `bmask' type.
24507 (bmasksi_vis): Likewise.
24508 * config/sparc/ultra3.md (us3_array): Likewise.
24509 * config/sparc/niagara7.md (n7_array): Likewise.
24510 * config/sparc/niagara4.md (n4_array): Likewise.
24511 * config/sparc/niagara2.md (niag2_vis): Likewise.
24512 (niag3_vis): Likewise.
24513 * config/sparc/niagara.md (niag_vis): Likewise.
24515 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24517 * ipa-comdats.c: Remove optimize check from gate.
24518 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24519 for functions not optimized.
24520 (ipa_fn_summary_read): Skip optimize check.
24521 (ipa_fn_summary_write): Likewise.
24522 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24524 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24526 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24527 (check_callers): Give up on caller which is not optimized.
24528 (inline_small_functions): Likewise.
24529 (ipa_inline): Do not give up when not optimizing.
24530 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24531 away unoptimizes cdtors.
24532 (whole_program_function_and_variable_visibility): Do
24533 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24534 * ipa.c (process_references): Do not check optimize.
24535 (symbol_table::remove_unreachable_nodes): Update optimize check.
24536 (set_writeonly_bit): Update optimize check.
24537 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24538 (pass_ipa_single_use::gate): Remove.
24540 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24542 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24543 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24544 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24545 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24546 permute_load, permute_store, adjust_extract, adjust_splat,
24547 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24548 replace_swap_with_copy, dump_swap_insn_table,
24549 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24550 recombine_lvx_pattern, recombine_stvx_pattern,
24551 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24552 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24553 to file rs6000-p8swap.c.
24554 * config/rs6000/rs6000-p8swap.c: New file.
24555 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24556 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24557 and rs6000*-*-* targets.
24559 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24561 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24563 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24565 * lto-wrapper.c (merge_and_complain): Do not merge
24566 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24567 fsigned_zeros, ftrapping_math, fwrapv.
24568 (append_compiler_options): Do not track these options.
24569 (append_linker_options): Likewie
24571 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24573 * cgraphunit.c (cgraph_node::finalize_function): When
24574 !flag_toplevel_reorde set no_reorder flag.
24575 (varpool_node::finalize_decl): Likewise.
24576 (symbol_table::compile): Drop no toplevel reorder path.
24578 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24580 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24581 edges; zero probability is not better than uninitialized.
24583 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24585 * asan.h (asan_sanitize_allocas_p): Declare.
24586 * asan.c (asan_sanitize_allocas_p): New function.
24587 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24588 (handle_builtin_alloca): Likewise.
24589 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24590 if !asan_sanitize_allocas_p.
24591 * params.def (asan-instrument-allocas): Add new option.
24592 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24593 * opts.c (common_handle_option): Disable allocas sanitization for
24596 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24598 * asan.c: Include gimple-fold.h.
24599 (get_last_alloca_addr): New function.
24600 (handle_builtin_stackrestore): Likewise.
24601 (handle_builtin_alloca): Likewise.
24602 (asan_emit_allocas_unpoison): Likewise.
24603 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24604 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24605 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24606 (instrument_builtin_call): Pass gimple iterator to
24607 get_mem_refs_of_builtin_call.
24608 (last_alloca_addr): New global.
24609 * asan.h (asan_emit_allocas_unpoison): Declare.
24610 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24611 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24612 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24613 if function calls alloca.
24614 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24615 * gimple-fold.h (replace_call_with_value): Declare.
24616 * internal-fn.c: Include asan.h.
24617 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24618 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24620 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24622 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24623 (C_SELFTEST_FLAGS): New.
24624 (CPP_SELFTEST_FLAGS): New.
24625 (SELFTEST_DEPS): New, from deps of s-selftest.
24626 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24627 (CPP_SELFTEST_DEPS): New.
24628 (selftest): Add dependency on s-selftest-c++.
24629 (s-selftest): Rename to...
24630 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24631 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24632 than SELFTEST_FLAGS.
24633 (selftest-gdb): Rename to...
24634 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24636 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24637 (selftest-valgrind): Rename to...
24638 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24640 (selftest-valgrind): Reintroduce as an alias for
24641 selftest-c-valgrind.
24642 (s-selftest-c++): New.
24643 (selftest-c++-gdb): New.
24644 (selftest-c++-valgrind): New.
24646 2017-07-06 Olivier Hainque <hainque@adacore.com>
24648 * gcc.c (process_command): When deciding if undefined variables
24649 should be ignored when processing specs, accept "gcc -v" as well.
24651 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24653 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24654 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24656 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24658 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24659 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24660 * config/arm/arm-tables.opt: Regenerate.
24661 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24663 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24665 2017-07-06 Carl Love <cel@us.ibm.com>
24667 * ChangeLog: Clean up from mid air collision
24669 2017-07-06 Carl Love <cel@us.ibm.com>
24671 * config/rs6000/rs6000-c.c: Add support for built-in functions
24672 vector signed int vec_subc (vector signed int, vector signed int);
24673 vector signed __int128 vec_subc (vector signed __int128,
24674 vector signed __int128);
24675 vector unsigned __int128 vec_subc (vector unsigned __int128,
24676 vector unsigned __int128);
24677 vector signed int vec_sube (vector signed int, vector signed int,
24678 vector signed int);
24679 vector unsigned int vec_sube (vector unsigned int,
24680 vector unsigned int,
24681 vector unsigned int);
24682 vector signed __int128 vec_sube (vector signed __int128,
24683 vector signed __int128,
24684 vector signed__int128);
24685 vector unsigned __int128 vec_sube (vector unsigned __int128,
24686 vector unsigned __int128,
24687 vector unsigned __int128);
24688 vector signed int vec_subec (vector signed int, vector signed int,
24689 vector signed int);
24690 vector unsigned int vec_subec (vector unsigned int,
24691 vector unsigned int,
24692 vector unsigned int);
24693 vector signed __int128 vec_subec (vector signed __int128,
24694 vector signed __int128,
24695 vector signed__int128);
24696 vector unsigned __int128 vec_subec (vector unsigned __int128,
24697 vector unsigned __int128,
24698 vector unsigned __int128);
24699 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24700 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24701 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24702 BU_ALTIVEC_OVERLOAD_X definitions.
24703 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24704 * doc/extend.texi: Update the built-in documentation file for the new
24705 built-in functions.
24707 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24710 * diagnostic-show-locus.c (layout::layout): Use start and finish
24711 spelling location for the start and finish of each range.
24712 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24713 Add unused aspect param.
24714 * input.c (expand_location_1): Add "aspect" param, and use it
24715 to access the correct part of the location.
24716 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24718 (expand_location_to_spelling_point): Likewise.
24719 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24720 param, and pass it to expand_location_1.
24722 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24724 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24725 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24726 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24727 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24728 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24729 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24730 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24731 _mm_maskz_getmant_ss): New intrinsics.
24732 (__builtin_ia32_getexpss128_mask): Changed to ...
24733 __builtin_ia32_getexpss128_round ... this.
24734 (__builtin_ia32_getexpsd128_mask): Changed to ...
24735 __builtin_ia32_getexpsd128_round ... this.
24736 * config/i386/i386-builtin-types.def
24737 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24738 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24739 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24740 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24741 __builtin_ia32_getmantss_mask_round): New builtins.
24742 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24743 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24744 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24745 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24746 * config/i386/sse.md
24747 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24748 avx512f_sgetexp<mode><mask_scalar_name>
24749 <round_saeonly_scalar_name> ... this.
24750 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24751 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24752 vgetexp<ssescalarmodesuffix>
24753 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24754 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24755 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24756 avx512f_vgetmant<mode><mask_scalar_name>
24757 <round_saeonly_scalar_name> ... this.
24758 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24759 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24760 vgetmant<ssescalarmodesuffix>
24761 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24762 %0<mask_scalar_operand4>, %1, %2
24763 <round_saeonly_scalar_mask_op4>, %3} ... this.
24764 * config/i386/subst.md (mask_scalar_operand4,
24765 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24766 round_saeonly_scalar_nimm_predicate): New subst attributes.
24768 2017-07-06 Julia Koval <julia.koval@intel.com>
24770 * config/i386/i386.c (ix86_erase_embedded_rounding):
24771 Remove code for old rounding pattern.
24773 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24775 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24777 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24779 * doc/sourcebuild.texi (Test Directives, Variants of
24780 dg-require-support): Add documentation for dg-require-stack-check.
24782 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24784 * config/i386/subst.md (mask_scalar, round_scalar,
24785 round_saeonly_scalar): New meta-templates.
24786 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24787 round_scalar_mask_operand3, round_scalar_mask_op3,
24788 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24789 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24790 round_saeonly_scalar_constraint,
24791 round_saeonly_scalar_prefix): New subst attribute.
24792 * config/i386/sse.md
24793 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24794 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24795 <round_scalar_name> ... this.
24796 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24797 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24798 <round_scalar_name> ... this.
24799 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24800 <sse>_vm<code><mode>3<mask_scalar_name>
24801 <round_saeonly_scalar_name> ... this.
24802 (v<plusminus_mnemonic><ssescalarmodesuffix>
24803 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24804 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24805 v<plusminus_mnemonic><ssescalarmodesuffix>
24806 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24807 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24808 (v<multdiv_mnemonic><ssescalarmodesuffix>
24809 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24810 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24811 v<multdiv_mnemonic><ssescalarmodesuffix>
24812 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24813 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24814 (v<maxmin_float><ssescalarmodesuffix>
24815 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24816 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24817 v<maxmin_float><ssescalarmodesuffix>
24818 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24819 %0<mask_scalar_operand3>, %1, %<iptr>2
24820 <round_saeonly_scalar_mask_op3>} ... this.
24822 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24824 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24825 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24827 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24828 Alan Hayward <alan.hayward@arm.com>
24829 David Sherwood <david.sherwood@arm.com>
24831 * combine.c (simplify_if_then_else): Remove "enum" before
24833 * compare-elim.c (can_eliminate_compare): Likewise.
24834 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24836 (aarch64_lookup_simd_builtin_type): Likewise.
24837 (aarch64_simd_builtin_type): Likewise.
24838 (aarch64_init_simd_builtin_types): Likewise.
24839 (aarch64_simd_expand_args): Likewise.
24840 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24842 (aarch64_reverse_mask): Likewise.
24843 (aarch64_simd_emit_reg_reg_move): Likewise.
24844 (aarch64_gen_adjusted_ldpstp): Likewise.
24845 (aarch64_ccmp_mode_to_code): Likewise.
24846 (aarch64_operands_ok_for_ldpstp): Likewise.
24847 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24848 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24850 (aarch64_min_divisions_for_recip_mul): Likewise.
24851 (aarch64_reassociation_width): Likewise.
24852 (aarch64_get_condition_code_1): Likewise.
24853 (aarch64_simd_emit_reg_reg_move): Likewise.
24854 (aarch64_simd_attr_length_rglist): Likewise.
24855 (aarch64_reverse_mask): Likewise.
24856 (aarch64_operands_ok_for_ldpstp): Likewise.
24857 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24858 (aarch64_gen_adjusted_ldpstp): Likewise.
24859 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24861 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24862 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24863 (arm_lookup_simd_builtin_type): Likewise.
24864 (arm_simd_builtin_type): Likewise.
24865 (arm_init_simd_builtin_types): Likewise.
24866 (arm_expand_builtin_args): Likewise.
24867 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24868 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24869 (ft32_setup_incoming_varargs): Likewise.
24870 (ft32_function_arg): Likewise.
24871 (ft32_function_arg_advance): Likewise.
24872 (ft32_pass_by_reference): Likewise.
24873 (ft32_arg_partial_bytes): Likewise.
24874 (ft32_valid_pointer_mode): Likewise.
24875 (ft32_addr_space_pointer_mode): Likewise.
24876 (ft32_addr_space_legitimate_address_p): Likewise.
24877 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24879 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24880 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24881 (iamcu_alignment): Likewise.
24882 (canonicalize_vector_int_perm): Likewise.
24883 (ix86_noce_conversion_profitable_p): Likewise.
24884 (ix86_mpx_bound_mode): Likewise.
24885 (ix86_operands_ok_for_move_multiple): Likewise.
24886 * config/microblaze/microblaze-protos.h
24887 (microblaze_expand_conditional_branch_reg): Likewise.
24888 * config/microblaze/microblaze.c
24889 (microblaze_expand_conditional_branch_reg): Likewise.
24890 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24892 (rs6000_reassociation_width): Likewise.
24893 (rs6000_invalid_binary_op): Likewise.
24894 (fusion_p9_p): Likewise.
24895 (emit_fusion_p9_load): Likewise.
24896 (emit_fusion_p9_store): Likewise.
24897 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24899 (riscv_hard_regno_mode_ok_p): Likewise.
24900 (riscv_address_insns): Likewise.
24901 (riscv_split_symbol): Likewise.
24902 (riscv_legitimize_move): Likewise.
24903 (riscv_function_value): Likewise.
24904 (riscv_hard_regno_nregs): Likewise.
24905 (riscv_expand_builtin): Likewise.
24906 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24907 (riscv_build_integer): Likewise.
24908 (riscv_split_integer): Likewise.
24909 (riscv_legitimate_constant_p): Likewise.
24910 (riscv_cannot_force_const_mem): Likewise.
24911 (riscv_regno_mode_ok_for_base_p): Likewise.
24912 (riscv_valid_base_register_p): Likewise.
24913 (riscv_valid_offset_p): Likewise.
24914 (riscv_valid_lo_sum_p): Likewise.
24915 (riscv_classify_address): Likewise.
24916 (riscv_legitimate_address_p): Likewise.
24917 (riscv_address_insns): Likewise.
24918 (riscv_load_store_insns): Likewise.
24919 (riscv_force_binary): Likewise.
24920 (riscv_split_symbol): Likewise.
24921 (riscv_force_address): Likewise.
24922 (riscv_legitimize_address): Likewise.
24923 (riscv_move_integer): Likewise.
24924 (riscv_legitimize_const_move): Likewise.
24925 (riscv_legitimize_move): Likewise.
24926 (riscv_address_cost): Likewise.
24927 (riscv_subword): Likewise.
24928 (riscv_output_move): Likewise.
24929 (riscv_canonicalize_int_order_test): Likewise.
24930 (riscv_emit_int_order_test): Likewise.
24931 (riscv_function_arg_boundary): Likewise.
24932 (riscv_pass_mode_in_fpr_p): Likewise.
24933 (riscv_pass_fpr_single): Likewise.
24934 (riscv_pass_fpr_pair): Likewise.
24935 (riscv_get_arg_info): Likewise.
24936 (riscv_function_arg): Likewise.
24937 (riscv_function_arg_advance): Likewise.
24938 (riscv_arg_partial_bytes): Likewise.
24939 (riscv_function_value): Likewise.
24940 (riscv_pass_by_reference): Likewise.
24941 (riscv_setup_incoming_varargs): Likewise.
24942 (riscv_print_operand): Likewise.
24943 (riscv_elf_select_rtx_section): Likewise.
24944 (riscv_save_restore_reg): Likewise.
24945 (riscv_for_each_saved_reg): Likewise.
24946 (riscv_register_move_cost): Likewise.
24947 (riscv_hard_regno_mode_ok_p): Likewise.
24948 (riscv_hard_regno_nregs): Likewise.
24949 (riscv_class_max_nregs): Likewise.
24950 (riscv_memory_move_cost): Likewise.
24951 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24952 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24953 (rl78_addr_space_address_mode): Likewise.
24954 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24956 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24957 (rs6000_reassociation_width): Likewise.
24958 (rs6000_invalid_binary_op): Likewise.
24959 (fusion_p9_p): Likewise.
24960 (emit_fusion_p9_load): Likewise.
24961 (emit_fusion_p9_store): Likewise.
24962 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24963 (ok_for_simple_move_operands): Likewise.
24964 (ok_for_simple_move_strict_operands): Likewise.
24965 (ok_for_simple_arith_logic_operands): Likewise.
24966 (visium_legitimize_reload_address): Likewise.
24967 (visium_select_cc_mode): Likewise.
24968 (output_cbranch): Likewise.
24969 (visium_split_double_move): Likewise.
24970 (visium_expand_copysign): Likewise.
24971 (visium_expand_int_cstore): Likewise.
24972 (visium_expand_fp_cstore): Likewise.
24973 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24974 (visium_function_arg): Likewise.
24975 (visium_function_arg_advance): Likewise.
24976 (visium_libcall_value): Likewise.
24977 (visium_setup_incoming_varargs): Likewise.
24978 (visium_legitimate_constant_p): Likewise.
24979 (visium_legitimate_address_p): Likewise.
24980 (visium_legitimize_address): Likewise.
24981 (visium_secondary_reload): Likewise.
24982 (visium_register_move_cost): Likewise.
24983 (visium_memory_move_cost): Likewise.
24984 (prepare_move_operands): Likewise.
24985 (ok_for_simple_move_operands): Likewise.
24986 (ok_for_simple_move_strict_operands): Likewise.
24987 (ok_for_simple_arith_logic_operands): Likewise.
24988 (visium_function_value_1): Likewise.
24989 (rtx_ok_for_offset_p): Likewise.
24990 (visium_legitimize_reload_address): Likewise.
24991 (visium_split_double_move): Likewise.
24992 (visium_expand_copysign): Likewise.
24993 (visium_expand_int_cstore): Likewise.
24994 (visium_expand_fp_cstore): Likewise.
24995 (visium_split_cstore): Likewise.
24996 (visium_select_cc_mode): Likewise.
24997 (visium_split_cbranch): Likewise.
24998 (output_cbranch): Likewise.
24999 (visium_print_operand_address): Likewise.
25000 * expmed.c (flip_storage_order): Likewise.
25001 * expmed.h (emit_cstore): Likewise.
25002 (flip_storage_order): Likewise.
25003 * genrecog.c (validate_pattern): Likewise.
25004 * hsa-gen.c (gen_hsa_addr): Likewise.
25005 * internal-fn.c (expand_arith_overflow): Likewise.
25006 * ira-color.c (allocno_copy_cost_saving): Likewise.
25007 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25008 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25009 (process_invariant_for_inheritance): Likewise.
25010 * lra-eliminations.c (move_plus_up): Likewise.
25011 * omp-low.c (lower_oacc_reductions): Likewise.
25012 * simplify-rtx.c (simplify_subreg): Likewise.
25013 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25014 (TARGET_CHKP_BOUND_MODE): Likewise..
25015 * targhooks.c (default_chkp_bound_mode): Likewise.
25016 (default_setup_incoming_vararg_bounds): Likewise.
25017 * targhooks.h (default_chkp_bound_mode): Likewise.
25018 (default_setup_incoming_vararg_bounds): Likewise.
25019 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25020 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25021 (have_whole_vector_shift): Likewise.
25022 * tree-vect-stmts.c (vectorizable_load): Likewise.
25023 * doc/tm.texi: Regenerate.
25025 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25027 Graceful degrade if Binutils PR21472 is not available.
25030 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25031 .rodata in flash test fails.
25032 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25033 * confgure: Regenerate.
25034 * config.in: Regenerate.
25035 * config/avr/avr.c (avr_asm_named_section)
25036 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25037 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25038 (avr_asm_init_sections): Same.
25040 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25042 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25043 (fma<VH:mode>4_intrinsic): Likewise.
25044 (*fmsub<VCVTF:mode>4): Likewise.
25045 (*fmsub<VH:mode>4_intrinsic): Likewise.
25047 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25050 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25051 Don't depend on "optimize > 0".
25052 (out_movhi_r_mr, out_movqi_mr_r): Same.
25053 (out_movhi_mr_r, out_movqi_r_mr): Same.
25054 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25055 io_address_operand on "optimize > 0".
25057 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25059 * tree-loop-distribution.c: Add general explanantion on the pass.
25060 (generate_loops_for_partition): Mark distributed loop.
25061 (pg_add_dependence_edges): New parameter. Handle alias data
25062 dependence specially and record it in the parameter if asked.
25063 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25064 (init_partition_graph_vertices, add_partition_graph_edge): New.
25065 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25066 (free_partition_graph_vdata, build_partition_graph): New.
25067 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25068 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25069 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25070 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25071 (version_for_distribution_p, finalize_partitions): New.
25072 (distribute_loop): Handle alias data dependence specially. Factor
25073 out loop fusion code as functions and call these functions.
25075 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25077 * tree-loop-distribution.c (classify_partition): New parameter and
25078 better handle reduction statement.
25079 (rdg_build_partitions): Revise comment.
25080 (distribute_loop): Compute statements in all partitions and pass it
25081 to classify_partition.
25083 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25085 * tree-loop-distribution.c (enum partition_type): New.
25086 (struct partition): New field type.
25087 (partition_merge_into): Add parameter. Update partition type.
25088 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25089 (build_rdg_partition_for_vertex): Compute partition type.
25090 (rdg_build_partitions): Dump partition type.
25091 (distribute_loop): Update calls to partition_merge_into.
25093 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25095 * tree-loop-distribution.c (struct ddr_hasher): New.
25096 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25098 (classify_partition): Call get_data_dependence.
25099 (pg_add_dependence_edges): Ditto.
25100 (distribute_loop): Release data dependence hash table.
25102 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25104 * tree-loop-distribution.c (ref_base_address): Delete.
25105 (similar_memory_accesses): Rename ...
25106 (share_memory_accesses): ... to this. Check if partitions access
25107 the same memory reference.
25108 (distribute_loop): Call share_memory_accesses.
25110 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25112 * tree-loop-distribution.c (struct partition): New field recording
25113 its data reference.
25114 (partition_alloc, partition_free): Init and release data refs.
25115 (partition_merge_into): Merge data refs.
25116 (build_rdg_partition_for_vertex): Collect data refs for partition.
25117 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25119 (distribute_loop): Remve data refs from vertice data of partition
25122 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25124 * tree-loop-distribution.c (params.h): Include header file.
25125 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25126 (datarefs_vec): New global var.
25127 (create_rdg_vertices): Use datarefs_vec directly.
25128 (free_rdg): Don't free data references.
25129 (build_rdg): Update use. Don't free data references.
25130 (distribute_loop): Compute global variable for data references.
25131 Bail out if there are too many data references.
25133 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25135 * tree-loop-distribution.c (loop_nest): New global var.
25136 (build_rdg): Use loop directly, rather than loop nest.
25137 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25139 (distribute_loop): Compute global variable loop nest. Update use.
25141 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25143 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25144 (partition_merge_into): New parameter. Dump reason for fusion.
25145 (distribute_loop): Update use of partition_merge_into.
25147 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25149 * tree-loop-distribution.c (bb_top_order_index): New.
25150 (bb_top_order_index_size, bb_top_order_cmp): New.
25151 (stmts_from_loop): Use topological order.
25152 (pass_loop_distribution::execute): Compute and release topological
25153 order for basic blocks.
25155 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25157 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25160 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25162 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25163 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25164 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25165 * internal-fn.def (LOOP_DIST_ALIAS): New.
25166 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25167 (fold_loop_internal_call): ... this.
25168 (vect_loop_dist_alias_call): New function.
25169 (set_uid_loop_bbs): Call fold_loop_internal_call.
25170 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25173 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25176 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25177 Require dead FLAGS_REG at the beginning of a peephole.
25179 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25182 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25183 arguments in the call to __builtin_ia32_sbb_u32.
25184 (_subborrow_u64): Swap _X and _Y arguments in the call to
25185 __builtin_ia32_sbb_u64.
25187 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25190 * tree-vrp.c (compare_assert_loc): Turn into a function template
25191 with stable template parameter. Only test if a->e is NULL,
25192 !a->e == !b->e has been verified already. Use e == NULL or
25193 e != NULL instead of e or ! e tests. If stable is true, don't use
25194 iterative_hash_expr, on the other side allow a or b or both NULL
25195 and sort the NULLs last.
25196 (process_assert_insertions): Sort using compare_assert_loc<false>
25197 instead of compare_assert_loc, later sort using
25198 compare_assert_loc<true> before calling process_assert_insertions_for
25199 in a loop. Use break instead of continue once seen NULL pointer.
25201 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25203 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25204 Cortex-R7 and Cortex-R8 processors.
25206 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25208 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25209 uninitialized while src is not.
25211 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25213 * common/config/arm/arm-common.c: Adjust include path for
25215 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25216 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25217 (arm-cpu-data.h): Likewise.
25218 (arm-cpu-cdata.h): Likewise.
25219 * config/arm/arm-cpu.h: Delete.
25220 * config/arm/arm-cpu-cdata.h: Delete.
25221 * config/arm/arm-cpu-data.h: Delete.
25223 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25225 * config/arm/arm-cpus.in (cortex-a55): New.
25226 (cortex-a75): Likewise.
25227 (cortex-a75.cortex-a55): Likewise.
25228 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25230 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25231 * config/arm/arm-cpu-cdata.h: Regenerate.
25232 * config/arm/arm-cpu-data.h: Regenerate.
25233 * config/arm/arm-cpu.h: Regenerate.
25234 * config/arm/arm-tables.opt: Regenerate.
25235 * config/arm/arm-tune.md: Regenerate.
25237 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25239 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25241 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25243 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25246 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25248 PR tree-optimization/81292
25249 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25250 full_string_p, also call adjust_related_strinfos if the adjustment
25251 is simple, otherwise invalidate related strinfos.
25253 2017-07-04 Martin Liska <mliska@suse.cz>
25256 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25257 newly created variable as DECL_IGNORED_P.
25259 2017-07-04 Martin Liska <mliska@suse.cz>
25262 * ipa-inline.c (inline_small_functions):
25263 Use xstrdup_for_dump.
25265 2017-07-04 Tom de Vries <tom@codesourcery.com>
25267 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25269 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25272 * config/darwin.c (darwin_function_switched_text_sections):
25275 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25277 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25279 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25281 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25283 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25285 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25286 min_profitable_iters, and th as inclusive lower bounds.
25287 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25288 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25289 for min_profitable_iters and min_profitable_estimate.
25290 (vect_transform_loop): Treat th as an inclusive lower bound.
25291 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25293 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25296 * config/darwin.c (darwin_function_switched_text_sections):
25297 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25298 in two pieces, and suppress the use of buf.
25300 2017-07-03 Nathan Sidwell <nathan@acm.org>
25302 * hash-table.h (hash_table_mod1): Fix indentation.
25304 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25306 PR middle-end/81290
25307 * predict.c (force_edge_cold): Be more careful about propagation
25309 * profile-count.h (profile_probability::guessed,
25310 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25312 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25314 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25316 * doc/invoke.texi (rcpc architecture extension): Document it.
25318 2017-07-03 Richard Biener <rguenther@suse.de>
25320 PR tree-optimization/60510
25321 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25322 the scalar reduction PHI and use it.
25323 (vectorizable_reduction): Properly guard the single_defuse_cycle
25324 path for non-SLP reduction chains where we cannot use it.
25325 Rework reduc_def/index and vector type deduction. Rework
25326 vector operand gathering during reduction op code-gen.
25327 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25328 chains dissolve the chain and leave it to non-SLP reduction
25331 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25333 * tree-data-ref.h (dr_alignment): Declare.
25334 * tree-data-ref.c (dr_alignment): New function.
25335 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25336 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25338 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25340 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25342 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25343 and base_misalignment fields.
25344 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25345 * tree-data-ref.c: Include builtins.h.
25346 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25347 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25348 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25349 * tree-vect-data-refs.c: Include tree-cfg.h.
25350 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25351 fields instead of calculating an alignment here.
25352 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25353 innermost_loop_behavior fields.
25355 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25357 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25359 (DR_STEP_ALIGNMENT): New macro.
25360 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25361 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25362 (create_data_ref): Print it.
25363 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25364 to tell whether the step preserves vector (mis)alignment.
25365 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25366 Move the check for an integer step and generalise to all INTEGER_CST.
25367 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25368 Print the outer step alignment.
25370 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25372 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25373 with offset_alignment.
25374 (DR_ALIGNED_TO): Delete.
25375 (DR_OFFSET_ALIGNMENT): New macro.
25376 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25377 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25378 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25379 (create_data_ref): Likewise.
25380 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25381 (vect_analyze_data_refs): Likewise.
25382 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25383 creating dummy innermost behavior.
25385 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25387 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25388 with a "innermost_loop_behavior *" and refeence tree.
25389 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25390 (create_data_ref): Update call accordingly.
25391 * tree-predcom.c (find_looparound_phi): Likewise.
25393 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25395 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25396 fields with dr_wrt_vec_loop.
25397 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25398 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25399 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25400 (vect_dr_behavior): New function.
25401 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25402 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25403 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25404 track whether the step preserves the misalignment.
25405 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25406 Use vect_dr_behavior.
25407 (vect_setup_realignment): Update call accordingly.
25408 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25409 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25410 call to vect_create_addr_base_for_vector_ref.
25411 (vect_create_cond_for_align_checks): Likewise.
25412 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25413 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25414 (vect_recog_mask_conversion_pattern): Likewise.
25415 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25416 (new_stmt_vec_info): Remove redundant zeroing.
25418 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25420 * common/config/arm/arm-common.c (arm_be8_option): New function.
25421 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25422 (ISA_ARMv6): Add isa_bit_be8.
25423 * config/arm/arm.h (arm_be8_option): Add prototype.
25424 (BE8_SPEC_FUNCTION): New define.
25425 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25426 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25427 (mlittle-endian): Similarly.
25428 (mbe8, mbe32): New options.
25429 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25430 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25432 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25434 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25436 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25438 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25439 (cleanup_tree_cfg_bb): Use it.
25440 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25442 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25444 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25447 * loop-doloop.c (add_test): Update profile.
25449 2017-07-03 Martin Liska <mliska@suse.cz>
25452 * sanopt.c (rewrite_usage_of_param): New function.
25453 (sanitize_rewrite_addressable_params): Likewise.
25454 (pass_sanopt::execute): Call rewrite_usage_of_param.
25456 2017-07-03 Richard Biener <rguenther@suse.de>
25458 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25459 back to using VIEW_CONVERT_EXPR.
25461 2017-07-03 Martin Liska <mliska@suse.cz>
25464 * doc/extend.texi: Document when a resolver function is
25465 generated for target_clones.
25467 2017-07-03 Martin Liska <mliska@suse.cz>
25469 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25470 and shadow memory of auto variables which are subject of
25471 use-after-scope sanitization.
25472 (asan_expand_mark_ifn): Add do set only when is_poison.
25474 2016-07-03 Richard Biener <rguenther@suse.de>
25476 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25478 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25479 (vectorizable_reduction): Perform reduction PHI creation when
25480 visiting a reduction PHI and adjust and simplify code generation
25481 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25482 (vect_transform_loop): Visit reduction PHIs.
25483 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25484 defs into the SLP tree.
25485 (vect_build_slp_tree): Reduction defs terminate the recursion.
25486 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25488 (vect_get_vec_defs_for_stmt_copy): Export.
25489 (vect_get_vec_defs): Likewise.
25490 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25492 (vect_get_vec_defs_for_stmt_copy): Declare.
25493 (vect_get_vec_defs): Likewise.
25495 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25497 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25498 parameter with a "loop" parameter and use it instead of the
25499 loop containing DR_STMT. Don't check simple_iv when doing
25500 BB analysis. Describe the two analysis modes in the comment.
25502 2017-07-03 Tom de Vries <tom@codesourcery.com>
25504 PR tree-optimization/69468
25505 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25506 (find_same_succ_bb): Handle ignore_edge_flags.
25508 2017-07-03 Tom de Vries <tom@codesourcery.com>
25510 PR tree-optimization/81192
25511 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25513 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25515 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25517 2017-07-03 Tom de Vries <tom@codesourcery.com>
25519 PR tree-optimization/81192
25520 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25521 BB_SAME_SUCC (bb) == NULL.
25523 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25525 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25528 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25530 * dumpfile.c: Include profile-count.h
25531 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25533 (insert_cond_bb): Update profile.
25534 * tree-cfg.h (insert_cond_bb): Update prototype.
25535 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25536 * tree-dump.c: Do not include tree-cfg.
25538 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25540 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25542 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25544 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25547 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25549 * tree-complex.c (expand_complex_div_wide): update profile.
25551 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25552 Alan Hayward <alan.hayward@arm.com>
25553 David Sherwood <david.sherwood@arm.com>
25555 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25556 (CORETYPES_H): New define.
25557 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25558 (insn-modes-inline.h, s-modes-inline-h): New rules.
25559 (generated_files): Add insn-modes-inline.h.
25560 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25561 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25562 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25563 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25564 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25565 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25566 (build/gencodes.o, build/genconditions.o): Likewise.
25567 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25568 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25569 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25570 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25571 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25572 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25573 * coretypes.h: Include everything up to real.h for generators.
25574 Include insn-modes.h first. Include wide-int-print.h after
25575 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25576 * machmode.h: Don't include insn-modes.h here.
25577 * function-tests.c: Remove includes of signop.h, machmode.h,
25578 double-int.h and wide-int.h.
25580 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25582 * optc-save-gen.awk: Likewise.
25583 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25584 * godump.c: Remove include of wide-int-print.h.
25585 * pretty-print.h: Likewise.
25586 * wide-int-print.cc: Likewise.
25587 * wide-int.cc: Likewise.
25588 * hash-map-tests.c: Remove include of signop.h.
25589 * hash-set-tests.c: Likewise.
25590 * rtl-tests.c: Likewise.
25591 * mkconfig.sh: Remove include of machmode.h.
25592 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25594 (emit_insn_modes_inline_h): ...this new function. Emit the code
25595 into an insn-modes-inline.h header file, adding appropriate
25596 include guards and end comments.
25597 (emit_insn_modes_c_header): Remove include of machmode.h.
25598 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25600 (main): Handle -i flag and call emit_insn_modes_inline_h when
25603 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25605 * tree-ssa-strlen.c (strinfo): Rename the length field to
25606 nonzero_chars. Add a full_string_p field.
25607 (compare_nonzero_chars, zero_length_string_p): New functions.
25608 (get_addr_stridx): Add an offset_out parameter.
25609 Use compare_nonzero_chars.
25610 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25611 (new_strinfo): Update after above changes to strinfo.
25612 (set_endptr_and_length): Set full_string_p.
25613 (get_string_length): Update after above changes to strinfo.
25614 (unshare_strinfo): Update call to new_strinfo.
25615 (maybe_invalidate): Likewise.
25616 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25617 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25618 as a uhwi instead of an shwi. Update after above changes to
25619 strinfo and new_strinfo.
25620 (zero_length_string): Assert that chainsi contains full strings.
25621 Use zero_length_string_p. Update call to new_strinfo.
25622 (adjust_related_strinfos): Update after above changes to strinfo.
25623 Copy full_string_p from origsi.
25624 (adjust_last_stmt): Use zero_length_string_p.
25625 (handle_builtin_strlen): Update after above changes to strinfo and
25626 new_strinfo. Install the lhs as the string length if the previous
25627 entry didn't describe a full string.
25628 (handle_builtin_strchr): Update after above changes to strinfo
25630 (handle_builtin_strcpy): Likewise.
25631 (handle_builtin_strcat): Likewise.
25632 (handle_builtin_malloc): Likewise.
25633 (handle_pointer_plus): Likewise.
25634 (handle_builtin_memcpy): Likewise. Track nonzero characters
25635 that aren't necessarily followed by a nul terminator.
25636 (handle_char_store): Likewise.
25638 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25640 PR tree-optimization/80769
25641 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25642 for malloc and calloc. Document the new invariant that all related
25643 strinfos have delayed lengths or none do.
25644 (verify_related_strinfos): Move earlier in file.
25645 (set_endptr_and_length): New function, split out from...
25646 (get_string_length): ...here. Also set the lengths of related
25648 (zero_length_string): Assert that chainsi has known (rather than
25650 (adjust_related_strinfos): Likewise.
25652 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25654 PR tree-optimization/81136
25655 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25656 assert that two references with the same misalignment have the same
25657 compile-time misalignment if those compile-time misalignments
25660 2017-07-01 Andi Kleen <ak@linux.intel.com>
25662 * print-tree.c (print_node): Print all attributes.
25664 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25666 * cfg.c (scale_bbs_frequencies): New function.
25667 * cfg.h (scale_bbs_frequencies): Declare it.
25668 * cfgloopanal.c (single_likely_exit): Cleanup.
25669 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25671 (scale_loop_profile): Likewise.
25672 (loop_version): Likewise.
25673 (create_empty_loop_on_edge): Update.
25674 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25675 scale_loop_frequencies, scale_loop_profile, loopify,
25676 loop_version): Update prototypes.
25677 * modulo-sched.c (sms_schedule): Update.
25678 * predict.c (unlikely_executed_edge_p): Also check probability.
25679 (probably_never_executed_edge_p): Fix typo.
25680 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25681 * tree-parloops.c (gen_parallel_loop): Update.
25682 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25683 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25684 * tree-ssa-loop-split.c (split_loop): Update.
25685 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25686 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25687 (vect_loop_versioning): Update.
25688 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25690 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25692 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25694 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25696 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25697 to keep profile consistent.
25699 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25701 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25702 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25703 * profile-count.h (max_safe_multiplier): Make unsigned.
25704 (profile_count::guessed_zero): New.
25706 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25708 * bb-reorder.c (fix_up_crossing_landing_pad,
25709 fix_crossing_conditional_branches): Use make_single_succ_edge
25710 to keep profile consistent.
25712 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25714 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25717 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25720 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25721 the right scopes, make sure cond_jump isn't preserved between multiple
25722 iterations. Search for fallthru edge whenever there are 3+ edges and
25723 use find_fallthru_edge for it.
25725 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25727 Patch by Alexander Monakov <amonakov@ispras.ru>
25728 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25729 probabilities consistently.
25731 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25733 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25734 emit_cmp_and_jump_insns.
25736 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25739 * tree-inline.c (expand_call_inline): Combine profile statuses.
25741 2017-06-30 Andrew Pinski <apinski@cavium.com>
25743 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25744 fold_stmt returned true.
25746 2017-06-30 Nathan Sidwell <nathan@acm.org>
25748 * ggc.h (empty_string): Delete.
25749 * cfgexpand.c (expand_asm_stmt): Use plain "".
25750 * optabs.c (expand_asm_memory_barrier): Likewise.
25751 * stringpool.c (empty_string): Delete.
25752 (digit_vector, digit_string): Delete.
25753 (ggc_alloc_string): Use plain "", don't optimize single digit
25754 strings. Use ggc_alloc_atomic.
25756 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25758 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25759 comparison set and one other set, use the cost of the non-comparison
25762 2017-06-30 Nathan Sidwell <nathan@acm.org>
25764 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25767 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25769 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25770 loops. Remove now unneeded calls to gimple_switch_set_label() that
25771 just set removed labels to NULL_TREE.
25773 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25775 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25776 (set_range_info): ...here. Only call set_range_info_raw if domain
25778 (set_nonzero_bits): Call set_range_info_raw.
25779 * tree-ssanames.h (set_range_info_raw): New.
25781 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25784 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25785 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25786 of nonimmediate_operand and <store_mask_constraint> instead of m
25787 for the input operand. For V8FI iterator, always split if input
25788 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25789 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25790 <store_mask_predicate> instead of register_operand and
25791 <store_mask_constraint> instead of v for the input operand. Make
25792 sure both operands aren't MEMs for if not <mask_applied>.
25794 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25796 * lto-wrapper.c (copy_file) Close both file descriptors before
25799 2017-06-30 Martin Liska <mliska@suse.cz>
25802 * multiple_target.c (create_dispatcher_calls): Make ifunc
25803 also for function that don't have calls or are not referenced.
25805 2017-06-30 Richard Biener <rguenther@suse.de>
25807 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25808 analyze the first scalar stmt. Move vector type computation
25809 for the BB case here from ...
25810 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25811 live operation processing in the SLP case properly.
25813 2017-06-30 Richard Biener <rguenther@suse.de>
25815 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25817 2017-06-30 Martin Liska <mliska@suse.cz>
25820 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25821 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25823 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25825 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25828 2017-06-30 Richard Biener <rguenther@suse.de>
25830 PR tree-optimization/81249
25831 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25832 condition reduction result to original scalar type.
25834 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25836 * profile-count.h (enum profile_quality): Fix typos and whitespace
25839 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25841 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25842 type for branch probabilities.
25844 2017-06-29 Julian Brown <julian@codesourcery.com>
25845 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25847 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25848 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25849 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25850 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25852 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25854 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25855 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25856 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25857 CC usage from generic code to here.
25858 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25859 CC usage into the target macros.
25861 2017-06-29 Maya Rashish <coypu@sdf.org>
25863 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25866 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25868 * arm/arm-builtins.c: Include profile-count.h
25869 * except.c (sjlj_emit_function_enter): Use
25870 profile_probability::unlikely.
25872 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25874 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25875 and tocrel_offset be pointer args rather than implicitly using
25877 (legitimate_constant_pool_address_p, rs6000_emit_move,
25878 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25879 tocrel_offset and use in toc_relative_expr_p call.
25880 (print_operand, print_operand_address): Use static tocrel_base_oac
25881 and tocrel_offset_oac.
25882 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25885 2017-06-29 Maya Rashish <coypu@sdf.org>
25887 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25889 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25891 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25892 objects, take into account only the alignment of 'op0' and 'mode1' if
25895 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25897 * ccmp.c (ccmp_tree_comparison_p): New function.
25898 (ccmp_candidate_p): Update to use above function.
25899 (get_compare_parts): New function.
25900 (expand_ccmp_next): Update to use new functions.
25901 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25903 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25904 take mode as argument.
25905 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25906 * expr.c (expand_expr_real_1): Pass mode as argument.
25908 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25910 * combine.c (combine_instructions): Print insns to dump_file, together
25913 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25915 * asan.c (asan_emit_stack_protection): Update.
25916 (create_cond_insert_point): Update.
25917 * auto-profile.c (afdo_propagate_circuit): Update.
25918 * basic-block.h (struct edge_def): Turn probability to
25919 profile_probability.
25920 (EDGE_FREQUENCY): Update.
25921 * bb-reorder.c (find_traces_1_round): Update.
25922 (better_edge_p): Update.
25923 (sanitize_hot_paths): Update.
25924 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25925 (make_single_succ_edge): Update.
25926 (check_bb_profile): Update.
25927 (dump_edge_info): Update.
25928 (update_bb_profile_for_threading): Update.
25929 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25930 probabilitycount to 0.
25931 * cfgbuild.c (compute_outgoing_frequencies): Update.
25932 * cfgcleanup.c (try_forward_edges): Update.
25933 (outgoing_edges_match): Update.
25934 (try_crossjump_to_edge): Update.
25935 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25936 (expand_gimple_tailcall): Update.
25937 (construct_init_block): Use make_single_succ_edge.
25938 (construct_exit_block): Use make_single_succ_edge.
25939 * cfghooks.c (verify_flow_info): Update.
25940 (redirect_edge_succ_nodup): Update.
25941 (split_edge): Update.
25942 (account_profile_record): Update.
25943 * cfgloopanal.c (single_likely_exit): Update.
25944 * cfgloopmanip.c (scale_loop_profile): Update.
25945 (set_zero_probability): Remove.
25946 (duplicate_loop_to_header_edge): Update.
25947 * cfgloopmanip.h (loop_version): Update prototype.
25948 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25949 (force_nonfallthru_and_redirect): Update.
25950 (update_br_prob_note): Update.
25951 (rtl_verify_edges): Update.
25952 (purge_dead_edges): Update.
25953 (rtl_lv_add_condition_to_bb): Update.
25954 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25955 * cgraphunit.c (init_lowered_empty_function): Update.
25956 (cgraph_node::expand_thunk): Update.
25957 * cilk-common.c: Include profile-count.h
25958 * dojump.c (inv): Remove.
25959 (jumpifnot): Update.
25960 (jumpifnot_1): Update.
25961 (do_jump_1): Update.
25963 (do_jump_by_parts_greater_rtx): Update.
25964 (do_compare_rtx_and_jump): Update.
25965 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25966 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25967 * dwarf2cfi.c: Include profile-count.h
25968 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25969 (sjlj_emit_dispatch_table): Likewise.
25970 * explow.c: Include profile-count.h
25971 * expmed.c (emit_store_flag_force): Update.
25972 (do_cmp_and_jump): Update.
25973 * expr.c (compare_by_pieces_d::generate): Update.
25974 (compare_by_pieces_d::finish_mode): Update.
25975 (emit_block_move_via_loop): Update.
25976 (store_expr_with_bounds): Update.
25977 (store_constructor): Update.
25978 (expand_expr_real_2): Update.
25979 (expand_expr_real_1): Update.
25980 * expr.h (try_casesi, try_tablejump): Update prototypes.
25981 * gimple-pretty-print.c (dump_probability): Update.
25982 (dump_profile): New.
25983 (dump_gimple_label): Update.
25984 (dump_gimple_bb_header): Update.
25985 * graph.c (draw_cfg_node_succ_edges): Update.
25986 * hsa-gen.c (convert_switch_statements): Update.
25987 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25988 (find_if_case_1): Update.
25989 (find_if_case_2): Update.
25990 * internal-fn.c (expand_arith_overflow_result_store): Update.
25991 (expand_addsub_overflow): Update.
25992 (expand_neg_overflow): Update.
25993 (expand_mul_overflow): Update.
25994 (expand_vector_ubsan_overflow): Update.
25995 * ipa-cp.c (good_cloning_opportunity_p): Update.
25996 * ipa-split.c (split_function): Use make_single_succ_edge.
25997 * ipa-utils.c (ipa_merge_profiles): Update.
25998 * loop-doloop.c (add_test): Update.
25999 (doloop_modify): Update.
26000 * loop-unroll.c (compare_and_jump_seq): Update.
26001 (unroll_loop_runtime_iterations): Update.
26002 * lra-constraints.c (lra_inheritance): Update.
26003 * lto-streamer-in.c (input_cfg): Update.
26004 * lto-streamer-out.c (output_cfg): Update.
26005 * mcf.c (adjust_cfg_counts): Update.
26006 * modulo-sched.c (sms_schedule): Update.
26007 * omp-expand.c (expand_omp_for_init_counts): Update.
26008 (extract_omp_for_update_vars): Update.
26009 (expand_omp_ordered_sink): Update.
26010 (expand_omp_for_ordered_loops): Update.
26011 (expand_omp_for_generic): Update.
26012 (expand_omp_for_static_nochunk): Update.
26013 (expand_omp_for_static_chunk): Update.
26014 (expand_cilk_for): Update.
26015 (expand_omp_simd): Update.
26016 (expand_omp_taskloop_for_outer): Update.
26017 (expand_omp_taskloop_for_inner): Update.
26018 * omp-simd-clone.c (simd_clone_adjust): Update.
26019 * optabs.c (expand_doubleword_shift): Update.
26020 (expand_abs): Update.
26021 (emit_cmp_and_jump_insn_1): Update.
26022 (expand_compare_and_swap_loop): Update.
26023 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26024 * predict.c (predictable_edge_p): Update.
26025 (edge_probability_reliable_p): Update.
26026 (set_even_probabilities): Update.
26027 (combine_predictions_for_insn): Update.
26028 (combine_predictions_for_bb): Update.
26029 (propagate_freq): Update.
26030 (estimate_bb_frequencies): Update.
26031 (force_edge_cold): Update.
26032 * profile-count.c (profile_count::dump): Add missing space into dump.
26033 (profile_count::debug): Add newline.
26034 (profile_count::differs_from_p): Explicitly convert to unsigned.
26035 (profile_count::stream_in): Update.
26036 (profile_probability::dump): New member function.
26037 (profile_probability::debug): New member function.
26038 (profile_probability::differs_from_p): New member function.
26039 (profile_probability::differs_lot_from_p): New member function.
26040 (profile_probability::stream_in): New member function.
26041 (profile_probability::stream_out): New member function.
26042 * profile-count.h (profile_count_quality): Rename to ...
26043 (profile_quality): ... this one.
26044 (profile_probability): New.
26045 (profile_count): Update.
26046 * profile.c (compute_branch_probabilities): Update.
26047 * recog.c (peep2_attempt): Update.
26048 * sched-ebb.c (schedule_ebbs): Update.
26049 * sched-rgn.c (find_single_block_region): Update.
26050 (compute_dom_prob_ps): Update.
26051 (schedule_region): Update.
26052 * sel-sched-ir.c (compute_succs_info): Update.
26053 * stmt.c (struct case_node): Update.
26054 (do_jump_if_equal): Update.
26055 (get_outgoing_edge_probs): Update.
26056 (conditional_probability): Update.
26057 (emit_case_dispatch_table): Update.
26058 (expand_case): Update.
26059 (expand_sjlj_dispatch_table): Update.
26060 (emit_case_nodes): Update.
26061 * targhooks.c: Update.
26062 * tracer.c (better_p): Update.
26063 (find_best_successor): Update.
26064 * trans-mem.c (expand_transaction): Update.
26065 * tree-call-cdce.c: Update.
26066 * tree-cfg.c (gimple_split_edge): Upate.
26067 (move_sese_region_to_fn): Upate.
26068 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26069 * tree-eh.c (lower_resx): Upate.
26070 (cleanup_empty_eh_move_lp): Upate.
26071 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26072 * tree-inline.c (copy_edges_for_bb): Update.
26073 (copy_cfg_body): Update.
26074 * tree-parloops.c (gen_parallel_loop): Update.
26075 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26076 (gimple_gen_time_profiler): Update.
26077 * tree-ssa-dce.c (remove_dead_stmt): Update.
26078 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26079 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26080 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26081 (unloop_loops): Update.
26082 (try_peel_loop): Update.
26083 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26084 * tree-ssa-loop-split.c (connect_loops): Update.
26085 (split_loop): Update.
26086 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26087 (hoist_guard): Update.
26088 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26089 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26090 (value_replacement): Update.
26091 * tree-ssa-reassoc.c (branch_fixup): Update.
26092 * tree-ssa-tail-merge.c (replace_block_by): Update.
26093 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26094 (create_edge_and_update_destination_phis): Update.
26095 (compute_path_counts): Update.
26096 (recompute_probabilities): Update.
26097 (update_joiner_offpath_counts): Update.
26098 (freqs_to_counts_path): Update.
26099 (duplicate_thread_path): Update.
26100 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26101 (struct switch_conv_info): Update.
26102 (gen_inbound_check): Update.
26103 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26104 (vect_do_peeling): Update.
26105 (vect_loop_versioning): Update.
26106 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26107 (optimize_mask_stores): Update.
26108 * ubsan.c (ubsan_expand_null_ifn): Update.
26109 * value-prof.c (gimple_divmod_fixed_value): Update.
26110 (gimple_divmod_fixed_value_transform): Update.
26111 (gimple_mod_pow2): Update.
26112 (gimple_mod_pow2_value_transform): Update.
26113 (gimple_mod_subtract): Update.
26114 (gimple_mod_subtract_transform): Update.
26115 (gimple_ic): Update.
26116 (gimple_stringop_fixed_value): Update.
26117 (gimple_stringops_transform): Update.
26118 * value-prof.h: Update.
26120 2017-06-29 Carl Love <cel@us.ibm.com>
26122 * config/rs6000/rs6000-c.c: Add support for built-in functions
26123 vector signed int vec_signed (vector float);
26124 vector signed long long vec_signed (vector double);
26125 vector signed int vec_signed2 (vector double, vector double);
26126 vector signed int vec_signede (vector double);
26127 vector signed int vec_signedo (vector double);
26128 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26129 instruction generator.
26130 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26131 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26132 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26134 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26135 vunsignede_v2df): Add define_expands.
26136 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26137 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26138 VEC_UNSIGNEDO): Add definitions.
26139 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26140 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26141 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26142 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26143 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26144 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26145 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26146 vec_unsignede, vec_unsignedo): Add builtin defines.
26147 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26149 * doc/extend.texi: Update the built-in documentation file for the
26150 new built-in functions.
26152 2017-06-29 Richard Biener <rguenther@suse.de>
26154 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26155 reduction chains to LOOP_VINFO_REDUCTIONS.
26156 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26157 SLP reductions after processing reduction chains.
26159 2017-06-29 Nathan Sidwell <nathan@acm.org>
26161 * builtins.c (fold_builtin_FUNCTION): Use
26162 lang_hooks.decl_printable_name.
26164 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26166 PR middle-end/81194
26167 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26168 with only one label.
26169 * stmt.c (expand_case): Assert NCASES is greater than one.
26171 2017-06-29 Richard Biener <rguenther@suse.de>
26173 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26175 (group_case_labels): Likewise.
26176 (find_taken_edge): Push sanity checking on val to workers...
26177 (find_taken_edge_cond_expr): ... here
26178 (find_taken_edge_switch_expr): ... and here, handle cases
26179 with just a default label.
26180 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26181 (group_case_labels): Likewise.
26182 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26183 group_case_labels does anything cleanup the CFG again.
26185 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26187 PR tree-optimization/81196
26188 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26189 exit condition comparing two IVs.
26191 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26193 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26194 profile to the dummy entry at the end of the list of architectures.
26195 * config/arm/arm-cpu-cdata.h: Regenerated.
26197 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26198 Michael Collison <michael.collison@arm.com>
26201 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26203 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26204 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26205 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26206 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26207 of shift when the shift amount is masked with constant equal to
26208 the size of the mode.
26209 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26212 2017-06-29 Martin Liska <mliska@suse.cz>
26214 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26216 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26219 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26222 (edge_growth_cache_entry::edge_growth_cache_entry): New
26224 (reset_edge_growth_cache): Update to use constructor.
26226 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26228 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26229 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26230 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26232 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26234 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26235 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26237 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26239 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26240 (*-linux-uclibc*): Add t-uclibc tmake_file.
26241 * config/t-musl: New.
26242 * config/t-uclibc: New.
26244 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26246 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26248 (gen_comm_data): Emit architectural setting of arch_prof.
26249 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26251 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26252 (armv8-m.base, armv8-m.main): Likewise.
26253 * arm-protos.h (arm_build_target): Add profile field.
26254 (arch_option): Likewise.
26255 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26257 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26258 arm_active_target.profile.
26260 2017-06-28 Richard Biener <rguenther@suse.de>
26262 PR middle-end/81227
26263 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26264 TYPE_OVERFLOW_WRAPS.
26265 * match.pd (negate_expr_p): Likewise.
26266 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26267 fold_build2, not fold_binary.
26269 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26271 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26272 Convert memory address to Pmode.
26273 (aarch64_print_operand): Assert MEM operands are always Pmode.
26275 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26278 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26279 Remove redundant if.
26280 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26281 * config/arm/aarch-common-protos.h
26282 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26283 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26285 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26288 * multiple_target.c (create_dispatcher_calls): Set the default
26289 clone to be static, not public.
26291 2017-06-28 Richard Biener <rguenther@suse.de>
26293 * tree-vect-loop.c (vectorizable_reduction): Move special
26294 cond reduction IV var creation ...
26295 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26296 parameter. Use STMT_VINFO_VECTYPE.
26297 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26300 2017-06-28 Martin Liska <mliska@suse.cz>
26303 * ipa-visibility.c (non_local_p): Handle visibility.
26305 2017-06-28 Martin Liska <mliska@suse.cz>
26308 * common.opt: Add IntegerRange to various options.
26309 * opt-functions.awk (integer_range_info): New function.
26310 * optc-gen.awk: Add integer_range_info to cl_options struct.
26311 * opts-common.c (decode_cmdline_option): Handle
26312 CL_ERR_INT_RANGE_ARG.
26313 (cmdline_handle_error): Likewise.
26314 * opts.c (print_filtered_help): Show valid interval in
26315 when --help is provided.
26316 * opts.h (struct cl_option): Add range_min and range_max fields.
26317 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26319 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26321 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26322 (x * C EQ/NE y * C): New transformation.
26324 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26326 * genmultilib (combination_space): Accept '+' in option names.
26328 2017-06-28 Martin Liska <mliska@suse.cz>
26331 * asan.c (instrument_derefs): Bail out inner references
26332 that are hard register variables.
26334 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26337 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26338 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26340 2017-06-28 Richard Biener <rguenther@suse.de>
26342 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26343 (vect_get_slp_defs): Adjust.
26344 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26346 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26348 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26349 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26350 (vectorizable_reduction): Adjust.
26351 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26353 (vect_get_slp_defs): Likewise.
26354 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26355 (vectorizable_bswap): Adjust.
26356 (vectorizable_call): Likewise.
26357 (vectorizable_conversion): Likewise.
26358 (vectorizable_assignment): Likewise.
26359 (vectorizable_shift): Likewise.
26360 (vectorizable_operation): Likewise.
26361 (vectorizable_store): Likewise.
26362 (vectorizable_condition): Likewise.
26363 (vectorizable_comparison): Likewise.
26365 2017-06-28 Michael Collison <michael.collison@arm.com>
26368 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26370 (arm_gen_movmemqi): Removed unused variable 'i'.
26371 Convert 'for' loop into 'while' loop.
26372 (arm_expand_prologue): Remove last unnecessary set of insn.
26373 (thumb_pop): Remove unused variable 'pushed_words'.
26374 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26376 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26378 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26379 * config/s390/s390-protos.h: Add prototype of
26380 s390_rel_address_ok_p.
26381 * config/s390/s390.c (s390_got_symbol): New function.
26382 (s390_rel_address_ok_p): New function.
26383 (legitimize_pic_address): Use s390_rel_address_ok_p.
26384 (s390_load_got): Use s390_got_symbol.
26385 (s390_option_override): Issue error if
26386 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26387 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26389 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26391 2017-06-27 Andrew Pinski <apinski@cavium.com>
26393 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26394 (X * copysign (1.0, X)): New pattern.
26395 (X * copysign (1.0, -X)): New pattern.
26396 (copysign (-1.0, CST)): New pattern.
26398 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26400 * genmultilib (combination_space): Remove variable.
26401 Validate reuse rules against regular expression for any sequence
26402 of multilib options in any order.
26404 2017-06-27 Michael Collison <michael.collison@arm.com>
26406 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26407 call aarch64_split_simd_combine.
26408 * (aarch64_combine_internal<mode>): Delete pattern.
26409 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26410 Allow register and subreg operands.
26412 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26414 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26415 specific need, just fallback on defaults.
26416 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26418 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26419 Olivier Hainque <hainque@adacore.com>
26421 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26423 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26424 targets. Pick a default if no particular attempt applied.
26425 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26428 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26430 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26431 (x86_64-wrs-vxworks7): Likewise.
26433 2017-06-27 Marek Polacek <polacek@redhat.com>
26436 * ubsan.c (instrument_null): Check get_base_address's result for null.
26438 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26440 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26442 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26444 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26445 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26446 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26447 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26448 New function types.
26449 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26450 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26451 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26452 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26453 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26454 BUILT_IN_FEUPDATEENV): New builtins.
26455 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26456 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26457 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26458 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26460 (builtin_structptr_types): Adjust size.
26461 * tree.c (builtin_structptr_types): Add four entries.
26463 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26464 Olivier Hainque <hainque@adacore.com>
26466 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26467 (TLS_SYM): New local macro, forcing reference to __tls__ on
26468 link command lines for VxWorks 7 RTPs, triggering initialization
26470 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26471 OS features TLS support, true for RTPs on VxWorks 7.
26472 * config/vxworks.c (vxworks_override_options): Setup emutls
26475 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26477 * predict.c (test_prediction_value_range): Use -1U instead of -1
26478 to avoid narrowing conversion warning.
26479 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26480 to avoid narrowing conversion warning.
26481 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26483 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26485 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26487 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26488 64bit configurations.
26489 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26490 (SIZE_TYPE): Likewise.
26491 * config/vxworks.c (vxworks_emutls_var_fields): Use
26492 long_unsigned_type_node instead of unsigned_type_node as the offset
26493 field type, which is "pointer" mode in emutls.c.
26495 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26498 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26500 PR middle-end/81207
26501 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26502 gimple_vuse copying separately from gimple_vdef copying.
26504 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26506 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26509 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26510 Olivier Hainque <hainque@adacore.com>
26512 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26513 for all vxworks7 targets.
26514 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26515 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26516 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26517 variations for VX6/VX7 and 32/64bits later on in ...
26518 (VXWORKS_LIB_SPEC): Leverage new macros.
26519 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26520 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26522 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26524 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26525 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26527 2017-06-26 Carl Love <cel@us.ibm.com>
26529 * config/rs6000/rs6000-c.c: Add support for built-in functions
26530 vector bool char vec_reve (vector bool char);
26531 vector signed char vec_reve (vector signed char);
26532 vector unsigned char vec_reve (vector unsigned char);
26533 vector bool int vec_reve (vector bool int);
26534 vector signed int vec_reve (vector signed int);
26535 vector unsigned int vec_reve (vector unsigned int);
26536 vector bool long long vec_reve (vector bool long long);
26537 vector signed long long vec_reve (vector signed long long);
26538 vector unsigned long long vec_reve (vector unsigned long long);
26539 vector bool short vec_reve (vector bool short);
26540 vector signed short vec_reve (vector signed short);
26541 vector double vec_reve (vector double);
26542 vector float vec_reve (vector float);
26543 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26544 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26545 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26546 (altivec_vreve): New pattern.
26547 * config/rs6000/altivec.h (vec_reve): New define.
26548 * doc/extend.texi (vec_rev): Update the built-in documentation file
26549 for the new built-in functions.
26551 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26553 PR tree-optimization/71815
26554 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26556 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26558 (slsr_process_phi): Likewise.
26559 (replace_uncond_cands_and_profitable_phis): Don't replace a
26560 multiply candidate with a stride of 1 (copy or cast).
26561 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26563 (lowest_cost_path): Likewise.
26564 (total_savings): Likewise.
26566 2017-06-26 Richard Biener <rguenther@suse.de>
26569 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26570 Use def_builtin_pure for all gather builtins.
26572 2017-06-26 Richard Biener <rguenther@suse.de>
26574 PR tree-optimization/81203
26575 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26576 non-dominating BBs.
26578 2017-06-26 Marek Polacek <polacek@redhat.com>
26581 * doc/invoke.texi: Document -Wmultistatement-macros.
26583 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26585 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26586 arm_neon_ok_no_float_abi effective target.
26588 2017-06-26 Richard Biener <rguenther@suse.de>
26590 PR tree-optimization/80928
26591 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26592 (copy_bbs): Set BB_DUPLICATED flag early.
26593 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26595 (execute_on_shrinking_pred): Likewise.
26596 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26597 BB_DUPLICATED blocks.
26598 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26599 iterate over all PHIs considering removal of *gsi.
26601 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26603 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26606 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26608 * config/rs6000/rs6000-string.c: (expand_block_clear,
26609 do_load_for_compare, select_block_compare_mode,
26610 compute_current_alignment, expand_block_compare,
26611 expand_strncmp_align_check, expand_strn_compare,
26612 expand_block_move, rs6000_output_load_multiple)
26613 Move functions related to string/block move/compare
26614 to a separate file.
26615 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26616 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26617 for this function which is now used in two files.
26618 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26619 * config.gcc: Add rs6000-string.o to extra_objs for
26620 targets powerpc*-*-* and rs6000*-*-*.
26622 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26625 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26626 32-bit, since indexed is not valid for DImode.
26627 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26628 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26629 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26630 (define_peephole2 for Altivec d-form store): Likewise.
26633 * multiple_target.c (create_dispatcher_calls): Only create the
26634 dispatcher call if the function is the default clone of a
26635 versioned function.
26637 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26639 PR middle-end/80902
26640 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26641 a call, force the call to not be a tail call.
26643 2017-06-23 Jeff Law <law@redhat.com>
26645 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26648 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26650 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26651 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26652 handling for early expansion of vector shifts (sl,sr,sra,rl).
26653 (builtin_function_type): Add vector shift right instructions
26654 to the unsigned argument list.
26656 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26658 rtl-optimizatoin/79286
26659 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26660 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26661 trap. PIC register plus a const unspec without offset can never trap.
26663 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26665 * tree.h (builtin_structptr_type): New type.
26666 (builtin_structptr_types): Declare new array.
26667 * tree.c (builtin_structptr_types): New array.
26668 (free_lang_data, build_common_tree_nodes): Use it.
26670 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26673 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26676 2017-06-22 Matt Turner <mattst88@gmail.com>
26678 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26679 Lake models to skylake case. Assume skylake for unknown
26680 models with clflushopt.
26682 2017-06-22 Jeff Law <law@redhat.com>
26684 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26685 frame sizes that do not satisfy aarch64_uimm12_shift.
26687 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26689 * profile-count.h (apply_probability,
26690 apply_scale, probability_in): Fix checks for zero.
26692 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26694 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26695 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26697 2016-06-22 Richard Biener <rguenther@suse.de>
26699 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26700 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26701 REDUC_MAX_EXPR support.
26702 (vectorizable_reduction): Likewise.
26703 (vect_create_epilog_for_reduction): Likewise.
26705 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26707 * match.pd (A / (1 << B) -> A >> B): New.
26708 * generic-match-head.c: Include optabs-tree.h.
26709 * gimple-match-head.c: Likewise.
26710 * optabs-tree.h (target_supports_op_p): New.
26711 * optabs-tree.c (target_supports_op_p): New.
26713 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26715 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26716 $gcc_cv_ld --help output.
26717 (gcc_cv_ld_demangle): Likewise.
26718 (gcc_cv_ld_eh_frame_hdr): Likewise.
26719 (gcc_cv_ld_pie): Likewise.
26720 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26721 (gcc_cv_ld_buildid): Likewise.
26722 (gcc_cv_ld_sysroot): Likewise.
26723 (ld_bndplt_support): Likewise.
26724 (ld_pushpopstate_support): Likewise.
26725 * configure: Regenerate.
26726 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26728 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26731 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26732 operands indexes to avoid gap between operands and match_dups.
26734 2017-06-21 Andrew Pinski <apinski@cavium.com>
26736 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26737 Increment Arith_shift and Arith_shift_reg by 1.
26738 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26740 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26741 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26742 (aarch64_strip_extend): Add new argument and test for it.
26743 (aarch64_cheap_mult_shift_p): New function.
26744 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26745 add a cost if it is true.
26746 Update calls to aarch64_strip_extend.
26747 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26749 2017-06-21 Andrew Pinski <apinski@cavium.com>
26751 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26753 (thunderxt88): Likewise.
26754 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26755 (thunderx_prefetch_tune): New variable.
26756 (thunderx2t99_prefetch_tune): Update for the correct values.
26757 (thunderxt88_tunings): New variable.
26758 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26759 generic_prefetch_tune.
26760 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26762 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26764 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26765 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26766 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26767 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26768 (aarch64_atomic_cas<mode>, GPI): Likewise.
26770 2017-06-21 Martin Liska <mliska@suse.cz>
26772 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26773 statements on cold and hot labels.
26774 * predict.c (tree_estimate_probability_bb): Remove the
26775 prediction from this place.
26777 2017-06-21 Martin Liska <mliska@suse.cz>
26779 PR tree-optimization/79489
26780 * gimplify.c (maybe_add_early_return_predict_stmt): New
26782 (gimplify_return_expr): Call the function.
26783 * predict.c (tree_estimate_probability_bb): Remove handling
26785 * predict.def: Update comment about early return predictor.
26786 * gimple-predict.h (is_gimple_predict): New function.
26787 * predict.def: Change default value of early return to 66.
26788 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26790 * passes.def: Put pass_strip_predict_hints to the beginning of
26793 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26795 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26796 FUNCTION_DECL declarations.
26797 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26799 (dwaf2out_decl): Likewise.
26800 * godump.c (go_early_global_decl): Skip call to the real debug hook
26801 for FUNCTION_DECL declarations.
26802 * passes.c (rest_of_decl_compilation): Skip call to the
26803 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26804 -fdump-go-spec is passed.
26806 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26808 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26809 Reorder for compactness.
26810 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26811 (def_builtin_pure, def_builtin_pure2): New functions.
26812 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26814 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26816 * match.pd (nop_convert): New predicate.
26817 ((A +- CST1) +- CST2): Allow some NOP conversions.
26819 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26822 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26823 with ctors/dtors if GOVD_SHARED is set.
26825 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26827 * config/aarch64/aarch64.md (movti_aarch64):
26828 Emit mov rather than orr.
26829 (movtf_aarch64): Likewise.
26830 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26831 Emit mov rather than orr.
26833 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26835 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26836 Swap alternatives, make integer dup more expensive.
26838 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26840 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26841 Return true for non-tls symbols.
26843 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26845 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26846 (cortex-a75): Likewise.
26847 (cortex-a75.cortex-a55): Likewise.
26848 * config/aarch64/aarch64-tune.md: Regenerate.
26849 * doc/invoke.texi (-mtune): Document new values for -mtune.
26851 2017-06-21 Tom de Vries <tom@codesourcery.com>
26853 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26854 stack_size feature.
26855 (Effective-Target Keywords, Other attributes): Suggest using
26856 dg-add-options stack_size feature to get stack limit in stack_size
26857 effective target documentation.
26859 2017-06-21 Julian Brown <julian@codesourcery.com>
26860 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26862 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26863 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26864 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26866 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26867 attribute type list for neon_multiply.
26868 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26869 attribute type list for neon_multiply.
26870 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26871 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26872 attribute type list for neon_multiply.
26873 * config/arm/types.md (crypto_pmull): Add.
26874 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26875 attribute type list.
26877 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26879 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26882 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26884 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26885 to make sure not to dereference a NULL cost_classes_ptr pointer.
26887 2017-06-20 Carl Love <cel@us.ibm.com>
26889 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26890 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26891 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26892 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26893 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26894 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26895 VMULOSW): New enum "unspec" values.
26896 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26897 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26898 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26899 altivec_vmulosw): New patterns.
26900 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26901 VMULOSW): Add definitions.
26903 2017-06-20 Julia Koval <julia.koval@intel.com>
26905 * config/i386/i386.c: Fix rounding expand for new pattern.
26906 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26908 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26910 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26911 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26913 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26915 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26918 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26920 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26921 architecture, then by implementer ID.
26922 * config/aarch64/aarch64-tune.md: Regenerate.
26924 2017-06-20 Richard Biener <rguenther@suse.de>
26926 PR middle-end/81097
26927 * fold-const.c (split_tree): Fold to type before negating.
26929 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26931 * diagnostic-show-locus.c
26932 (selftest::test_fixit_deletion_affecting_newline): New function.
26933 (selftest::diagnostic_show_locus_c_tests): Call it.
26935 2017-06-20 Andreas Schwab <schwab@suse.de>
26938 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26941 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26943 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26944 __ARM_FEATURE_COPROC according to support.
26946 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26948 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26949 Rewritten to avoid overflow for > 32-bit pointers.
26952 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26953 by removing enum keyword.
26954 (ubsan_type_descriptor): Likewise. Formatting fix.
26957 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26958 splitter): Require TARGET_SSE2 in the condition.
26960 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26963 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26964 for doing vector set of SFmode on ISA 3.0.
26965 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26966 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26968 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26969 SFmode value into a V4SF variable that was extracted from another
26970 V4SF variable without converting the element to double precision
26971 and back to single precision vector format.
26972 (vsx_insert_extract_v4sf_p9_2): Likewise.
26974 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26976 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26977 in UWHI to avoid undefined overflow.
26980 * ubsan.h (enum ubsan_encode_value_phase): New.
26981 (ubsan_encode_value): Change second argument to
26982 enum ubsan_encode_value_phase with default value of
26983 UBSAN_ENCODE_VALUE_GENERIC.
26984 * ubsan.c (ubsan_encode_value): Change second argument to
26985 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26986 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26987 create_tmp_var_raw instead of create_tmp_var and use a
26989 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26990 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26991 ubsan_encode_value callers.
26994 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26995 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26996 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26998 2017-06-19 Richard Biener <rguenther@suse.de>
27000 PR middle-end/81118
27001 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27002 estimates if we changed anything.
27004 2017-06-19 Richard Biener <rguenther@suse.de>
27006 PR tree-optimization/80887
27007 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27008 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27009 simplified lookups, then reset mprts_hook.
27010 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27012 (try_to_simplify): Likewise.
27014 2017-06-19 Martin Liska <mliska@suse.cz>
27017 * gimplify.c (gimplify_switch_expr):
27018 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27020 2017-06-19 Martin Liska <mliska@suse.cz>
27022 * doc/install.texi: Document that PGO runs in 4 stages.
27024 2017-06-19 Martin Liska <mliska@suse.cz>
27027 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27028 to dispatcher function name.
27029 * multiple_target.c (replace_function_decl): New function.
27030 (create_dispatcher_calls): Redirect both edges and references.
27032 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27034 * profile-count.c (profile_count::dump): Dump quality.
27035 (profile_count::differs_from_p): Update for unsigned val.
27036 * profile-count.h (profile_count_quality): New enum.
27037 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27039 2017-06-19 Richard Biener <rguenther@suse.de>
27041 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27042 struct function as arg.
27043 (estimate_numbers_of_iterations): Export overload with loop arg.
27044 (free_numbers_of_iterations_estimates_loop): Use an overload of
27045 free_numbers_of_iterations_estimates instead.
27046 * tree-cfg.c (remove_bb): Adjust.
27047 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27048 * tree-parloops.c (gen_parallel_loop): Likewise.
27049 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27051 (tree_unroll_loops_completely): Likewise.
27052 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27053 Use an overload instead and export.
27054 (estimated_loop_iterations): Adjust.
27055 (max_loop_iterations): Likewise.
27056 (likely_max_loop_iterations): Likewise.
27057 (estimate_numbers_of_iterations): Take struct function as arg
27059 (loop_exits_before_overflow): Adjust.
27060 (free_numbers_of_iterations_estimates_loop): Use an overload.
27061 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27062 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27064 2017-06-19 Richard Biener <rguenther@suse.de>
27067 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27068 RANGE_EXPR conservatively.
27070 2017-06-16 Carl Love <cel@us.ibm.com>
27072 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27073 definitions for vec_float, vec_float2, vec_floato,
27074 vec_floate built-ins.
27075 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27076 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27078 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27079 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27080 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27081 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27082 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27083 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27084 vec_floato): Add builtin defines.
27085 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27086 Update the built-in documentation file for the new built-in
27089 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27091 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27092 (mthumb): Mark as the negative of -marm.
27094 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27096 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27098 (ARM Options, -mtune): Document that this accepts the same
27099 extension options as -mcpu.
27100 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27102 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27104 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27105 permitted extensions.
27107 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27109 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27110 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27111 (armv8-m.main): Add option +nodsp.
27112 * config/arm/arm-cpu-cdata.h: Regenerated.
27114 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27116 * config/arm/t-fuchsia: New file.
27117 * config.gcc (arm*-*-fuchsia*): Use it.
27119 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27121 * config/arm/t-symbian: Rewrite for new option infrastructure.
27123 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27125 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27126 (MULTILIB_REQUIRED): Likewise.
27128 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27130 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27131 (MULTILIB_RESUE): Likewise.
27132 (MULTILIB_MATCHES): Likewise.
27133 (MULTLIB_REQUIRED): Likewise.
27135 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27137 * config/arm/t-rtems: Rewrite for new option framework.
27139 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27141 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27142 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27143 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27144 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27145 * config/arm/t-multilib: ... here.
27146 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27147 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27148 armv7-a and armv8*-a when A-profile libraries have not been built.
27149 * config/arm/t-rmprofile: Rewrite.
27151 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27153 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27154 with a backslash. Remove the backslash after substituting unescaped
27156 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27158 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27160 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27161 the driver to pass through the default setting of -mfloat-abi.
27162 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27164 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27165 (all_feat_combs): New rule.
27166 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27168 * config/arm/t-aprofile: Rewrite.
27170 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27172 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27173 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27174 fpu is not specified by the user/command-line.
27175 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27176 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27177 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27178 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27179 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27180 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27184 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27185 * config/arm/t-arm-elf: Rewritten.
27187 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27189 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27190 have some floating-point instructions.
27191 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27192 (TARGET_MAYBE_HARD_FLOAT): New macro.
27193 * config/arm/arm-builtins.c (arm_init_builtins): Use
27194 TARGET_MAYBE_HARD_FLOAT.
27195 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27197 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27199 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27200 (configargs.h): Include it.
27201 (arm_print_hint_for_fpu_option): New function.
27202 (arm_parse_fpu_option): New function.
27203 (candidate_extension): New class.
27204 (arm_canon_for_multilib): New function.
27205 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27206 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27207 (ARCH_CANONICAL_SPECS): New macro.
27208 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27210 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27212 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27213 are set after handling multilib fragments. Set target_cpu_default2
27216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27218 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27220 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27221 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27223 (check_arch): Likewise.
27224 * config/arm/arm.c (arm_configure_build_target): Handle
27225 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27226 options in the default.
27228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27230 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27231 when an option is an alias of another.
27232 * config/arm/parsecpu.awk (optalias): New parser token.
27233 (gen_comm_data): Mark non-alias options as such. Emit entries
27234 for extension aliases.
27235 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27236 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27237 (armv6kz, armv6zk, armv6t2): Likewise.
27238 (armv7): Make vfpv3-d16 an alias.
27239 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27241 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27242 Sort in canonical order.
27243 (armv8-a): Sort in canonical order.
27244 (armv8.1-a, armv8.2-a): Likewise.
27245 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27247 (cortex-a9): Sort in canonical order.
27248 * config/arm/arm.c (selftests.h): Include it.
27249 (arm_test_cpu_arch_data): New function.
27250 (arm_run_self_tests): New function.
27251 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27252 (targetm): Move declaration to the end of the file.
27253 * arm-cpu-cdata.h: Regenerated.
27255 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27257 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27258 call to target_mode_check describing the type of option passed.
27259 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27260 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27261 arm_parse_cpu_option_name to match parameters against list of
27263 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27264 arm_arch_core_flags data structure.
27265 * config/arm/arm-cpu_cdata.h: Regenerated.
27267 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27269 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27271 (arm_print_hint_for_cpu_option): Likewise.
27272 (arm_print_hint_for_arch_option): Likewise.
27273 (arm_parse_cpu_option_name): Likewise.
27274 (arm_parse_arch_option_name): Likewise.
27275 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27276 of entries in the all_fpus list.
27277 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27278 (arm_parse_cpu_option_name): Declare.
27279 (arm_parse_arch_option_name): Declare.
27280 (arm_parse_option_features): Declare.
27281 (arm_intialize_isa): Declare.
27282 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27284 (gen_comm_data): ... here. Make definitions non-static.
27285 * config/arm/arm-cpu-data.h: Regenerated.
27286 * config/arm/arm-cpu-cdata.h: Regenerated.
27288 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27290 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27291 (cpu_arch_extension): New structure.
27292 (cpu_arch_option, arch_option, cpu_option): New structures.
27293 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27294 architecture types.
27295 (gen_data): Generate new format data tables.
27296 * config/arm/arm.c (cpu_tune): New structure.
27297 (cpu_option, processors): Delete.
27298 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27299 (arm_print_hint_for_cpu_option): ... this and ...
27300 (arm_print_hint_for_arch_option): ... this.
27301 (arm_parse_arch_cpu_name): Delete. Replace with ...
27302 (arm_parse_cpu_option_name): ... this and ...
27303 (arm_parse_arch_option_name): ... this.
27304 (arm_unrecognized_feature): Change type of target parameter to
27306 (arm_parse_arch_cpu_features): Delete. Replace with ...
27307 (arm_parse_option_features): ... this.
27308 (arm_configure_build_target): Rework to use new configuration data
27310 (arm_print_tune_info): Rework for new configuration data tables.
27311 * config/arm/arm-cpu-data.h: Regenerated.
27312 * config/arm/arm-cpu.h: Regenerated.
27314 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27316 * Makefile.in (OBJS): Move sbitmap.o from here ...
27317 (OBJS-libcommon): ... to here.
27319 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27321 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27322 (ISA_ALL_CRYPTO): New macro.
27323 (ISA_ALL_SIMD): New macro
27324 (ISA_ALL_FP): New macro.
27325 * config/arm/arm.c (fpu_bitlist): Update initializer.
27326 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27328 (arm9e): Add fpu. Add option for nofp
27329 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27330 (arm926ej-s, arm1026ej-s): Likewise.
27331 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27332 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27333 neon-fp16, neon-vfpv4, nofp and nosimd.
27334 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27335 (cortex-a8): Add fpu. Add option for nofp.
27336 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27337 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27338 (cortex-r4f): Add fpu.
27339 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27340 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27342 (cortex-r8): Likewise.
27343 (cortex-m4): Add fpu. Add option for nofp.
27344 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27345 (cortex-a17.cortex-a7): Likewise.
27346 (cortex-a32): Add fpu. Add options for crypto and nofp.
27347 (cortex-a35, cortex-a53): Likewise.
27348 (cortex-a57): Add fpu. Add option for crypto.
27349 (cortex-a72, cortex-a73): Likewise.
27350 (exynos-m1): Likewise.
27351 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27352 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27353 (cortex-m33): Add fpu. Add option for nofp.
27354 * config/arm/arm-cpu-cdata.h: Regenerated
27355 * config/arm/arm-cpu-data.h: Regenerated.
27357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27359 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27360 (armv5te, armv5tej): Likewise.
27361 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27362 (armv7): Add options fp and vfpv3-d16.
27363 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27364 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27366 (armv7ve): Likewise.
27367 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27368 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27369 (armv8-a): Add nocrypto option.
27370 (armv8.1-a, armv8.2-a): Likewise.
27371 (armv8-m.main): add options fp, fp.dp and nofp.
27373 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27375 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27377 (armv8-a+crc): Delete.
27378 (armv8.1-a): Add options simd, crypto and nofp.
27379 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27380 (armv8.2-a+fp16): Delete.
27381 (armv8-m.main): Add option dsp.
27382 (armv8-m.main+dsp): Delete.
27383 (cortex-a8): Add fpu. Add nofp option.
27384 (cortex-a9): Add fpu. Add nofp and nosimd options.
27385 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27386 link to main cpu and architecture data structures.
27387 (gen_comm_data): Only put isa attributes from the main architecture
27389 (option): New statement for architecture and CPU entries.
27390 * arm.c (struct cpu_option): New structure.
27391 (struct processors): Add entry for options.
27392 (arm_unrecognized_feature): New function.
27393 (arm_parse_arch_cpu_name): Ignore any characters after the first
27395 (arm_parse_arch_cpu_feature): New function.
27396 (arm_configure_build_target): Separate out any CPU and architecture
27397 features and parse separately. Don't error out if -mfpu=auto is
27398 used with only an architecture string.
27399 (arm_print_asm_arch_directives): New function.
27400 (arm_file_start): Call it.
27401 * config/arm/arm-cpu-cdata.h: Regenerated.
27402 * config/arm/arm-cpu-data.h: Likewise.
27403 * config/arm/arm-tables.opt: Likewise.
27405 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27407 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27408 assembler when it is not -mfpu=auto.
27410 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27412 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27413 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27414 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27415 (ASM_CPU_SPEC): Rewrite.
27416 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27417 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27418 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27419 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27420 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27421 copied string is NUL-terminated. Also strip any characters prefixed
27423 (arm_rewrite_selected_arch): New function.
27424 (arm_rewrite_march): New function.
27426 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27428 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27429 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27430 (march, mcpu, mtune): Convert to string-based options.
27431 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27432 (arm_parse_arch_cpu_name): New function.
27433 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27434 identify selected architecture or CPU.
27435 (arm_option_save): New function.
27436 (TARGET_OPTION_SAVE): Redefine.
27437 (arm_option_restore): Restore string options.
27438 (arm_option_print): Print string options.
27440 2017-06-16 Martin Sebor <msebor@redhat.com>
27442 PR tree-optimization/80933
27443 PR tree-optimization/80934
27444 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27445 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27446 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27447 (gimple_fold_builtin): Call them.
27449 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27451 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27452 as unlikely; update profile.
27454 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27456 * predict.c (force_edge_cold): Handle declaring edges impossible
27459 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27461 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27463 (try_unroll_loop_completely): Fix reporting.
27465 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27467 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27469 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27472 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27473 if given a non-constant argument for an intrinsic which requires a
27476 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27478 * profile.c (compare_freqs): New function.
27479 (branch_prob): Sort edge list.
27480 (find_spanning_tree): Assume that the list is priority sorted.
27482 2017-06-16 Richard Biener <rguenther@suse.de>
27484 PR tree-optimization/81090
27485 * passes.def (pass_record_bounds): Remove.
27486 * tree-pass.h (make_pass_record_bounds): Likewise.
27487 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27488 make_pass_record_bounds): Likewise.
27489 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27490 not free niter estimates at the beginning but at the end.
27491 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27493 2017-06-16 Richard Biener <rguenther@suse.de>
27495 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27496 initializer to workaround ICE in host GCC 4.8.
27498 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27500 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27502 (clone_inlined_nodes): Update.
27504 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27506 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27507 prefetch settings, and enable prefetching by default at -O3.
27509 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27511 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27512 Set flag_prefetch_loop_arrays according to tuning data.
27514 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27516 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27517 New tune structure.
27518 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27519 [Unrelated to main purpose of the patch] Place the pointer field last
27520 to enable type checking errors when tune structure are wrongly merged.
27521 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27522 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27523 (thunderx2t99_prefetch_tune): New tune constants.
27524 (tune_params *_tunings): Update all tunings (no functional change).
27525 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27526 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27527 from tunings structures.
27529 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27532 * ubsan.c (instrument_null): Add T argument, use it instead
27533 of computing it based on IS_LHS.
27534 (instrument_object_size): Likewise.
27535 (pass_ubsan::execute): Adjust instrument_null and
27536 instrument_object_size callers to pass gimple_get_lhs or
27537 gimple_assign_rhs1 result to it. Use instrument_null instead of
27538 calling get_base_address and instrument_mem_ref. Handle
27539 aggregate call arguments for object-size sanitization.
27541 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27543 PR tree-optimization/81089
27544 * tree-vrp.c (is_masked_range_test): Validate operands of
27547 2017-06-15 Martin Sebor <msebor@redhat.com>
27550 * dumpfile.c (dump_register): Avoid calling memset to initialize
27551 a class with a default ctor.
27552 * gcc.c (struct compiler): Remove const qualification.
27553 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27554 * hash-table.h: Ditto.
27555 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27557 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27558 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27560 * params.h (struct param_info): Make struct members non-const.
27561 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27562 with default initialization.
27563 * vec.h (vec_copy_construct, vec_default_construct): New helper
27565 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27566 with vec_copy_construct.
27567 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27568 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27569 * doc/invoke.texi (-Wclass-memaccess): Document.
27571 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27573 * emit-rtl.h (is_leaf): Update comment about local
27574 register allocator.
27576 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27579 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27580 for a variable to have a section before checking if the section has a
27582 Set section to.persistent if persistent attribute is set.
27583 Warn if .persistent attribute is used on an automatic variable.
27585 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27587 PR rtl-optimization/80474
27588 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27590 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27592 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27595 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27597 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27598 (casesi): Emit bounds checking as RTL.
27599 (casesi_internal_mips16_<mode>): Remove bounds checking.
27601 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27603 * config/xtensa/xtensa.c (xtensa_option_override): Append
27604 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27605 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27606 xtensa_doloop_hooks): Define unconditionally.
27607 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27609 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27610 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27611 for it in xtensa_option_override.
27612 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27613 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27615 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27617 * doc/cppopts.texi: Document '-' special value to -MF.
27619 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27621 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27622 (cortex_a53_fconst): Likewise.
27623 (cortex_a53_fpmul): Likewise.
27624 (cortex_a53_f_load_64): Likewise.
27625 (cortex_a53_f_load_many): Likewise.
27626 (cortex_a53_advsimd_alu): Likewise.
27627 (cortex_a53_advsimd_alu_q): Likewise.
27628 (cortex_a53_advsimd_mul): Likewise.
27629 (cortex_a53_advsimd_mul_q): Likewise.
27630 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27631 Add missing fmul, r2f_cvt and fconst cases.
27633 2017-06-14 Richard Biener <rguenther@suse.de>
27635 PR middle-end/81088
27636 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27638 (fold_binary_loc): When associating do not treat pre-existing
27639 TREE_OVERFLOW on literal constants as a reason to allow
27640 TREE_OVERFLOW on associated literal constants.
27642 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27644 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27645 (MASK_FEATURES): New macro.
27646 * config/sparc/sparc.c (sparc_option_override): Remove the special
27647 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27649 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27651 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27652 a division of 0 if non-call exceptions are enabled.
27654 2017-06-14 Andrew Pinski <apinski@cavium.com>
27655 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27658 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27659 Improve vector initialization code gen for only variable case.
27661 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27663 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27665 2017-06-14 Richard Biener <rguenther@suse.de>
27667 PR tree-optimization/81083
27668 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27671 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27673 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27674 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27675 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27676 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27677 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27678 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27680 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27682 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27683 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27685 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27687 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27689 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27691 * config/rs6000/t-rtems: Don't handle SPE.
27693 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27695 * config/rs6000/t-linux: Don't handle SPE.
27697 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27699 * config/rs6000/eabispe.h: Delete file.
27701 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27703 * config/rs6000/t-spe: Delete file.
27705 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27707 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27708 (rs6000_legitimate_offset_address_p): Return false for anything in
27709 V2SImode or V2SFmode.
27711 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27713 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27714 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27715 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27717 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27718 (rs6000_legitimate_offset_address_p): Ditto.
27719 (rs6000_emit_move): Ditto.
27720 (rs6000_init_builtins): Remove V4HI_type_node.
27722 2017-06-13 Martin Liska <mliska@suse.cz>
27725 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27726 (gate_asan): Likewise.
27727 * asan.h (asan_no_sanitize_address_p): Remove the function.
27728 (sanitize_flags_p): New function.
27729 * builtins.def: Fix coding style.
27730 * common.opt: Use renamed enum value.
27731 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27732 * doc/extend.texi: Document no_sanitize attribute.
27733 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27734 to SANITIZE_UNDEFINED_NONDEFAULT.
27735 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27736 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27737 Use sanitize_flags_p.
27738 * gimplify.c (gimplify_function_tree): Likewise.
27739 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27740 * opts.c (parse_no_sanitize_attribute): New function.
27741 (common_handle_option): Use renamed enum value.
27742 * opts.h (parse_no_sanitize_attribute): Declare.
27743 * tree.c (sanitize_flags_p): New function.
27744 * tree.h: Declared here.
27745 * tsan.c: Use sanitize_flags_p.
27746 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27747 (instrument_mem_ref): Likewise.
27748 (instrument_bool_enum_load): Likewise.
27749 (do_ubsan_in_current_function): Remove the function.
27750 (pass_ubsan::execute): Use sanitize_flags_p.
27751 * ubsan.h: Remove do_ubsan_in_current_function
27752 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27753 (dump_function_to_file): Use it here.
27755 2017-06-13 Martin Jambor <mjambor@suse.cz>
27757 PR tree-optimization/80803
27758 PR tree-optimization/81063
27759 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27760 (propagate_subaccesses_across_link): Enqueue subtree whenever
27761 necessary instead of relying on the caller.
27763 2017-06-13 Martin Jambor <mjambor@suse.cz>
27765 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27766 that have a first_link.
27767 (sort_and_splice_var_accesses): Do not check first_link before
27769 (subtree_mark_written_and_enqueue): Likewise.
27770 (propagate_all_subaccesses): Likewise and do not stop at first
27771 parent with a first_link.
27773 2017-06-13 Martin Jambor <mjambor@suse.cz>
27775 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27778 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27780 * match.pd: New pattern.
27782 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27784 * tree-vrp.c (is_masked_range_test): New function.
27785 (register_edge_assert_for): Determine ranges for
27788 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27790 PR tree-optimization/67328
27791 * fold-const.c (maskable_range_p): New function.
27792 (build_range_check): Generate bittests if possible.
27794 2017-06-13 Martin Liska <mliska@suse.cz>
27796 * gimple-pretty-print.c (dump_probability): Add new argument.
27797 (dump_edge_probability): Dump both probability and count.
27798 (dump_gimple_label): Likewise.
27799 (dump_gimple_bb_header): Likewise.
27801 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27804 * config/avr/avr-devices.c: Fix indentation.
27805 * config/avr/gen-avr-mmcu-specs.c: Dito.
27807 2017-06-13 Richard Biener <rguenther@suse.de>
27809 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27810 instead get vector type from stmt_info.
27811 (vectorizable_reduction): Adjust. Remove dead code.
27813 2017-06-13 Richard Biener <rguenther@suse.de>
27815 PR middle-end/81065
27816 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27817 case of C * (x * C2 + C3).
27818 (fold_addr_of_array_ref_difference): Properly fold index difference.
27820 2017-06-12 David S. Miller <davem@davemloft.net>
27823 * config/sparc/sparc.md (return expander): Emit frame blockage if
27824 function uses alloca.
27826 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27828 * combine.c (make_field_assignment): Check len rather than the mode
27829 precision when calling force_to_mode.
27831 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27833 Support multilibs and devices that see flash in RAM address range.
27836 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27837 (avr_mcu_t) <flash_pm_offset>: New field.
27838 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27839 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27840 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27841 (AVR_TINY_PM_OFFSET): Remove macro.
27842 * config/avr/avr.opt (-mshort-calls): New option.
27843 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27844 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27845 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27846 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27847 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27848 instead of avr_arch->have_jmp_call.
27849 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27850 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27851 avr_arch->flash_pm_offset to define.
27852 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27853 new field flash_pm_offset. Add entry for avrxmega3.
27854 (avr_texinfo): Add entry for avrxmega3.
27855 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27856 attiny212, attiny214,
27857 attiny412, attiny414, attiny416, attiny417,
27858 attiny814, attiny816, attiny817,
27859 attiny1614, attiny1616, attiny1617,
27860 attiny3214, attiny3216, attiny3217.
27861 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27862 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27863 (avr_print_operand_address) [AVR_TINY]: Same.
27864 (avr_asm_init_sections) <readonly_data_section>: Only patch
27865 callback if avr_arch->flash_pm_offset = 0.
27866 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27867 for rodata if avr_arch->flash_pm_offset != 0.
27868 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27869 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27870 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27871 (m_options): Append opt_rcall.
27872 (m_dirnames): Append dir_rcall.
27873 * config/avr/t-multilib: Regenerate.
27875 * configure.ac [target=avr]: Check whether avrxmega3 default
27876 linker description file works as needed.
27877 * configure: Regenerate.
27878 * doc/avr-mmcu.texi: Regenerate.
27879 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27880 <__AVR_ARCH__>: Document avrxmega3 and 103.
27881 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27882 <__AVR_SHORT_CALLS__>: Document it.
27883 <__AVR_PM_BASE_ADDRESS__>: Document it.
27884 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27885 (AVR Variable Attributes) <progmem>: Document this is
27886 not needed for avrxmega3.
27887 (AVR Named Address Spaces) <__flash>: Dito.
27889 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27891 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27893 2017-06-12 Doug Rupp <rupp@adacore.com>
27895 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27896 Append vxworks-stdint.h to the tm_file list.
27897 * config/vxworks-stdint.h: New file.
27899 2017-06-12 Martin Liska <mliska@suse.cz>
27901 PR tree-optimization/81041
27902 * tree-profile.c (gimple_gen_ic_func_profiler):
27903 Create an extra BB in profile-generate
27904 (gimple_gen_time_profiler): Likewise.
27906 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27908 PR tree-optimization/81003
27909 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27910 (update_range_test): Use it instead of force_gimple_operand_gsi.
27912 2017-06-12 Richard Biener <rguenther@suse.de>
27914 PR tree-optimization/81053
27915 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27916 with backedge value not defined in loop. Simplify def stmt
27919 2017-06-11 Tom de Vries <tom@codesourcery.com>
27922 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27924 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27925 nvptx_cannot_force_const_mem.
27927 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27929 * opts.c (finish_options): Move test for flag_split_stack after
27930 it has been initialized.
27932 2017-06-11 Jason Merrill <jason@redhat.com>
27934 * tree.h (id_equal): New.
27935 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27936 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27937 instead of strcmp of IDENTIFIER_POINTER.
27939 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27941 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27942 (mark_all_inlined_calls_cdtor): Fix formating.
27943 (inline_transform): Rescale profile before inlining.
27945 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27947 * cgraph.h (cgraph_edge::clone): Update prototype.
27948 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27949 (cgraph_node::create_clone): Update.
27950 (cgraph_node::create_version_clone): Update.
27951 * tree-inline.c (copy_bb): Update.
27952 (expand_call_inline): Update.
27954 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27956 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27957 factored out from ...
27958 (rs6000_emit_prologue): ... here.
27960 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27962 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27963 factored out from ...
27964 (rs6000_emit_prologue): ... here.
27966 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27968 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27970 (handle_missing_profiles): Fix computation of tp_first_run.
27971 (counts_to_freqs): Do not touch freqs when count is 0.
27973 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27975 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27978 2017-06-10 Tom de Vries <tom@codesourcery.com>
27980 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27981 attributes): Document signal effective target.
27983 2017-06-10 Tom de Vries <tom@codesourcery.com>
27985 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27986 Document effective target stack_size.
27988 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27990 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27991 to the edit_context if they can be auto-applied.
27993 2017-06-9 Ian Lance Taylor <iant@golang.org>
27995 * opts.c (finish_options): If -fsplit-stack, disable implicit
27996 -forder-blocks-and-partition.
27997 * doc/invoke.texi (Optimize Options): Document that when using
27998 -fsplit-stack -forder-blocks-and-partition is not implicitly
28001 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28003 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28004 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28005 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28006 * builtins.def (abort, trap, unreachable): Declare cold.
28007 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28008 * tree-core.h (ECF_COLD): New.
28009 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28010 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28012 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28014 * predict.c (unlikely_executed_stmt_p): Cleanup.
28016 2017-06-09 Richard Biener <rguenther@suse.de>
28018 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28019 model if the ref is always written to.
28021 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28023 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28025 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28027 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28030 2017-06-09 Tom de Vries <tom@codesourcery.com>
28033 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28034 "target cannot support label values" when encountering LABEL_REF.
28036 2017-06-09 Martin Liska <mliska@suse.cz>
28038 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28039 (gimple_gen_ic_func_profiler): Emit direct comparison
28040 of __gcov_indirect_call_callee with NULL.
28041 (gimple_gen_time_profiler): Change probability from
28042 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28044 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28046 * profile.c (edge_gcov_counts): Turn to pointer.
28047 (compute_branch_probabilities, compute_branch_probabilities): Update.
28048 (branch_prob): Do not clear edge_gcov_count.
28049 * profile.h (edge_gcov_counts): Turn to pointer.
28050 (edge_gcov_count): Update.
28052 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28054 * gimple.h (gimple_check_failed): Mark cold.
28056 2017-06-09 Richard Biener <rguenther@suse.de>
28058 PR tree-optimization/66623
28059 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28060 refactor check_reduction into two parts, properly computing
28061 whether we have to check reduction validity for outer loop
28064 2017-06-09 Richard Biener <rguenther@suse.de>
28066 PR tree-optimization/79483
28067 * graphite-scop-detection.c (order): New global.
28068 (get_order): Compute bb to order mapping that satisfies code
28069 generation constraints.
28070 (cmp_pbbs): New helper.
28071 (build_scops): Start domwalk at entry block, sort generated
28074 2017-06-09 Richard Biener <rguenther@suse.de>
28076 PR middle-end/81007
28077 * ipa-polymorphic-call.c
28078 (ipa_polymorphic_call_context::restrict_to_inner_class):
28079 Skip FIELD_DECLs with error_mark_node type.
28080 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28083 2017-06-09 Martin Liska <mliska@suse.cz>
28085 * predict.c (struct branch_predictor): New struct.
28086 (test_prediction_value_range): New test.
28087 (predict_c_tests): New function.
28088 * selftest-run-tests.c (selftest::run_tests): Run the function.
28089 * selftest.h: Declare new tests.
28091 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28094 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28095 gen_add3_insn did not fail.
28096 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28097 r0, construct that number in a temporary reg and add that reg to r0.
28098 If asked to put the result in r0 as well, fail.
28100 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28102 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28103 for early expansion of vec_eqv.
28105 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28107 PR middle-end/81005
28108 * ubsan.c (instrument_null): Avoid pointless code temporary.
28109 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28111 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28115 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28118 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28121 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28123 * predict.c (unlikely_executed_edge_p): Move ahead.
28124 (probably_never_executed_edge_p): Use it.
28126 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28128 PR middle-end/79988
28129 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28130 gimple_call_builtin_p call.
28132 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28134 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28135 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28136 rtl_check_failed_type2, rtl_check_failed_code1,
28137 rtl_check_failed_code2, rtl_check_failed_code_mode,
28138 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28139 rtvec_check_failed_bounds, rtl_check_failed_flag,
28140 _fatal_insn_not_found, _fatal_insn): Likewise.
28141 * tree.h (tree_contains_struct_check_failed,
28142 tree_check_failed, tree_not_check_failed,
28143 tree_class_check_failed, tree_range_check_failed,
28144 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28145 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28146 tree_operand_check_failed, omp_clause_check_failed,
28147 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28150 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28152 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28153 flag_branch_probabilities.
28154 * ipa-inline.c (edge_badness): Likewise.
28155 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28156 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28157 * predict.c (maybe_hot_frequency_p): Likewise.
28158 (probably_never_executed): Likewise.
28159 * sched-ebb.c (schedule_ebbs): Likewise.
28160 * sched-rgn.c (find_single_block_region): Likewise.
28161 * tracer.c (tail_duplicate): Likewise.
28163 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28165 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28166 longer requires x_flag_profile_use.
28168 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28170 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28171 instead of flag_reorder_blocks_and_partition.
28172 * dbxout.c (dbxout_function_end): Likewise.
28173 * dwarf2out.c (gen_subprogram_die): Likewise.
28174 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28175 * hw-doloop.c (reorg_loops): Likewise.
28176 * varasm.c (assemble_start_function,
28177 assemble_end_function): Likewise.
28178 (decide_function_section): Do not check for
28179 flag_reorder_blocks_and_partition.
28181 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28183 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28185 (chkp_get_hard_register_fake_addr_expr): Ditto.
28186 (chkp_build_addr_expr): Add check for hard reg case.
28187 (chkp_parse_array_and_component_ref): Ditto.
28188 (chkp_find_bounds_1): Ditto.
28189 (chkp_process_stmt): Don't generate bounds store for
28192 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28194 * predict.c (maybe_hot_bb_p): Do not check profile status.
28195 (maybe_hot_edge_p): Likewise.
28196 (probably_never_executed): Check for zero counts even if profile
28198 (unlikely_executed_edge_p): New function.
28199 (unlikely_executed_stmt_p): New function.
28200 (unlikely_executed_bb_p): New function.
28201 (set_even_probabilities): Use unlikely predicates.
28202 (combine_predictions_for_bb): Likewise.
28203 (predict_paths_for_bb): Likewise.
28204 (predict_paths_leading_to_edge): Likewise.
28205 (determine_unlikely_bbs): New function.
28206 (estimate_bb_frequencies): Use it.
28207 (compute_function_frequency): Use zero counts even if profile is
28209 * profile-count.h: Fix typo.
28211 2017-08-08 Julia Koval <julia.koval@intel.com>
28213 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28214 _mm512_mask_cvtsepi16_storeu_epi8,
28215 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28216 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28217 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28218 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28219 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28220 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28221 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28222 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28223 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28224 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28225 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28226 __builtin_ia32_pmovuswb256mem_mask,
28227 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28228 __builtin_ia32_pmovwb512mem_mask): New builtins.
28230 2017-08-08 Julia Koval <julia.koval@intel.com>
28232 PR target/73350,80862
28233 * config/i386/subst.md (round): Fix round pattern.
28234 * config/i386/i386.c (ix86_erase_embedded_rounding):
28235 Fix erasing rounding for the fixed pattern.
28237 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28239 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28241 2017-06-08 Martin Liska <mliska@suse.cz>
28243 PR gcov-profile/80911
28244 * gcov.c (block_info::block_info): New constructor.
28246 2017-06-07 Carl Love <cel@us.ibm.com>
28248 * config/rs6000/rs6000-c: The return type of the following
28249 built-in functions was implemented as int not long long. Fix sign
28250 of return value for the unsigned version of vec_mulo and vec_mule.
28251 vector unsigned long long vec_bperm (vector unsigned long long,
28252 vector unsigned char)
28253 vector signed long long vec_mule (vector signed int,
28255 vector unsigned long long vec_mule (vector unsigned int,
28256 vector unsigned int)
28257 vector signed long long vec_mulo (vector signed int,
28259 vector unsigned long long vec_mulo (vector unsigned int,
28260 vector unsigned int)
28261 * doc/extend.texi: Fix the documentation for the built-in
28264 2017-06-07 Carl Love <cel@us.ibm.com>
28267 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28270 2017-06-07 Carl Love <cel@us.ibm.com>
28272 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28273 support, Generate doublehv for signed int/float for BE case only.
28275 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28277 * doc/invoke.texi (mcx16): Rewrite.
28279 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28281 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28282 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28283 *mov<mode>_softfloat, and an anonymous splitter): Use
28284 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28286 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28288 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28290 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28291 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28292 (rs6000_debug_reg_global): Adjust.
28293 (rs6000_init_hard_regno_mode_ok): Adjust.
28294 (rs6000_dbx_register_number): Adjust.
28295 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28296 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28297 Remove SPE_ACC and SPEFSCR.
28298 (REG_ALLOC_ORDER): Ditto.
28299 (FRAME_POINTER_REGNUM): Change to 111.
28300 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28301 (REG_CLASS_NAMES): Ditto.
28302 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28303 (REGISTER_NAMES): Ditto.
28304 (ADDITIONAL_REG_NAMES): Ditto.
28305 (rs6000_reg_names): Ditto.
28306 * config/rs6000/rs6000.md: Renumber some register number
28309 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28311 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28313 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28314 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28316 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28317 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28318 Delete the SPE high registers.
28319 (REG_ALLOC_ORDER): Ditto.
28320 (enum reg_class): Remove SPE_HIGH_REGS.
28321 (REG_CLASS_NAMES): Ditto.
28322 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28323 (REGISTER_NAMES): Ditto.
28324 (rs6000_reg_names): Ditto.
28325 * doc/tm.texi.in: Remove SPE as example.
28326 * doc/tm.texi: Regenerate.
28328 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28330 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28331 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28332 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28333 * config/rs6000/rs6000.md (type): Remove "brinc".
28335 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28337 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28338 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28339 * config/rs6000/linuxspe.h: Delete file.
28340 * config/rs6000/rs6000.md: Don't include spe.md.
28341 * config/rs6000/spe.h: Delete file.
28342 * config/rs6000/spe.md: Delete file.
28343 * config/rs6000/t-rs6000: Remove spe.md.
28345 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28347 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28348 (reg_or_none500mem_operand): Delete.
28349 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28350 instead of reg_or_none500mem_operand.
28352 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28354 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28355 handling of SPE flags.
28356 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28358 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28360 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28362 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28363 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28364 paired_divv2sf3): Similar.
28365 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28366 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28367 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28369 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28370 Rename the paired_* instruction patterns.
28371 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28373 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28374 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28375 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28376 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28377 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28378 PAIRED_VECTOR_MODE.
28379 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28380 (spe_func_has_64bit_regs_p): Delete.
28381 (spe_expand_predicate_builtin): Delete.
28382 (spe_expand_evsel_builtin): Delete.
28383 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28384 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28385 (invalid_e500_subreg): Delete.
28386 (rs6000_legitimize_address): Always force_reg op2 as well, for
28387 paired single memory accesses.
28388 (rs6000_member_type_forces_blk): Delete.
28389 (rs6000_spe_function_arg): Delete.
28390 (rs6000_expand_unop_builtin): Delete SPE handling.
28391 (rs6000_expand_binop_builtin): Ditto.
28392 (spe_expand_stv_builtin): Delete.
28393 (bdesc_2arg_spe): Delete.
28394 (spe_expand_builtin): Delete.
28395 (spe_expand_predicate_builtin): Delete.
28396 (spe_expand_evsel_builtin): Delete.
28397 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28398 (spe_init_builtins): Delete.
28399 (spe_func_has_64bit_regs_p): Delete.
28400 (savres_routine_name): Delete "info" parameter. Adjust callers.
28401 (rs6000_emit_stack_reset): Ditto.
28402 (rs6000_dwarf_register_span): Delete.
28403 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28404 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28405 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28407 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28409 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28410 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28411 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28412 mulv2sf3, divv2sf3): Delete expanders.
28414 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28416 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28418 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28420 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28421 * config/rs6000/rs6000.c: Ditto.
28423 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28425 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28426 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28427 comparison_operator.
28429 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28431 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28432 * config/rs6000/rs6000.opt: Ditto.
28433 * config/rs6000/t-rtems: Ditto.
28435 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28437 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28438 TARGET_E500_SINGLE by 0, simplify.
28439 * config/rs6000/rs6000.c: Ditto.
28440 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28441 (spe_build_register_parallel): Delete.
28442 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28443 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28444 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28445 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28446 (E500_CONVERT): Delete.
28447 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28449 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28451 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28452 * config/rs6000/dfp.md: Ditto.
28453 (negdd2, *negdd2_fpr): Merge.
28454 (absdd2, *absdd2_fpr): Merge.
28455 (negtd2, *negtd2_fpr): Merge.
28456 (abstd2, *abstd2_fpr): Merge.
28457 * config/rs6000/e500.h: Delete file.
28458 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28459 TARGET_FPRS by 1 and simplify.
28460 * config/rs6000/rs6000-c.c: Ditto.
28461 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28462 TARGET_DF_SPE by 0.
28463 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28465 * config/rs6000/rs6000.md: Ditto.
28466 (floatdidf2, *floatdidf2_fpr): Merge.
28467 (move_from_CR_gt_bit): Delete.
28468 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28469 (E500_CR_IOR_COMPARE): Delete.
28470 (All patterns that require !TARGET_FPRS): Delete.
28471 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28473 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28475 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28477 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28479 * graphds.c (add_edge): Intitialize edge's attached data.
28480 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28481 pointer parameter. Call pointed function on each edge during
28482 graph traversing. Skip traversing the edge when the function
28484 (graphds_dfs, graphds_scc): Ditto.
28485 (for_each_edge): New parameter. Pass the new parameter to callback
28487 * graphds.h (skip_edge_callback): New function pointer type.
28488 (graphds_dfs, graphds_scc): New function pointer parameter.
28489 (graphds_edge_callback, for_each_edge): New parameter.
28491 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28493 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28494 out code checking if runtime alias check is possible to below ...
28495 Call the new function.
28496 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28497 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28499 2017-06-07 Marek Polacek <polacek@redhat.com>
28502 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28503 TYPE_OVERFLOW_WRAPS check.
28505 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28507 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28508 if versioning is required.
28509 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28510 peeling with the check for versioning.
28512 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28514 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28515 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28516 Set true to new parameter if new ssa variable is defined.
28517 (vect_gen_vector_loop_niters): Refactor. Set range information
28518 for the new vector loop bound variable.
28519 (vect_do_peeling): Ditto.
28521 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28523 * tree-affine.c (ssa.h): Include header file.
28524 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28525 has wrapping overflow behavior.
28527 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28529 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28531 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28533 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28534 (tree_to_aff_combination): ... here.
28536 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28538 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28539 reg_pressure model function.
28540 (ivopts_global_cost_for_size): Delete.
28541 (determine_set_costs, iv_ca_recount_cost): Call new model function
28542 ivopts_estimate_reg_pressure.
28544 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28546 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28547 expensive than udiv. Remove floating point cases from mod.
28549 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28551 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28552 Increase idiv cost.
28554 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28556 * config/aarch64/aarch64.md
28557 (copysignsf3): Fix mask generation.
28559 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28561 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28563 (class dump_manager): Add register_dumps method.
28564 * dumpfile.c: Include langhooks.h.
28565 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28566 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28567 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28568 (dump_manager::dump_register): Start with 512 entries instead of 32.
28569 (dump_manager::register_dumps): New method.
28570 * toplev.c (general_init): Instead of invoking register_dumps
28571 langhook, invoke register_dumps method on the dump manager.
28572 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28575 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28577 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28578 Say that # requires an associated define_split to exist, and that
28579 the define_split must be suitable for use after register allocation.
28581 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28583 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28584 (compute_outgoing_frequencies): Also initialize zero counts.
28585 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28586 around loops; preserve more of profile when nothing changes.
28588 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28590 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28592 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28593 * config/arm/arm-cpu-cdata.h: Regenerate.
28594 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28595 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28596 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28597 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28598 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28600 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28602 * config/arm/t-rmprofile: Likewise.
28603 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28605 2017-06-06 David S. Miller <davem@davemloft.net>
28608 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28609 blockage if function uses alloca.
28611 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28613 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28614 New "uid" fields to hold pretty-print IDs of group and ref.
28615 Memory references are now identified as <group_id>:<ref_id>
28616 instead of using [random] addresses.
28617 (dump_mem_details): Simplify, no functional change.
28618 (dump_mem_ref): Simplify and make output more concise.
28619 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28620 (find_or_create_group): Initialize group uid.
28621 (record_ref): Initialize ref uid. Improve debug output.
28622 (prune_group_by_reuse, should_issue_prefetch_p,)
28623 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28624 (mark_nontemporal_store, determine_loop_nest_reuse):
28625 Improve debug output.
28627 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28629 * dbgcnt.def (prefetch): New debug counter.
28630 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28631 (schedule_prefetches): Stop issueing prefetches if debug counter
28634 2017-06-06 Tom de Vries <tom@codesourcery.com>
28636 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28637 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28639 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28641 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28642 Use aarch64_reg_or_zero predicate for operand 4.
28643 (aarch64_compare_and_swap<mode> define_insn_and_split):
28644 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28645 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28647 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28649 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28650 (arm_compute_save_core_reg_mask): This.
28651 (thumb1_compute_save_reg_mask): Rename into ...
28652 (thumb1_compute_save_core_reg_mask): This.
28653 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28654 (arm_compute_frame_layout): Likewise.
28656 2017-06-06 Richard Biener <rguenther@suse.de>
28658 PR tree-optimization/80974
28659 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28660 keep or clear leaders SSA info.
28662 2017-06-06 Tom de Vries <tom@codesourcery.com>
28664 * config/nvptx/nvptx.c (split_mode_p): New function.
28665 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28667 2017-06-06 Tom de Vries <tom@codesourcery.com>
28669 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28671 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28674 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28677 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28679 * shrink-wrap.c (handle_simple_exit): Update profile.
28680 (try_shrink_wrapping): Upate profile.
28682 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28684 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28685 (tree_guess_outgoing_edge_probabilities): New.
28686 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28687 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28689 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28691 * ipa-split.c (split_function): Initialize return bb profile.
28693 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28695 * profile.c (compute_branch_probabilities): Also initialize
28696 EXIT_BLOCK profile.
28698 2017-06-06 Richard Biener <rguenther@suse.de>
28700 PR tree-optimization/80928
28701 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28702 (vect_analyze_loop_operations): Properly guard analysis for
28704 (vect_transform_loop): Likewise.
28705 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28706 (vect_model_induction_cost): Do not cost for pure SLP.
28707 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28708 of induction in inner loop vectorization.
28709 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28710 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28711 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28713 (vect_analyze_slp_cost_1): Cost induction.
28714 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28715 (vect_get_slp_vect_defs): Likewise.
28716 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28717 (vect_transform_stmt): Handle SLP reductions.
28718 * tree-vectorizer.h (vectorizable_induction): Adjust.
28720 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28722 * config/rs6000/rs6000.c (make_resolver_func): Update
28723 init_lowered_empty_function call.
28725 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28727 * doc/invoke.texi: Document the -fprofile-abs-path option.
28728 * common.opt (fprofile-abs-path): New option.
28729 * gcov-io.h (gcov_write_filename): Declare.
28730 * gcov-io.c (gcov_write_filename): New function.
28731 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28732 * profile.c (output_location): Likewise.
28734 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28736 * shring-wrap.c: Revert accidental commit.
28738 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28740 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28742 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28744 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28746 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28747 profile in callgraph edge.
28748 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28749 (apply_scale): Likewise.
28750 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28751 Also scale profile when inlining function with zero profile.
28752 (initialize_cfun): Update exit block profile even when it is zero.
28753 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28754 when profile is read.
28756 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28758 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28759 (CLONE_*): New constants to define the processors we can generate
28760 code for with the target_clone attribute.
28761 (rs6000_clone_map): New array to identify which clone processors
28762 the current program is running on.
28763 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28764 target_clone attribute.
28765 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28766 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28767 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28768 (cpu_expand_builtin): Add support for target_clone attribute.
28769 (rs6000_valid_attribute_p): Allow "default" attribute.
28770 (get_decl_name): New debug function to simplify printing the
28771 current function name in debugging statements.
28772 (rs6000_clone_priority): New functions to support the target_clone
28773 attribute, and be able to generate code to switch between ISA 2.05
28774 through ISA 3.0 (power6 through power9).
28775 (rs6000_compare_version_priority): Likewise.
28776 (rs6000_get_function_versions_dispatcher): Likewise.
28777 (make_resolver_func): Likewise.
28778 (add_condition_to_bb): Likewise.
28779 (dispatch_function_versions): Likewise.
28780 (rs6000_generate_version_dispatcher_body): Likewise.
28781 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28782 (fusion_gpr_load_p): Fix a spacing issue.
28783 * doc/extend.texi (Common Function Attributes): Document that the
28784 PowerPC supports the target_clone attribute.
28786 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28788 * config/arm/arm.h: explain F symbol found in description of ARM
28789 register allocation in its legend.
28791 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28793 * config/mips/frame-header-opt.c: Include profile-count.h.
28794 * config/riscv/riscv.c: Include profile-count.h
28796 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28798 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28800 (sm_set_flag_if_changed): Add bbs field.
28801 (execute_sm_if_changed_flag_set): Pass BBS.
28802 (execute_sm): Update.
28804 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28806 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28809 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28811 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28812 (peephole2): New peephole2 to emit the above.
28813 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28815 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28817 * config/aarch64/aarch64.c (define_peephole2 above
28818 *sub_<shift>_<mode>): New peephole.
28820 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28822 * config/i386/i386.c (make_resolver_func): Update.
28823 * Makefile.in: Add profile-count.h and profile-count.o
28824 * auto-profile.c (afdo_indirect_call): Update to new API.
28825 (afdo_set_bb_count): Update.
28826 (afdo_propagate_edge): Update.
28827 (afdo_propagate_circuit): Update.
28828 (afdo_calculate_branch_prob): Update.
28829 (afdo_annotate_cfg): Update.
28830 * basic-block.h: Include profile-count.h
28831 (struct edge_def): Turn count to profile_count.
28832 (struct basic_block_def): Likewie.
28833 (REG_BR_PROB_BASE): Move to profile-count.h
28834 (RDIV): Move to profile-count.h
28835 * bb-reorder.c (max_entry_count): Turn to profile_count.
28836 (find_traces): Update.
28837 (rotate_loop):Update.
28838 (connect_traces):Update.
28839 (sanitize_hot_paths):Update.
28840 * bt-load.c (migrate_btr_defs): Update.
28841 * cfg.c (RDIV): Remove.
28842 (init_flow): Use alloc_block.
28843 (alloc_block): Uninitialize count.
28844 (unchecked_make_edge): Uninitialize count.
28845 (check_bb_profile): Update.
28846 (dump_edge_info): Update.
28847 (dump_bb_info): Update.
28848 (update_bb_profile_for_threading): Update.
28849 (scale_bbs_frequencies_int): Update.
28850 (scale_bbs_frequencies_gcov_type): Update.
28851 (scale_bbs_frequencies_profile_count): New.
28852 * cfg.h (update_bb_profile_for_threading): Update.
28853 (scale_bbs_frequencies_profile_count): Declare.
28854 * cfgbuild.c (compute_outgoing_frequencies): Update.
28855 (find_many_sub_basic_blocks): Update.
28856 * cfgcleanup.c (try_forward_edges): Update.
28857 (try_crossjump_to_edge): Update.
28858 * cfgexpand.c (expand_gimple_tailcall): Update.
28859 (construct_exit_block): Update.
28860 * cfghooks.c (verify_flow_info): Update.
28861 (dump_bb_for_graph): Update.
28862 (split_edge): Update.
28863 (make_forwarder_block): Update.
28864 (duplicate_block): Update.
28865 (account_profile_record): Update.
28866 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28867 (get_estimated_loop_iterations): Update.
28868 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28869 (single_likely_exit): Update.
28870 * cfgloopmanip.c (scale_loop_profile): Update.
28872 (set_zero_probability): Update.
28873 (lv_adjust_loop_entry_edge): Update.
28874 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28875 (purge_dead_edges): Update.
28876 (rtl_account_profile_record): Update.
28877 * cgraph.c (cgraph_node::create): Uninitialize count.
28878 (symbol_table::create_edge): Uninitialize count.
28879 (cgraph_update_edges_for_call_stmt_node): Update.
28880 (cgraph_edge::dump_edge_flags): Update.
28881 (cgraph_node::dump): Update.
28882 (cgraph_edge::maybe_hot_p): Update.
28883 * cgraph.h: Include profile-count.h
28884 (create_clone), create_edge, create_indirect_edge): Update.
28885 (cgraph_node): Turn count to profile_count.
28886 (cgraph_edge0: Likewise.
28887 (make_speculative, clone): Update.
28888 (create_edge): Update.
28889 (init_lowered_empty_function): Update.
28890 * cgraphclones.c (cgraph_edge::clone): Update.
28891 (duplicate_thunk_for_node): Update.
28892 (cgraph_node::create_clone): Update.
28893 * cgraphunit.c (cgraph_node::analyze): Update.
28894 (cgraph_node::expand_thunk): Update.
28895 * final.c (dump_basic_block_info): Update.
28896 * gimple-streamer-in.c (input_bb): Update.
28897 * gimple-streamer-out.c (output_bb): Update.
28898 * graphite.c (print_global_statistics): Update.
28899 (print_graphite_scop_statistics): Update.
28900 * hsa-brig.c: Include basic-block.h.
28901 * hsa-dump.c: Include basic-block.h.
28902 * hsa-gen.c (T sum_slice): Update.
28903 (convert_switch_statements):Update.
28904 * hsa-regalloc.c: Include basic-block.h.
28905 * ipa-chkp.c (chkp_produce_thunks): Update.
28906 * ipa-cp.c (struct caller_statistics): Update.
28907 (init_caller_stats): Update.
28908 (gather_caller_stats): Update.
28909 (ipcp_cloning_candidate_p): Update.
28910 (good_cloning_opportunity_p): Update.
28911 (get_info_about_necessary_edges): Update.
28912 (dump_profile_updates): Update.
28913 (update_profiling_info): Update.
28914 (update_specialized_profile): Update.
28915 (perhaps_add_new_callers): Update.
28916 (decide_about_value): Update.
28917 (ipa_cp_c_finalize): Update.
28918 * ipa-devirt.c (struct odr_type_warn_count): Update.
28919 (struct decl_warn_count): Update.
28920 (struct final_warning_record): Update.
28921 (possible_polymorphic_call_targets): Update.
28922 (ipa_devirt): Update.
28923 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28924 * ipa-icf.c (sem_function::merge): Update.
28925 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28926 * ipa-inline.c (compute_uninlined_call_time): Update.
28927 (compute_inlined_call_time): Update.
28928 (want_inline_small_function_p): Update.
28929 (want_inline_self_recursive_call_p): Update.
28930 (edge_badness): Update.
28931 (lookup_recursive_calls): Update.
28932 (recursive_inlining): Update.
28933 (inline_small_functions): Update.
28934 (dump_overall_stats): Update.
28935 (dump_inline_stats): Update.
28936 * ipa-profile.c (ipa_profile_generate_summary): Update.
28937 (ipa_propagate_frequency): Update.
28938 (ipa_profile): Update.
28939 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28940 * ipa-utils.c (ipa_merge_profiles): Update.
28941 * loop-doloop.c (doloop_modify): Update.
28942 * loop-unroll.c (report_unroll): Update.
28943 (unroll_loop_runtime_iterations): Update.
28944 * lto-cgraph.c (lto_output_edge): Update.
28945 (lto_output_node): Update.
28946 (input_node): Update.
28947 (input_edge): Update.
28948 (merge_profile_summaries): Update.
28949 * lto-streamer-in.c (input_cfg): Update.
28950 * lto-streamer-out.c (output_cfg): Update.
28951 * mcf.c (create_fixup_graph): Update.
28952 (adjust_cfg_counts): Update.
28953 (sum_edge_counts): Update.
28954 * modulo-sched.c (sms_schedule): Update.
28955 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28956 * predict.c (maybe_hot_count_p): Update.
28957 (probably_never_executed): Update.
28958 (dump_prediction): Update.
28959 (combine_predictions_for_bb): Update.
28960 (propagate_freq): Update.
28961 (handle_missing_profiles): Update.
28962 (counts_to_freqs): Update.
28963 (rebuild_frequencies): Update.
28964 (force_edge_cold): Update.
28965 * predict.h: Include profile-count.h
28966 (maybe_hot_count_p, counts_to_freqs): UPdate.
28967 * print-rtl-function.c: Do not include cfg.h
28968 * print-rtl.c: Include basic-block.h
28969 * profile-count.c: New file.
28970 * profile-count.h: New file.
28971 * profile.c (is_edge_inconsistent): Update.
28972 (correct_negative_edge_counts): Update.
28973 (is_inconsistent): Update.
28974 (set_bb_counts): Update.
28975 (read_profile_edge_counts): Update.
28976 (compute_frequency_overlap): Update.
28977 (compute_branch_probabilities): Update; Initialize and deinitialize
28979 (branch_prob): Update.
28980 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28981 (edge_gcov_count): New.
28982 (bb_gcov_count): New.
28983 * shrink-wrap.c (try_shrink_wrapping): Update.
28984 * tracer.c (better_p): Update.
28985 * trans-mem.c (expand_transaction): Update.
28986 (ipa_tm_insert_irr_call): Update.
28987 (ipa_tm_insert_gettmclone_call): Update.
28988 * tree-call-cdce.c: Update.
28989 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28990 (gimple_duplicate_sese_tail): Update.
28991 (gimple_account_profile_record): Update.
28992 (execute_fixup_cfg): Update.
28993 * tree-inline.c (copy_bb): Update.
28994 (copy_edges_for_bb): Update.
28995 (initialize_cfun): Update.
28996 (freqs_to_counts): Update.
28997 (copy_cfg_body): Update.
28998 (expand_call_inline): Update.
28999 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29000 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29001 (try_unroll_loop_completely): Update.
29002 (try_peel_loop): Update.
29003 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29004 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29005 * tree-ssa-loop-split.c (connect_loops): Update.
29006 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29007 * tree-ssa-reassoc.c (branch_fixup): Update.
29008 * tree-ssa-tail-merge.c (replace_block_by): Update.
29009 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29010 (compute_path_counts): Update.
29011 (update_profile): Update.
29012 (recompute_probabilities): Update.
29013 (update_joiner_offpath_counts): Update.
29014 (estimated_freqs_path): Update.
29015 (freqs_to_counts_path): Update.
29016 (clear_counts_path): Update.
29017 (ssa_fix_duplicate_block_edges): Update.
29018 (duplicate_thread_path): Update.
29019 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29020 (struct switch_conv_info): Update.
29021 * tree-tailcall.c (decrease_profile): Update.
29022 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29023 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29024 * value-prof.c (check_counter): Update.
29025 (gimple_divmod_fixed_value): Update.
29026 (gimple_mod_pow2): Update.
29027 (gimple_mod_subtract): Update.
29028 (gimple_ic_transform): Update.
29029 (gimple_stringop_fixed_value): Update.
29030 * value-prof.h (gimple_ic): Update.
29032 2017-06-02 Carl Love <cel@us.ibm.com>
29034 * config/rs6000/rs6000-c: Add support for built-in functions
29035 vector double vec_doublee (vector signed int);
29036 vector double vec_doublee (vector unsigned int);
29037 vector double vec_doublee (vector float);
29038 vector double vec_doubleh (vector signed int);
29039 vector double vec_doubleh (vector unsigned int);
29040 vector double vec_doubleh (vector float);
29041 vector double vec_doublel (vector signed int);
29042 vector double vec_doublel (vector unsigned int);
29043 vector double vec_doublel (vector float);
29044 vector double vec_doubleo (vector signed int);
29045 vector double vec_doubleo (vector unsigned int);
29046 vector double vec_doubleo (vector float);.
29047 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29048 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29050 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29051 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29052 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29054 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29055 vec_doublel, vec_doubleh.
29056 * doc/extend.texi: Update the built-in documentation file for the
29057 new built-in functions.
29059 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29062 * ipa-inline-analysis.c (free_growth_caches): Set
29063 edge_removal_hook_holder to NULL after removing it.
29065 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29067 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29068 comparision with zero.
29070 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29071 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29072 for early expansion of vec_min and vec_max builtins.
29073 (builtin_function_type): Add min/max unsigned variants to those
29074 identified as having unsigned arguments.
29076 2017-06-02 Olivier Hainque <hainque@adacore.com>
29078 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29080 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29082 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29083 Use VALL_F16 iterator rather than VALL.
29085 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29087 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29088 Emit CBNZ inside loop when doing a strong exchange and comparing
29089 against zero. Generate the CC flags after the loop.
29091 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29093 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29094 (dl_section_ref): New.
29095 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29096 On AIX, append an expression to subtract the size of the
29097 section length to dl_section_ref.
29099 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29101 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29102 for early expansion of vector absolute builtins.
29104 2017-06-02 Richard Biener <rguenther@suse.de>
29106 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29107 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29109 2017-06-02 Richard Biener <rguenther@suse.de>
29111 PR tree-optimization/80948
29112 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29113 stmt order as well.
29115 2017-06-02 Richard Biener <rguenther@suse.de>
29117 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29119 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29120 for inductions as relevant.
29122 2017-06-02 Richard Biener <rguenther@suse.de>
29124 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29125 (vectorizable_induction): ... this. Remove dead code.
29127 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29129 * builtins. (expand_builtin_alloca): Remove second parameter and
29130 infer its value from the first parameter instead.
29131 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29133 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29135 PR rtl-optimization/80903
29136 * loop-doloop.c (add_test): Unshare sequence.
29138 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29140 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29142 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29144 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29146 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29147 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29148 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29149 (xlogue_layout::get_stub_rtx): Make static.
29150 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29151 (xlogue_layout::compute_stub_managed_regs): Rename to...
29152 (xlogue_layout::count_stub_managed_regs): ...this.
29153 (xlogue_layout::is_stub_managed_reg): New function.
29154 (xlogue_layout::m_stub_names): Rename to...
29155 (xlogue_layout::s_stub_names): ...this, make static.
29156 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29157 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29158 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29159 xlogue_layout::s_stub_names): Instantiate statics.
29160 (stub_managed_regs): Remove.
29161 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29162 (disable_call_ms2sysv_xlogues): Rename to...
29163 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29164 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29166 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29167 change after reload_completed.
29168 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29170 (ix86_expand_prologue): Likewise.
29171 (ix86_expand_epilogue): Likewise.
29172 (ix86_expand_split_stack_prologue): Likewise.
29173 (ix86_compute_frame_layout): Remove frame parameter ...
29174 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29175 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29177 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29178 (ix86_frame): Move from here ...
29179 * config/i386/i386.h (ix86_frame): ... to here.
29180 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29181 complete ix86_frame data structure instead. Remove some_ld_name.
29183 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29185 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29186 symbols that hold a DECL_VALUE_EXPR.
29188 2017-06-01 Martin Jambor <mjambor@suse.cz>
29190 PR tree-optimization/80898
29191 * tree-sra.c (process_subtree_disqualification): Removed.
29192 (disqualify_candidate): Do not acll
29193 process_subtree_disqualification.
29194 (subtree_mark_written_and_enqueue): New function.
29195 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29196 RHS has been disqualified and re-queue LHS if necessary. Apart
29197 from that, ignore disqualified RHS.
29199 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29201 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29202 address fetch for z10 or later.
29204 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29206 * config/arc/arc.md (tst_movb): Add guard when splitting.
29208 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29210 * config/arc/arc.c (arc_can_eliminate): Test against
29211 arc_frame_pointer_needed.
29213 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29215 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29216 to prevent store reordering.
29217 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29218 (type): Add block type.
29219 (stack_tie): Define special instruction to be used in
29222 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29224 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29225 constraint. It is not valid for the pattern.
29226 (noncommutative_binary_comparison): Likewise.
29228 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29230 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29233 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29235 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29236 be used by the reg-alloc.
29238 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29240 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29241 reg-alloc when having mul64 or mul32x16 instructions.
29242 (mulsidi3): Likewise.
29243 (umulsidi3): Likewise.
29244 (mulsi32x16): New pattern.
29245 (mulsi64): Likewise.
29246 (mulsidi64): Likewise.
29247 (umulsidi64): Likewise.
29248 (MUL32x16_REG): Define.
29249 (mul64_600): Use MUL32x16_REG.
29250 (mac64_600): Likewise.
29251 (umul64_600): Likewise.
29252 (umac64_600): Likewise.
29254 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29256 * config/arc/arc.md (mulsi3_700): Make it commutative.
29258 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29260 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29262 (*sign_extendsidi2_insn): Likewise for movstosw.
29264 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29266 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29267 the type of the input discriminant value. Convert the
29268 discriminant value of signedness vary.
29270 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29272 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29273 Add to -Wall section.
29275 2017-06-01 Richard Biener <rguenther@suse.de>
29277 PR middle-end/66313
29278 * fold-const.c (fold_plusminus_mult_expr): If the factored
29279 factor may be zero use a wrapping type for the inner operation.
29280 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29281 and handle moved defs.
29282 (process_assignment): Properly guard the unary op case. Return a
29283 tri-state indicating that moving the stmt before the call may allow
29284 to continue. Pass through to_move.
29285 (find_tail_calls): Handle moving unrelated defs before
29288 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29291 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29292 splitter result in the canonical way.
29294 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29296 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29297 also for 32bit target. Update insn attributes.
29298 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29300 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29302 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29303 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29304 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29305 (_mm_maskz_min_ss): New intrinsics.
29307 2017-05-31 Martin Liska <mliska@suse.cz>
29309 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29310 Change comment style to one we normally use.
29311 (vectorizable_reduction): Likewise.
29312 (vectorizable_induction): Likewise.
29313 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29314 (vectorizable_call): Likewise.
29315 (vectorizable_simd_clone_call): Likewise.
29316 (vectorizable_conversion): Likewise.
29317 (vectorizable_assignment): Likewise.
29318 (vectorizable_shift): Likewise.
29319 (vectorizable_operation): Likewise.
29320 (vectorizable_store): Likewise.
29321 (vectorizable_load): Likewise.
29322 * tree-vectorizer.h: Likewise.
29324 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29326 * passes.c (emergency_dump_function): New.
29327 * tree-pass.h (emergency_dump_function): Declare.
29328 * plugin.c (plugins_internal_error_function): Remove.
29329 * plugin.h (plugins_internal_error_function): Remove declaration.
29330 * toplev.c (internal_error_function): New static function. Use it...
29331 (general_init): ...here.
29333 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29335 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29336 (arc_rtx_costs): Add costs for new patterns.
29337 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29338 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29340 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29342 * tree-ssa-strlen.c (get_next_strinfo): New function.
29343 (get_stridx_plus_constant): Use it.
29344 (zero_length_string): Likewise.
29345 (adjust_related_strinfos): Likewise.
29346 (adjust_last_stmt): Likewise.
29348 2017-05-31 Richard Biener <rguenther@suse.de>
29351 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29352 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29354 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29356 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29357 loop_vinfo argument and use of dependence distance vectors.
29358 Check instead whether the two references differ only in their
29359 initial value and assume that they have the same alignment if the
29360 difference is a multiple of the vector alignment.
29361 (vect_analyze_data_refs_alignment): Update call accordingly.
29363 2017-05-31 Martin Liska <mliska@suse.cz>
29366 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29368 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29370 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29371 (create_intersect_range_checks): Move from ...
29372 * tree-data-ref.c (create_intersect_range_checks_index)
29373 (create_intersect_range_checks): ... to here.
29374 (create_runtime_alias_checks): New function factored from ...
29375 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29376 here. Call above function.
29377 * tree-data-ref.h (create_runtime_alias_checks): New function.
29379 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29381 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29382 segment length for dr_b and compute it in wide_int.
29384 2017-05-31 Richard Biener <rguenther@suse.de>
29386 PR tree-optimization/80906
29387 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29388 and pass through iv_map.
29389 (copy_bb_and_scalar_dependences): Adjust.
29390 (translate_pending_phi_nodes): Likewise.
29391 (copy_loop_close_phi_args): Handle code-generating IVs instead
29394 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29396 * diagnostic-color.c (color_dict): Add "type-diff".
29397 (parse_gcc_colors): Update comment.
29398 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29399 -fdiagnostics-show-template-tree and -fno-elide-type.
29400 (GCC_COLORS): Add type-diff to example.
29402 (-fdiagnostics-show-template-tree): New.
29403 (-fno-elide-type): New.
29404 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29405 the pp_format_decoder callback. Call any m_format_postprocessor's
29407 (pretty_printer::pretty_printer): Initialize
29408 m_format_postprocessor.
29409 (pretty_printer::~pretty_printer): Delete any
29410 m_format_postprocessor.
29411 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29412 (class format_postprocessor): New class.
29413 (struct pretty_printer::format_decoder): Document the new parameters.
29414 (struct pretty_printer::m_format_postprocessor): New field.
29415 * tree-diagnostic.c (default_tree_printer): Update for new
29416 bool and const char ** params.
29417 * tree-diagnostic.h (default_tree_printer): Likewise.
29419 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29421 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29422 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29423 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29424 rs6000_gen_cell_microcode code.
29425 (rs6000_final_prescan_insn): Delete.
29426 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29427 "warn-cell-microcode" entries.
29428 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29429 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29430 throughout. Change cc_reg_not_micro_cr0_operand to
29431 cc_reg_not_cr0_operand throughout.
29432 (*extendhi<mode>2_noload): Delete.
29433 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29434 (mwarn-cell-microcode): Delete.
29435 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29436 -mgen-cell-microcode and -mwarn-cell-microcode.
29438 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29441 * config/i386/constraints.md (Yd): New constraint.
29443 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29444 and (?Yd, r) alternatives. Update insn attributes.
29445 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29446 and (?*Yd, r) alternatives. Update insn attributes.
29447 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29449 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29451 * gimplify.c (gimplify_modify_expr): Don't create a
29452 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29455 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29457 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29459 2017-05-30 Richard Biener <rguenther@suse.de>
29461 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29462 and reduc_def fields.
29463 (STMT_VINFO_REDUC_TYPE): New define.
29464 (STMT_VINFO_REDUC_DEF): Likewise.
29465 (vect_force_simple_reduction): Adjust prototype.
29466 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29467 (vect_is_simple_reduction): Remove check_reduction argument.
29468 (vect_force_simple_reduction): Adjust and set
29469 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29470 (vectorizable_reduction): Do not re-do reduction analysis
29471 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29472 * tree-parloops.c (gather_scalar_reductions): Adjust.
29474 2017-05-30 Richard Biener <rguenther@suse.de>
29476 PR middle-end/80901
29477 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29480 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29482 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29483 Introduce unknown_misalignment parameter and remove vf.
29484 (vect_peeling_hash_get_lowest_cost):
29485 Pass unknown_misalignment parameter.
29486 (vect_enhance_data_refs_alignment):
29487 Fix unsupportable data ref treatment.
29489 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29491 * tree-vect-data-refs.c (vect_get_data_access_cost):
29492 Workaround for SLP handling.
29493 (vect_enhance_data_refs_alignment):
29494 Compute costs for doing no peeling at all, compare to the best
29495 peeling costs so far and avoid peeling if cheaper.
29497 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29499 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29500 Return peeling info and set costs to zero for unlimited cost
29502 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29503 with unknown misalignment. Compute and costs for unknown
29504 misalignment, compare them to the costs for known misalignment
29505 and choose the cheapest for peeling.
29507 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29509 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29510 (vect_get_peeling_costs_all_drs): Create function.
29511 (vect_peeling_hash_get_lowest_cost):
29512 Use vect_get_peeling_costs_all_drs.
29513 (vect_peeling_supportable): Create function.
29514 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29516 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29518 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29519 DR_HAS_NEGATIVE_STEP.
29520 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29521 (vect_enhance_data_refs_alignment): Use.
29522 (vect_duplicate_ssa_name_ptr_info): Use.
29523 * tree-vectorizer.h (dr_misalignment): Use.
29524 (known_alignment_for_access_p): Use.
29526 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29529 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29531 (has_section_name): New function.
29533 2017-05-30 Martin Liska <mliska@suse.cz>
29536 * auto-profile.c (get_function_decl_from_block): Fix
29539 2017-05-30 Richard Biener <rguenther@suse.de>
29541 PR middle-end/80876
29542 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29544 2017-05-30 Martin Liska <mliska@suse.cz>
29546 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29547 * dumpfile.h (struct dump_file_info): Remove ctors.
29549 2017-05-30 Martin Liska <mliska@suse.cz>
29551 * predict.def: Fix GNU coding style.
29553 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29555 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29556 Mark 'to' argument with ATTRIBUTE_UNUSED.
29558 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29560 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29561 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29562 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29565 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29567 * doc/install.texi (Options specification): Restore entry of
29568 --enable-sjlj-exceptions.
29570 2017-05-27 Michael Eager <eager@eagercon.com>
29573 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29575 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29577 * config/microblaze/microblaze.h
29578 (FIXED_REGISTERS): Update in macro.
29579 (CALL_USED_REGISTERS): Update in macro.
29581 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29583 * doc/install.texi: Add links to macOS binary distributions.
29585 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29589 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29591 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29593 2017-05-26 Martin Liska <mliska@suse.cz>
29595 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29597 2017-05-26 Martin Liska <mliska@suse.cz>
29599 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29600 always leading ';; '.
29601 (dump_bb_info): Likewise.
29602 (brief_dump_cfg): Likewise.
29603 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29604 * dumpfile.c: Remove usage of TDF_VERBOSE.
29605 * dumpfile.h (enum dump_kind): Likewise.
29606 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29607 * print-tree.c (debug_verbose): Remove.
29608 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29609 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29610 * tree-diagnostic.c (default_tree_printer): Replace
29611 TDF_DIAGNOSTIC with TDF_SLIM.
29613 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29615 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29616 in parameter loop, rather than loop_vinfo.
29617 (create_intersect_range_checks): Ditto.
29618 (vect_create_cond_for_alias_checks): Update call to above functions.
29620 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29622 PR tree-optimization/80815
29623 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29624 for merging runtime alias checks. Handle negative DR_STEPs.
29626 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29628 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29630 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29631 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29632 out code pruning runtime alias checks.
29633 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29634 factored out from above.
29635 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29637 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29639 (prune_runtime_alias_test_list): New decalaration.
29641 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29643 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29644 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29645 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29646 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29647 (operator==, comp_dr_with_seg_len_pair): Ditto.
29648 (vect_prune_runtime_alias_test_list): Ditto.
29650 2017-05-26 Martin Liska <mliska@suse.cz>
29653 * params.def: Bound partial-inlining-entry-probability param.
29655 2017-05-26 Marek Polacek <polacek@redhat.com>
29658 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29661 2017-05-26 Richard Biener <rguenther@suse.de>
29663 PR tree-optimization/80842
29664 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29667 2017-05-26 Richard Biener <rguenther@suse.de>
29669 PR tree-optimization/80844
29670 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29672 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29674 * doc/md.texi (Machine Constraints): Update x86 family
29675 machine constraints section to match 'config/i386/constraints.md'.
29677 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29679 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29681 2017-05-25 Nathan Sidwell <nathan@acm.org>
29683 * doc/invoke.texi (--enable-languages): Update documentation.
29685 2017-05-25 Martin Liska <mliska@suse.cz>
29687 * dumpfile.c: Add TDF_FOLDING.
29688 * dumpfile.h (enum dump_kind): Likewise.
29689 * genmatch.c (dt_simplify::gen_1): Use it.
29691 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29693 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29695 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29697 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29698 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29700 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29702 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29703 * match.pd (X == C): Rewrite it here.
29704 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29705 with_certain_nonzero_bits2): New predicates.
29706 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29708 2017-05-24 Nathan Sidwell <nathan@acm.org>
29710 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29713 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29716 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29718 * config/powerpcspe: New port. Files are copied from the rs6000
29719 port, with "rs6000" in filenames replaced by "powerpcspe".
29721 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29723 PR rtl-optimization/80754
29724 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29726 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29728 * config/sparc/sparc.md (length): Return the correct value for -mflat
29729 sibcalls to match output_sibcall.
29731 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29735 * config/rs6000/rs6000.c (struct machine_function): Add new field
29737 (rs6000_get_separate_components): Init that field, use it.
29738 (rs6000_components_for_bb): Use the field.
29740 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29742 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29744 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29746 PR middle-end/80823
29747 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29749 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29752 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29753 address against address_operand predicate.
29754 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29756 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29758 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29759 parameters passed indirectly.
29761 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29763 * config/i386/i386.md (*movdi_internal): Remove SSE4
29764 alternative 18 (?r, *v). Update insn attributes.
29765 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29766 Update insn attributes.
29767 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29768 Update insn attributes.
29769 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29770 alternative 1 (r, v). Remove isa attribute.
29771 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29772 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29773 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29775 2017-05-23 Tom de Vries <tom@codesourcery.com>
29777 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29780 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29782 * cgraphunit.c (symbol_table::process_new_functions): Update.
29783 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29784 (inline_generate_summary): Rename to ...
29785 (ipa_fn_summary_generate): ... this one.
29786 (inline_read_summary): Rename to ...
29787 (ipa_fn_summary_read): ... this one.
29788 (inline_write_summary): Rename to ...
29789 (ipa_fn_summary_write): ... this one.
29790 (inline_free_summary): Rename to ...
29791 (ipa_free_fn_summary): ... this one.
29792 (pass_data_local_fn_summary, pass_local_fn_summary,
29793 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29794 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29795 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29796 make_pass_ipa_fn_summary): New.
29797 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29798 inline_write_summary, inline_free_summary): Remove.
29799 (ipa_free_fn_summary) : New.
29800 * ipa-inline.c (ipa_inline): Update.
29801 (pass_ipa_inline): Do not generate summaries.
29802 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29804 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29805 and add pass_ipa_fn_summary.
29806 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29808 (make_pass_inline_parameters): Remove.
29810 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29812 * omp-low.c (struct omp_context): Remove "default_kind" member.
29815 * omp-offload.c (execute_oacc_device_lower): Remove the
29816 parallelism dimensions function attributes for unparallelized
29817 OpenACC kernels constructs.
29819 2017-05-23 Martin Liska <mliska@suse.cz>
29821 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29823 (cgraph_edge::make_speculative): Likewise.
29824 (cgraph_edge::resolve_speculation): Likewise.
29825 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29826 (cgraph_node::dump): Likewise.
29827 * cgraph.h: Likewise.
29828 * cgraphunit.c (analyze_functions): Likewise.
29829 (symbol_table::compile): Likewise.
29830 * ipa-cp.c (print_all_lattices): Likewise.
29831 (determine_versionability): Likewise.
29832 (initialize_node_lattices): Likewise.
29833 (ipcp_verify_propagated_values): Likewise.
29834 (estimate_local_effects): Likewise.
29835 (update_profiling_info): Likewise.
29836 (create_specialized_node): Likewise.
29837 (perhaps_add_new_callers): Likewise.
29838 (decide_about_value): Likewise.
29839 (decide_whether_version_node): Likewise.
29840 (identify_dead_nodes): Likewise.
29841 (ipcp_store_bits_results): Likewise.
29842 * ipa-devirt.c (dump_targets): Likewise.
29843 (ipa_devirt): Likewise.
29844 * ipa-icf.c (sem_item::dump): Likewise.
29845 (sem_function::equals): Likewise.
29846 (sem_variable::equals): Likewise.
29847 (sem_item_optimizer::read_section): Likewise.
29848 (sem_item_optimizer::execute): Likewise.
29849 (congruence_class::dump): Likewise.
29850 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29851 (dump_inline_summary): Likewise.
29852 (estimate_node_size_and_time): Likewise.
29853 (inline_analyze_function): Likewise.
29854 * ipa-inline-transform.c (inline_call): Likewise.
29855 * ipa-inline.c (report_inline_failed_reason): Likewise.
29856 (want_early_inline_function_p): Likewise.
29857 (edge_badness): Likewise.
29858 (update_edge_key): Likewise.
29859 (inline_small_functions): Likewise.
29860 * ipa-profile.c (ipa_profile): Likewise.
29861 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29862 (ipa_make_edge_direct_to_target): Likewise.
29863 (remove_described_reference): Likewise.
29864 (ipa_impossible_devirt_target): Likewise.
29865 (propagate_controlled_uses): Likewise.
29866 (ipa_print_node_params): Likewise.
29867 (ipcp_transform_function): Likewise.
29868 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29869 (propagate_pure_const): Likewise.
29870 * ipa-reference.c (generate_summary): Likewise.
29871 (read_write_all_from_decl): Likewise.
29872 (propagate): Likewise.
29873 (ipa_reference_read_optimization_summary): Likewise.
29874 * ipa-utils.c (ipa_merge_profiles): Likewise.
29875 * ipa.c (walk_polymorphic_call_targets): Likewise.
29876 (symbol_table::remove_unreachable_nodes): Likewise.
29877 (ipa_single_use): Likewise.
29878 * passes.c (execute_todo): Likewise.
29879 * predict.c (drop_profile): Likewise.
29880 * symtab.c (symtab_node::get_dump_name): New function.
29881 (symtab_node::dump_name): Likewise.
29882 (symtab_node::dump_asm_name): Likewise.
29883 (symtab_node::dump_references): Likewise.
29884 (symtab_node::dump_referring): Likewise.
29885 (symtab_node::dump_base): Likewise.
29886 (symtab_node::debug_symtab): Likewise.
29887 * tree-sra.c (convert_callers_for_node): Likewise.
29888 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29889 * value-prof.c (init_node_map): Likewise.
29891 2017-05-23 Martin Liska <mliska@suse.cz>
29893 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29894 and symtab_node::debug_symtab to symbol_table::debug.
29895 * cgraphunit.c (analyze_functions): Use the renamed function.
29896 (symbol_table::compile): Likewise.
29897 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29898 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29899 * passes.c (execute_todo): Likewise.
29900 * symtab.c (symbol_table::dump): New function.
29901 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29903 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29905 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29906 that nonconst implies exec.
29908 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29910 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29911 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29912 (inline_edge_summary_vec): Turn into ...
29913 (ipa_call_summaries): ... this one.
29914 (redirect_to_unreachable, edge_set_predicate,
29915 evaluate_properties_for_edge, inline_summary_alloc,
29916 reset_ipa_call_summary, reset_inline_summary,
29917 inline_summary_t::duplicate): Update.
29918 (inline_edge_duplication_hook): Turn to ...
29919 (ipa_call_summary_t::duplicate): ... this one.
29920 (inline_edge_removal_hook): Turn to ...
29921 (ipa_call_summary_t::remove): ... this one.
29922 (dump_inline_edge_summary): Turn to ...
29923 (dump_ipa_call_summary): ... this one.
29924 (estimate_function_body_sizes): Update.
29925 (inline_update_callee_summaries): Update.
29926 (remap_edge_change_prob): Update.
29927 (remap_edge_summaries): Update.
29928 (inline_merge_summary): Update.
29929 (do_estimate_edge_time): Update.
29930 (inline_generate_summary): Update.
29931 (inline_read_section): Update.
29932 (inline_read_summary): Update.
29933 (inline_free_summary): Update.
29934 * ipa-inline.c (can_inline_edge_p): Update.
29935 (compute_inlined_call_time): Update.
29936 (want_inline_small_function_p): Update.
29937 (edge_badness): Update.
29938 (early_inliner): Update.
29939 * ipa-inline.h (inline_edge_summary): Turn to ...
29940 (ipa_call_summary): ... this one.
29941 (ipa_call_summary_t): New class.
29942 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29943 (ipa_call_summaries): New.
29944 (inline_edge_summary): Remove.
29945 (estimate_edge_growth): Update.
29946 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29947 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29948 * ipa-split.c (execute_split_functions): Update.
29949 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29951 2017-05-23 Tom de Vries <tom@codesourcery.com>
29953 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29954 attributes): Document rdrand effective target.
29956 2017-05-23 Tom de Vries <tom@codesourcery.com>
29958 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29959 attributes): Sort alphabetically.
29961 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29963 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29965 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29968 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29969 V2DF/V2DI splat into two separate patterns, one that handles
29970 registers, and the other that only handles memory. Drop support
29971 for splatting from a GPR on ISA 2.07 and then splitting the
29972 splat into direct move and splat.
29973 (vsx_splat_<mode>_reg): Likewise.
29974 (vsx_splat_<mode>_mem): Likewise.
29976 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29978 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29980 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29982 PR middle-end/80809
29983 * omp-low.c (finish_taskreg_remap): New function.
29984 (finish_taskreg_scan): If unit size of ctx->record_type
29985 is non-constant, unshare the size expression and replace
29986 decls in it with possible outer var refs.
29988 PR middle-end/80809
29989 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29990 GOVD_SHARED rather than GOVD_PRIVATE with it.
29991 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29992 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29994 PR middle-end/80853
29995 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29996 as last argument to build_outer_var_ref for pointer bases of array
29997 section reductions.
29999 2017-05-19 Martin Sebor <msebor@redhat.com>
30001 * print-tree.c (print_node): Print DECL_READ_P flag.
30003 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30005 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30006 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30007 * cgraph.c: Likewise.
30008 * cgraphunit.c: Likewise.
30009 * gengtype.c: Likewise.
30010 * ipa-cp.c: Likewise.
30011 * ipa-devirt.c: Likewise.
30012 * ipa-icf.c: Likewise.
30013 * ipa-predicate.c: Likewise.
30014 * ipa-profile.c: Likewise.
30015 * ipa-prop.c: Likewise.
30016 * ipa-split.c: Likewise.
30018 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30019 edge_predicate_pool, dump_inline_hints,
30020 inline_summary::account_size_time, redirect_to_unreachable,
30021 edge_set_predicate, set_hint_predicate,
30022 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30023 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30024 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30025 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30026 ipa_call_summary_t::remove, initialize_growth_caches,
30027 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30028 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30029 mark_modified, unmodified_parm_1, unmodified_parm,
30030 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30031 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30032 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30033 will_be_nonconstant_predicate, record_modified_bb_info,
30034 get_minimal_bb, record_modified, param_change_prob,
30035 phi_result_unknown_predicate, predicate_for_phi_result,
30036 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30037 estimate_function_body_sizes, compute_inline_parameters,
30038 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30039 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30040 inline_update_callee_summaries, remap_edge_change_prob,
30041 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30042 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30043 inline_analyze_function, inline_summary_t::insert,
30044 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30045 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30046 inline_free_summary): Move to ipa-fnsummary.h
30047 (predicate_t): Remove.
30048 * ipa-fnsummary.c: New file.
30049 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30050 (enum inline_hints_vals, inline_hints, agg_position_info,
30051 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30052 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30053 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30054 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30055 inline_read_summary, inline_write_summary, inline_free_summary,
30056 inline_analyze_function, initialize_inline_failed,
30057 inline_merge_summary, inline_update_overall_summary,
30058 compute_inline_parameters): Move to ipa-fnsummary.h
30059 * ipa-fnsummary.h: New file.
30060 * ipa-inline-transform.h: Include ipa-inline.h.
30061 * ipa-inline.c: LIkewise.
30063 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30065 * ipa-inline.c (edge_badness): Use inlined_time instead of
30066 inline_summaries->get.
30068 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30070 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30072 2017-05-22 Nathan Sidwell <nathan@acm.org>
30074 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30075 (fdump-lang): Document 'raw' option.
30076 * dumpfile.h (TDI_tu): Delete.
30077 * dumpfile.c (dump_files): Remove translation-unit.
30078 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30080 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30082 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30083 command option from $(AWK) call.
30084 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30085 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30086 [FORMAT]: Remove handling of variable.
30087 * config/avr/t-multilib: Regenerate.
30089 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30091 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30093 (dump_inline_summary): Do not print self_time.
30094 (estimate_function_body_sizes): Do not set self_time.
30095 (compute_inline_parameters): Likewise.
30096 (inline_read_section, inline_write_summary): Do not stream self_time.
30097 * ipa-inline.h (inline_summary): Drop self_time.
30099 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30101 * ipa-inline-analysis.c (account_size_time): Rename to ...
30102 (inline_summary::account_size_time): ... this one.
30103 (reset_ipa_call_summary): Turn to ...
30104 (ipa_call_summary::reset): ... this one.
30105 (reset_inline_summary): Turn to ...
30106 (inline_summary::reset): ... this one.
30107 (inline_summary_t::remove): Update.
30108 (inline_summary_t::duplicate): Update.
30109 (ipa_call_summary_t::remove): Update.
30110 (dump_inline_summary): Update.
30111 (estimate_function_body_sizes): Update.
30112 (compute_inline_parameters): Update.
30113 (estimate_node_size_and_time): Update.
30114 (inline_merge_summary): Update.
30115 (inline_update_overall_summary): Update.
30116 (inline_read_section): Update.
30117 (inline_write_summary): Update.
30118 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30119 add account_size_time and reset member functions.
30120 (ipa_call_summary): Add reset function.
30121 * ipa-predicate.h (predicate::operator &): Constify.
30123 2017-05-22 Richard Biener <rguenther@suse.de>
30125 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30127 2017-05-19 Jason Merrill <jason@redhat.com>
30129 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30131 2017-05-19 Marek Polacek <polacek@redhat.com>
30134 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30135 TYPE_OVERFLOW_WRAPS checks.
30137 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30139 * tree-core.h (enum omp_clause_default_kind): Add
30140 "OMP_CLAUSE_DEFAULT_PRESENT".
30141 * tree-pretty-print.c (dump_omp_clause): Handle it.
30142 * gimplify.c (enum gimplify_omp_var_data): Add
30143 "GOVD_MAP_FORCE_PRESENT".
30144 (gimplify_adjust_omp_clauses_1): Map it to
30145 "GOMP_MAP_FORCE_PRESENT".
30146 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30148 * gimplify.c (oacc_default_clause): Clarify.
30150 2017-05-19 Nathan Sidwell <nathan@acm.org>
30152 LANG_HOOK_REGISTER_DUMPS
30153 * toplev.c (general_init): Call register dump lang hook.
30154 * doc/invoke.texi: Document -fdump-lang option family.
30155 * dumpfile.c (dump_files): Remove class dump here.
30156 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30157 * dumpfile.h (tree_dump_index): Remove TDI_class.
30158 * langhooks-def.h (lhd_register_dumps): Declare.
30159 (LANG_HOOKS_REGISTER_DUMPS): Define.
30160 (LANG_HOOKS_INITIALIZER): Add it.
30161 * langhooks.c (lhd_register_dumps): Define.
30162 * langhooks.h (struct lang_hooks): Add register_dumps.
30164 2017-05-19 Nathan Sidwell <nathan@acm.org>
30166 * context.h (context::set_passes): New.
30167 * context.c (context::context): Do not create pass manager.
30168 * toplev.c (general_init): Create pass manager here.
30170 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30172 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30173 use this splitter if two add or or instructions would also work for
30174 the constant we want to generate.
30176 2017-05-19 Richard Biener <rguenther@suse.de>
30179 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30180 predicate evaluation.
30182 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30184 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30186 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30189 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30191 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30192 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30193 (inline_edge_summary_vec): Turn into ...
30194 (ipa_call_summaries): ... this one.
30195 (redirect_to_unreachable, edge_set_predicate,
30196 evaluate_properties_for_edge, inline_summary_alloc,
30197 reset_ipa_call_summary, reset_inline_summary,
30198 inline_summary_t::duplicate): Update.
30199 (inline_edge_duplication_hook): Turn to ...
30200 (ipa_call_summary_t::duplicate): ... this one.
30201 (inline_edge_removal_hook): Turn to ...
30202 (ipa_call_summary_t::remove): ... this one.
30203 (dump_inline_edge_summary): Turn to ...
30204 (dump_ipa_call_summary): ... this one.
30205 (estimate_function_body_sizes): Update.
30206 (inline_update_callee_summaries): Update.
30207 (remap_edge_change_prob): Update.
30208 (remap_edge_summaries): Update.
30209 (inline_merge_summary): Update.
30210 (do_estimate_edge_time): Update.
30211 (inline_generate_summary): Update.
30212 (inline_read_section): Update.
30213 (inline_read_summary): Update.
30214 (inline_free_summary): Update.
30215 * ipa-inline.c (can_inline_edge_p): Update.
30216 (compute_inlined_call_time): Update.
30217 (want_inline_small_function_p): Update.
30218 (edge_badness): Update.
30219 (early_inliner): Update.
30220 * ipa-inline.h (inline_edge_summary): Turn to ...
30221 (ipa_call_summary): ... this one.
30222 (ipa_call_summary_t): New class.
30223 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30224 (ipa_call_summaries): New.
30225 (inline_edge_summary): Remove.
30226 (estimate_edge_growth): Update.
30227 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30228 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30229 * ipa-split.c (execute_split_functions): Update.
30230 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30232 2017-05-19 Richard Biener <rguenther@suse.de>
30234 PR middle-end/80764
30235 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30237 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30239 * config/rs6000/rs6000.c (struct machine_function): Add field
30240 fpr_is_wrapped_separately.
30241 (rs6000_get_separate_components): Use 64 components. Handle the
30242 new FPR components.
30243 (rs6000_components_for_bb): Handle the FPR components.
30244 (rs6000_emit_prologue_components): Handle the FPR components.
30245 (rs6000_emit_epilogue_components): Handle the FPR components.
30246 (rs6000_set_handled_components): Handle the FPR components.
30247 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30248 that are already separately shrink-wrapped.
30249 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30250 that are already separately shrink-wrapped.
30252 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30255 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30258 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30259 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30260 cases where the register allocator uses a move and an offsettable
30261 memory operation to/from a FPR register on ISA 2.06/2.07.
30262 (define_peephole2 for Altivec d-form store): Likewise.
30264 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30267 * config/i386/mmx.md (*mov<mode>_internal): Enable
30268 alternatives 11, 12, 13 and 14 also for 32bit targets.
30269 Remove alternatives 15, 16, 17 and 18.
30270 * config/i386/sse.md (vec_concatv2di): Change
30271 alternative (!x, *y) to (x, ?!*Yn).
30273 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30275 * dumpfile.h (enum dump_kind): Remove stray comma.
30277 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30279 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30280 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30281 predicate::num_conditions
30282 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30283 (CHANGED): turn into predicate::changed.
30284 (agg_position_info): Move to ipa-predicate.h
30285 (add_condition, predicate::add_clause, predicate::operator &=,
30286 predicate::or_with, predicate::evaluate, predicate::probability,
30287 dump_condition, dump_clause, predicate::dump,
30288 predicate::remap_after_duplication, predicate::remap_after_inlining,
30289 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30290 (evaluate_conditions_for_known_args): Update.
30291 (set_cond_stmt_execution_predicate): Update.
30292 * ipa-inline.h: Include ipa-predicate.h
30293 (condition, inline_param_summary, conditions, agg_position_info,
30294 predicate): Move to ipa-predicate.h
30295 * ipa-predicate.c: New file.
30296 * ipa-predicate.h: New file.
30298 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30300 * final.c (leaf_function_p): Check we are not in a sequence.
30302 2017-05-18 Martin Liska <mliska@suse.cz>
30304 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30305 * dumpfile.c (dump_register): Use new enum dump_kind.
30306 (get_dump_file_name): Likewise.
30307 (dump_enable_all): Likewise.
30308 (dump_switch_p_1): Likewise.
30309 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30310 * dumpfile.h (enum dump_kind): New enum type.
30311 (struct dump_file_info): Create constructor and
30312 format fields and comments.
30313 * passes.c (pass_manager::register_one_dump_file):
30315 * statistics.c (statistics_early_init): Likewise.
30316 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30317 TDF_TREE with TDF_SLIM.
30318 (gather_memory_references_ref): Likewise.
30320 2017-05-18 Martin Liska <mliska@suse.cz>
30322 * vec.h (struct vnull): Use it.
30324 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30326 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30327 (true_predicate, false_predicate, true_predicate_p,
30328 false_predicate_p): Remove.
30329 (single_cond_predicate, not_inlined_predicate): Turn to member function
30331 (add_condition): Update.
30332 (add_clause): Turn to...
30333 (predicate::add_clause): ... this one; update; allow passing NULL
30335 (and_predicates): Turn to ...
30336 (predicate::operator &=): ... this one.
30337 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30338 (or_predicates): Turn to ...
30339 (predicate::or_with): ... this one.
30340 (evaluate_predicate): Turn to ...
30341 (predicate::evaluate): ... this one.
30342 (predicate_probability): Turn to ...
30343 (predicate::probability): ... this one.
30344 (dump_condition): Update.
30345 (dump_predicate): Turn to ...
30346 (predicate::dump): ... this one.
30347 (account_size_time): Update.
30348 (edge_set_predicate): Update.
30349 (set_hint_predicate): UPdate.
30350 (evaluate_conditions_for_known_args): Update.
30351 (evaluate_properties_for_edge): Update.
30352 (remap_predicate_after_duplication): Turn to...
30353 (predicate::remap_after_duplication): ... this one.
30354 (remap_hint_predicate_after_duplication): Update.
30355 (inline_summary_t::duplicate): UPdate.
30356 (dump_inline_edge_summary): Update.
30357 (dump_inline_summary): Update.
30358 (set_cond_stmt_execution_predicate): Update.
30359 (set_switch_stmt_execution_predicate): Update.
30360 (compute_bb_predicates): Update.
30361 (will_be_nonconstant_expr_predicate): Update.
30362 (will_be_nonconstant_predicate): Update.
30363 (phi_result_unknown_predicate): Update.
30364 (predicate_for_phi_result): Update.
30365 (array_index_predicate): Update.
30366 (estimate_function_body_sizes): Update.
30367 (estimate_node_size_and_time): Update.
30368 (estimate_ipcp_clone_size_and_time): Update.
30369 (remap_predicate): Rename to ...
30370 (predicate::remap_after_inlining): ... this one.
30371 (remap_hint_predicate): Update.
30372 (inline_merge_summary): Update.
30373 (inline_update_overall_summary): Update.
30374 (estimate_size_after_inlining): Update.
30375 (read_predicate): Rename to ...
30376 (predicate::stream_in): ... this one.
30377 (read_inline_edge_summary): Update.
30378 (write_predicate): Rename to ...
30379 (predicate::stream_out): ... this one.
30380 (write_inline_edge_summary): Update.
30381 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30382 (clause_t): Turn to uint32_t
30383 (predicate): Turn to class; implement constructor and operators
30385 (size_time_entry): Update.
30386 (inline_summary): Update.
30387 (inline_edge_summary): Update.
30389 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30391 * fold-const.c (fold_binary_loc): Move transformation...
30392 * match.pd (C - X CMP X): ... here.
30394 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30396 * config/sparc/sparc.c (sparc_option_override): Set function
30397 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30398 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30400 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30402 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30404 2017-05-18 Marek Polacek <polacek@redhat.com>
30407 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30408 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30409 gimple_assign_load_p.
30411 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30413 PR middle-end/80692
30414 * real.c (do_compare): Give decimal_do_compare preference over
30415 comparing just the signs.
30417 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30419 * doc/md.texi (Canonicalization of Instructions): Describe the
30420 canonical form of instructions that inherently set a condition
30423 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30425 PR middle-end/80775
30426 * tree-cfg.c: Move deletion of unreachable case statements to after
30427 the merging of consecutive case labels.
30429 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30431 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30432 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30433 restoring of callee-saved registers.
30435 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30437 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30438 * config/visium/visium.c (single_set_and_flags): Likewise.
30439 * config/visium/visium.md (Substitutions): Likewise.
30441 2017-05-17 Martin Liska <mliska@suse.cz>
30443 * cfg.c: Introduce dump_flags_t type and
30444 use it instead of int type.
30446 * cfghooks.c: Likewise.
30447 * cfghooks.h (struct cfg_hooks): Likewise.
30448 * cfgrtl.c: Likewise.
30449 * cfgrtl.h: Likewise.
30450 * cgraph.c (cgraph_node::get_body): Likewise.
30451 * coretypes.h: Likewise.
30452 * domwalk.c: Likewise.
30453 * domwalk.h: Likewise.
30454 * dumpfile.c (struct dump_option_value_info): Likewise.
30455 (dump_enable_all): Likewise.
30456 (dump_switch_p_1): Likewise.
30457 (opt_info_switch_p): Likewise.
30458 * dumpfile.h (enum tree_dump_index): Likewise.
30459 (struct dump_file_info): Likewise.
30460 * genemit.c: Likewise.
30461 * generic-match-head.c: Likewise.
30462 * gengtype.c (open_base_files): Likewise.
30463 * gimple-pretty-print.c: Likewise.
30464 * gimple-pretty-print.h: Likewise.
30465 * graph.c (print_graph_cfg): Likewise.
30466 * graphite-scop-detection.c (dot_all_sese): Likewise.
30467 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30468 * loop-unroll.c (report_unroll): Likewise.
30469 * passes.c (pass_manager::register_one_dump_file): Likewise.
30470 * print-tree.c: Likewise.
30471 * statistics.c: Likewise.
30472 * tree-cfg.c: Likewise.
30473 * tree-cfg.h: Likewise.
30474 * tree-dfa.c: Likewise.
30475 * tree-dfa.h: Likewise.
30476 * tree-dump.c (dump_function): Likewise.
30477 * tree-dump.h (struct dump_info): Likewise.
30478 * tree-pretty-print.c: Likewise.
30479 * tree-pretty-print.h: Likewise.
30480 * tree-ssa-live.c: Likewise.
30481 * tree-ssa-live.h: Likewise.
30482 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30483 * tree-vect-loop.c: Likewise.
30484 * tree-vect-slp.c: Likewise.
30486 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30487 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30489 PR tree-optimization/80457
30490 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30491 of all arguments to a statement as scalar_to_vec operations.
30492 (vectorizable_call): Adjust call to vect_model_simple_cost for
30494 (vectorizable_conversion): Likewise.
30495 (vectorizable_assignment): Likewise.
30496 (vectorizable_shift): Likewise.
30497 (vectorizable_operation): Likewise.
30498 (vectorizable_comparison): Likewise.
30499 (vect_is_simple_cond): Record the def types for operands.
30500 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30501 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30502 for statement argument count.
30504 2017-05-16 Carl Love <cel@us.ibm.com>
30506 * config/rs6000/rs6000-c: Add support for built-in functions
30507 vector unsigned long long vec_bperm (vector unsigned long long,
30508 vector unsigned char)
30509 vector signed long long vec_mule (vector signed int,
30511 vector unsigned long long vec_mule (vector unsigned int,
30512 vector unsigned int)
30513 vector signed long long vec_mulo (vector signed int,
30515 vector unsigned long long vec_mulo (vector unsigned int,
30516 vector unsigned int)
30517 vector signed char vec_sldw (vector signed char,
30518 vector signed char,
30520 vector unsigned char vec_sldw (vector unsigned char,
30521 vector unsigned char,
30523 vector signed short vec_sldw (vector signed short,
30524 vector signed short,
30526 vector unsigned short vec_sldw (vector unsigned short,
30527 vector unsigned short,
30529 vector signed int vec_sldw (vector signed int,
30532 vector unsigned int vec_sldw (vector unsigned int,
30533 vector unsigned int,
30535 vector signed long long vec_sldw (vector signed long long,
30536 vector signed long long,
30538 vector unsigned long long vec_sldw (vector unsigned long long,
30539 vector unsigned long long,
30541 * config/rs6000/rs6000-c: Add support for built-in functions
30542 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30543 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30544 * doc/extend.texi: Update the built-in documentation for the
30545 new built-in functions.
30547 2017-05-16 Marek Polacek <polacek@redhat.com>
30551 * tree.c (save_expr): Don't fold the expression.
30553 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30555 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30556 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30557 and (?*y,m). Update insn attributes.
30559 2017-05-16 Martin Liska <mliska@suse.cz>
30561 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30562 flags argument of print_gimple_stmt, print_gimple_expr,
30563 print_generic_stmt and print_generic_expr.
30564 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30565 * coretypes.h: Likewise.
30566 * except.c (dump_eh_tree): Likewise.
30567 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30568 * gimple-pretty-print.h: Likewise.
30569 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30570 (backprop::push_to_worklist): Likewise.
30571 (backprop::pop_from_worklist): Likewise.
30572 (backprop::process_use): Likewise.
30573 (backprop::intersect_uses): Likewise.
30574 (note_replacement): Likewise.
30575 * gimple-ssa-store-merging.c
30576 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30577 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30578 (pass_store_merging::execute): Likewise.
30579 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30580 (ssa_base_cand_dump_callback): Likewise.
30581 (dump_incr_vec): Likewise.
30582 (replace_refs): Likewise.
30583 (replace_mult_candidate): Likewise.
30584 (create_add_on_incoming_edge): Likewise.
30585 (create_phi_basis): Likewise.
30586 (insert_initializers): Likewise.
30587 (all_phi_incrs_profitable): Likewise.
30588 (introduce_cast_before_cand): Likewise.
30589 (replace_one_candidate): Likewise.
30590 * gimplify.c (gimplify_expr): Likewise.
30591 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30592 (set_rename): Likewise.
30593 (rename_uses): Likewise.
30594 (copy_loop_phi_nodes): Likewise.
30595 (add_close_phis_to_merge_points): Likewise.
30596 (copy_loop_close_phi_args): Likewise.
30597 (copy_cond_phi_args): Likewise.
30598 (graphite_copy_stmts_from_block): Likewise.
30599 (translate_pending_phi_nodes): Likewise.
30600 * graphite-poly.c (print_pdr): Likewise.
30601 (dump_gbb_cases): Likewise.
30602 (dump_gbb_conditions): Likewise.
30603 (print_scop_params): Likewise.
30604 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30605 (build_cross_bb_scalars_use): Likewise.
30606 (gather_bbs::before_dom_children): Likewise.
30607 * hsa-dump.c (dump_hsa_immed): Likewise.
30608 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30609 (get_replacement_map): Likewise.
30610 * ipa-inline-analysis.c (dump_condition): Likewise.
30611 (estimate_function_body_sizes): Likewise.
30612 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30613 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30614 * ipa-prop.c (ipa_dump_param): Likewise.
30615 (ipa_print_node_jump_functions_for_edge): Likewise.
30616 (ipa_modify_call_arguments): Likewise.
30617 (ipa_modify_expr): Likewise.
30618 (ipa_dump_param_adjustments): Likewise.
30619 (ipa_dump_agg_replacement_values): Likewise.
30620 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30621 * ipa-pure-const.c (check_stmt): Likewise.
30622 (pass_nothrow::execute): Likewise.
30623 * ipa-split.c (execute_split_functions): Likewise.
30624 * omp-offload.c (dump_oacc_loop_part): Likewise.
30625 (dump_oacc_loop): Likewise.
30626 * trans-mem.c (tm_log_emit): Likewise.
30627 (tm_memopt_accumulate_memops): Likewise.
30628 (dump_tm_memopt_set): Likewise.
30629 (dump_tm_memopt_transform): Likewise.
30630 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30631 (print_loop): Likewise.
30632 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30633 (chkp_gather_checks_info): Likewise.
30634 (chkp_get_check_result): Likewise.
30635 (chkp_remove_check_if_pass): Likewise.
30636 (chkp_use_outer_bounds_if_possible): Likewise.
30637 (chkp_reduce_bounds_lifetime): Likewise.
30638 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30639 (chkp_mark_completed_bounds): Likewise.
30640 (chkp_register_incomplete_bounds): Likewise.
30641 (chkp_mark_invalid_bounds): Likewise.
30642 (chkp_maybe_copy_and_register_bounds): Likewise.
30643 (chkp_build_returned_bound): Likewise.
30644 (chkp_get_bound_for_parm): Likewise.
30645 (chkp_build_bndldx): Likewise.
30646 (chkp_get_bounds_by_definition): Likewise.
30647 (chkp_generate_extern_var_bounds): Likewise.
30648 (chkp_get_bounds_for_decl_addr): Likewise.
30649 * tree-chrec.c (chrec_apply): Likewise.
30650 * tree-data-ref.c (dump_data_reference): Likewise.
30651 (dump_subscript): Likewise.
30652 (dump_data_dependence_relation): Likewise.
30653 (analyze_overlapping_iterations): Likewise.
30654 * tree-inline.c (expand_call_inline): Likewise.
30655 (tree_function_versioning): Likewise.
30656 * tree-into-ssa.c (dump_defs_stack): Likewise.
30657 (dump_currdefs): Likewise.
30658 (dump_names_replaced_by): Likewise.
30659 (dump_update_ssa): Likewise.
30660 (update_ssa): Likewise.
30661 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30662 * tree-parloops.c (build_new_reduction): Likewise.
30663 (try_create_reduction_list): Likewise.
30664 (ref_conflicts_with_region): Likewise.
30665 (oacc_entry_exit_ok_1): Likewise.
30666 (oacc_entry_exit_single_gang): Likewise.
30667 * tree-pretty-print.h: Likewise.
30668 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30669 (get_scalar_evolution): Likewise.
30670 (add_to_evolution): Likewise.
30671 (get_loop_exit_condition): Likewise.
30672 (analyze_evolution_in_loop): Likewise.
30673 (analyze_initial_condition): Likewise.
30674 (analyze_scalar_evolution): Likewise.
30675 (instantiate_scev): Likewise.
30676 (number_of_latch_executions): Likewise.
30677 (gather_chrec_stats): Likewise.
30678 (final_value_replacement_loop): Likewise.
30679 (scev_const_prop): Likewise.
30680 * tree-sra.c (dump_access): Likewise.
30681 (disqualify_candidate): Likewise.
30682 (create_access): Likewise.
30683 (reject): Likewise.
30684 (maybe_add_sra_candidate): Likewise.
30685 (create_access_replacement): Likewise.
30686 (analyze_access_subtree): Likewise.
30687 (analyze_all_variable_accesses): Likewise.
30688 (sra_modify_assign): Likewise.
30689 (initialize_constant_pool_replacements): Likewise.
30690 (find_param_candidates): Likewise.
30691 (decide_one_param_reduction): Likewise.
30692 (replace_removed_params_ssa_names): Likewise.
30693 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30694 * tree-ssa-copy.c (dump_copy_of): Likewise.
30695 (copy_prop_visit_cond_stmt): Likewise.
30696 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30697 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30698 (record_equivalences_from_stmt): Likewise.
30699 * tree-ssa-dse.c (compute_trims): Likewise.
30700 (delete_dead_call): Likewise.
30701 (delete_dead_assignment): Likewise.
30702 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30703 (forward_propagate_into_cond): Likewise.
30704 (pass_forwprop::execute): Likewise.
30705 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30706 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30708 (move_computations_worker): Likewise.
30709 (execute_sm): Likewise.
30710 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30711 (remove_exits_and_undefined_stmts): Likewise.
30712 (remove_redundant_iv_tests): Likewise.
30713 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30714 (adjust_iv_update_pos): Likewise.
30715 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30716 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30717 (value_replacement): Likewise.
30718 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30719 * tree-ssa-pre.c (print_pre_expr): Likewise.
30720 (get_representative_for): Likewise.
30721 (create_expression_by_pieces): Likewise.
30722 (insert_into_preds_of_block): Likewise.
30723 (eliminate_insert): Likewise.
30724 (eliminate_dom_walker::before_dom_children): Likewise.
30725 (eliminate): Likewise.
30726 (remove_dead_inserted_code): Likewise.
30727 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30728 * tree-ssa-reassoc.c (get_rank): Likewise.
30729 (eliminate_duplicate_pair): Likewise.
30730 (eliminate_plus_minus_pair): Likewise.
30731 (eliminate_not_pairs): Likewise.
30732 (undistribute_ops_list): Likewise.
30733 (eliminate_redundant_comparison): Likewise.
30734 (update_range_test): Likewise.
30735 (optimize_range_tests_var_bound): Likewise.
30736 (optimize_vec_cond_expr): Likewise.
30737 (rewrite_expr_tree): Likewise.
30738 (rewrite_expr_tree_parallel): Likewise.
30739 (linearize_expr): Likewise.
30740 (break_up_subtract): Likewise.
30741 (linearize_expr_tree): Likewise.
30742 (attempt_builtin_powi): Likewise.
30743 (attempt_builtin_copysign): Likewise.
30744 (transform_stmt_to_copy): Likewise.
30745 (transform_stmt_to_multiply): Likewise.
30746 (dump_ops_vector): Likewise.
30747 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30748 (print_scc): Likewise.
30749 (set_ssa_val_to): Likewise.
30750 (visit_reference_op_store): Likewise.
30751 (visit_use): Likewise.
30752 (sccvn_dom_walker::before_dom_children): Likewise.
30753 (run_scc_vn): Likewise.
30754 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30756 (expr_hash_elt::print): Likewise.
30757 (const_and_copies::pop_to_marker): Likewise.
30758 (const_and_copies::record_const_or_copy_raw): Likewise.
30759 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30760 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30761 (dump_predicates): Likewise.
30762 (find_uninit_use): Likewise.
30763 (warn_uninitialized_phi): Likewise.
30764 (pass_late_warn_uninitialized::execute): Likewise.
30765 * tree-ssa.c (verify_vssa): Likewise.
30766 (verify_ssa): Likewise.
30767 (maybe_optimize_var): Likewise.
30768 * tree-vrp.c (dump_value_range): Likewise.
30769 (dump_all_value_ranges): Likewise.
30770 (dump_asserts_for): Likewise.
30771 (register_edge_assert_for_2): Likewise.
30772 (vrp_visit_cond_stmt): Likewise.
30773 (vrp_visit_switch_stmt): Likewise.
30774 (vrp_visit_stmt): Likewise.
30775 (vrp_visit_phi_node): Likewise.
30776 (simplify_cond_using_ranges_1): Likewise.
30777 (fold_predicate_in): Likewise.
30778 (evrp_dom_walker::before_dom_children): Likewise.
30779 (evrp_dom_walker::push_value_range): Likewise.
30780 (evrp_dom_walker::pop_value_range): Likewise.
30781 (execute_early_vrp): Likewise.
30783 2017-05-16 Richard Biener <rguenther@suse.de>
30785 * dwarf2out.c (loc_list_from_tree_1): Do not create
30786 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30788 2017-05-16 Richard Biener <rguenther@suse.de>
30790 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30792 (note_variable_value_in_expr): If we resolved the decl ref
30793 do not push to the stack.
30795 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30797 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30798 operations in fast-math mode.
30799 (vaddq_f16): Likewise.
30800 (vmul_f16): Likewise.
30801 (vmulq_f16): Likewise.
30802 (vsub_f16): Likewise.
30803 (vsubq_f16): Likewise.
30804 * config/arm/neon.md (add<mode>3): New.
30806 (fma:<VH:mode>3): New. Also remove outdated comment.
30809 2017-05-16 Martin Liska <mliska@suse.cz>
30813 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30814 (odr_types_equivalent_p): Likewise.
30816 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30818 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30820 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30823 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30824 non-interunit SSE move alternatives with '?'.
30825 (zero-extendsidi peephole2): New peephole to skip intermediate
30826 general register in SSE zero-extend sequence.
30828 2017-05-15 Jeff Law <law@redhat.com>
30830 * reorg.c (relax_delay_slots): Create a new variable to hold
30831 the temporary target rather than clobbering TARGET_LABEL.
30833 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30834 missing argument to extract_bit_field call.
30835 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30837 2017-05-15 Martin Liska <mliska@suse.cz>
30840 * gcc.c (process_command): Do not allow empty argument of -o option.
30842 2017-05-15 Renlin Li <renlin.li@arm.com>
30844 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30845 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30846 * config/aarch64/constraints.md (Usf): Add long call check.
30847 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30848 (call_value): Likewise.
30849 (sibcall): Likewise.
30850 (sibcall_value): Likewise.
30852 (call_value_insn): New.
30853 (sibcall_insn): Update rtx pattern.
30854 (sibcall_value_insn): Likewise.
30855 (call_internal): Remove.
30856 (call_value_internal): Likewise.
30857 (sibcall_internal): Likewise.
30858 (sibcall_value_internal): Likewise.
30859 (call_reg): Likewise.
30860 (call_symbol): Likewise.
30861 (call_value_reg): Likewise.
30862 (call_value_symbol): Likewise.
30864 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30867 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30869 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30871 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30872 compatible with CCGOCmode and with CCZmode.
30874 2017-05-14 Martin Sebor <msebor@redhat.com>
30876 PR middle-end/77671
30877 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30878 (gimple_fold_builtin_snprintf): Same.
30879 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30880 (gimple_fold_builtin_snprintf): Same.
30881 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30882 of character types.
30883 (is_call_safe): New function.
30884 (try_substitute_return_value): Call it.
30885 (try_simplify_call): New function.
30886 (pass_sprintf_length::handle_gimple_call): Call it.
30888 2017-05-14 Martin Sebor <msebor@redhat.com>
30890 PR middle-end/80669
30891 * builtins.c (expand_builtin_stpncpy): Simplify.
30893 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30895 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30896 * config/i386/i386.h
30897 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30898 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30899 (struct machine_function): Add new members call_ms2sysv,
30900 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30901 (struct machine_frame_state): New fields sp_realigned and
30902 sp_realigned_offset.
30903 * config/i386/i386.c
30904 (enum xlogue_stub): New enum.
30905 (enum xlogue_stub_sets): New enum.
30906 (class xlogue_layout): New class.
30907 (struct ix86_frame): New fields stack_realign_allocate_offset,
30908 stack_realign_offset and outlined_save_offset. Modify comments to
30909 detail stack layout when using out-of-line stubs.
30910 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30911 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30912 -mcall-ms2sysv-xlogues.
30913 (stub_managed_regs): New static variable.
30914 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30915 registers managed by out-of-line stub.
30916 (disable_call_ms2sysv_xlogues): New function.
30917 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30918 m->call_ms2sysv when appropriate and compute frame layout for
30920 (sp_valid_at, fp_valid_at): New inline functions.
30921 (choose_basereg): New function.
30922 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30924 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30925 Use align parameter of choose_baseaddr to generated aligned SSE movs
30927 (pro_epilogue_adjust_stack): Modify to track
30928 machine_frame_state::sp_realigned.
30929 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30930 (ix86_nsaved_sseregs): Likewise.
30931 (ix86_emit_save_regs): Likewise.
30932 (ix86_emit_save_regs_using_mov): Likewise.
30933 (ix86_emit_save_sse_regs_using_mov): Likewise.
30934 (get_scratch_register_on_entry): Likewise.
30935 (gen_frame_set): New function.
30936 (gen_frame_load): Likewise.
30937 (gen_frame_store): Likewise.
30938 (emit_outlined_ms2sysv_save): Likewise.
30939 (emit_outlined_ms2sysv_restore): Likewise.
30940 (ix86_expand_prologue): Modify stack re-alignment code and call
30941 emit_outlined_ms2sysv_save when appropriate.
30942 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30943 parameter rtx_insn *insn, which allows the function to be used to only
30944 generate the notes.
30945 (ix86_expand_epilogue): Modify validity checks of frame and stack
30946 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30947 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30948 * config/i386/predicates.md
30949 (save_multiple): New predicate.
30950 (restore_multiple): Likewise.
30951 * config/i386/sse.md
30952 (save_multiple<mode>): New pattern.
30953 (save_multiple_realign<mode>): Likewise.
30954 (restore_multiple<mode>): Likewise.
30955 (restore_multiple_and_return<mode>): Likewise.
30956 (restore_multiple_leave_return<mode>): Likewise.
30957 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30959 2017-05-14 Julia Koval <julia.koval@intel.com>
30961 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30962 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30963 (__builtin_ia32_xsetbv): New builtins.
30964 * config/i386/i386.c (ix86_expand_special_args_builtin):
30966 (ix86_expand_builtin): Special expand for new intrinsics.
30967 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30968 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30969 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30971 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30973 * cfganal.c (inverted_post_order_compute): Change argument type
30975 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30976 * df-core.c (rest_of_handle_df_initialize): Adjust.
30977 (rest_of_handle_df_finish): Likewise.
30978 (df_analyze_1): Likewise.
30979 (df_analyze): Likewise.
30980 (loop_inverted_post_order_compute): Change argument to be a vec *.
30981 (df_analyze_loop): Adjust.
30982 (df_get_n_blocks): Likewise.
30983 (df_get_postorder): Likewise.
30984 * df.h (struct df_d): Change field to be a vec.
30985 * lcm.c (compute_laterin): Adjust.
30986 (compute_available): Likewise.
30987 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30988 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30989 * tree-ssa-pre.c (compute_antic): Likewise.
30991 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30993 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30994 (depth_first_search::depth_first_search): Change structure init
30995 function to this constructor.
30996 (depth_first_search::add_bb): Rename function to this member.
30997 (depth_first_search::execute): Likewise.
30998 (flow_dfs_compute_reverse_finish): Adjust.
31000 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31002 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31003 (longest_simple_path): Likewise.
31004 * shrink-wrap.c (spread_components): Likewise.
31005 (disqualify_problematic_components): Likewise.
31006 (emit_common_heads_for_components): Likewise.
31007 (emit_common_tails_for_components): Likewise.
31008 (insert_prologue_epilogue_for_components): Likewise.
31010 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31012 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31015 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31017 * df-core.c (df_set_blocks): Start using auto_bitmap.
31018 (df_compact_blocks): Likewise.
31019 * df-problems.c (df_rd_confluence_n): Likewise.
31020 * df-scan.c (df_insn_rescan_all): Likewise.
31021 (df_process_deferred_rescans): Likewise.
31022 (df_update_entry_block_defs): Likewise.
31023 (df_update_exit_block_uses): Likewise.
31024 (df_entry_block_bitmap_verify): Likewise.
31025 (df_exit_block_bitmap_verify): Likewise.
31026 (df_scan_verify): Likewise.
31027 * lra-constraints.c (lra_constraints): Likewise.
31028 (undo_optional_reloads): Likewise.
31029 (lra_undo_inheritance): Likewise.
31030 * lra-remat.c (calculate_gen_cands): Likewise.
31031 (do_remat): Likewise.
31032 * lra-spills.c (assign_spill_hard_regs): Likewise.
31033 (spill_pseudos): Likewise.
31034 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31035 (bitmap_set_subtract_values): Likewise.
31037 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31039 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31040 management with auto_bitmap.
31041 (fix_inter_tick): Likewise.
31042 (fix_recovery_deps): Likewise.
31043 * ira.c (add_store_equivs): Likewise.
31044 (find_moveable_pseudos): Likewise.
31045 (split_live_ranges_for_shrink_wrap): Likewise.
31046 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31047 (rtx_reuse_manager::seen_def_p): Likewise.
31048 (rtx_reuse_manager::set_seen_def): Likewise.
31049 * print-rtl.h (class rtx_reuse_manager): Likewise.
31051 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31053 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31055 (migrate_btr_def): Likewise.
31056 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31057 * df-core.c (loop_post_order_compute): Likewise.
31058 (loop_inverted_post_order_compute): Likewise.
31059 * hsa-common.h: Likewise.
31060 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31061 * init-regs.c (initialize_uninitialized_regs): Likewise.
31062 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31063 (inline_small_functions): Likewise.
31064 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31065 * ira.c (combine_and_move_insns): Likewise.
31066 (build_insn_chain): Likewise.
31067 * loop-invariant.c (find_invariants): Likewise.
31068 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31069 * predict.c (tree_predict_by_opcode): Likewise.
31070 (predict_paths_leading_to): Likewise.
31071 (predict_paths_leading_to_edge): Likewise.
31072 (estimate_loops_at_level): Likewise.
31073 (estimate_loops): Likewise.
31074 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31075 (spread_components): Likewise.
31076 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31077 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31078 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31079 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31080 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31081 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31082 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31083 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31084 (mark_threaded_blocks): Likewise.
31085 (thread_through_all_blocks): Likewise.
31086 * tree-ssa.c (verify_ssa): Likewise.
31087 (execute_update_addresses_taken): Likewise.
31088 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31090 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31092 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31094 (post_order_compute): Likewise.
31095 (inverted_post_order_compute): Likewise.
31096 (pre_and_rev_post_order_compute_fn): Likewise.
31098 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31100 * genrecog.c (int_set::int_set): Explicitly construct our
31101 auto_vec base class.
31102 * vec.h (auto_vec::auto_vec): New constructor.
31104 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31106 * bitmap.h (class auto_bitmap): New constructor taking
31107 bitmap_obstack * argument.
31109 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31111 * bitmap.h (class auto_bitmap): Change type of m_bits to
31112 bitmap_head, and adjust ctor / dtor and member operators.
31114 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31116 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31117 when returned register mode doesn't match original mode.
31119 2017-05-12 Jeff Law <law@redhat.com>
31120 Jakub Jelinek <jakub@redhat.com>
31122 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31123 we look for cc setter after the compare-elim changes.
31124 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31125 within the vector to match what compare-elim now expects.
31126 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31127 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31129 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31130 after the compare-elim changes.
31131 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31132 the vector to match what compare-elim now expects.
31133 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31134 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31135 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31136 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31137 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31139 * config/visium/visium.c (single_set_and_flags): Fix where
31140 we look for cc setter after the compare-elim changes.
31141 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31142 with the vector to match what compare-elim now expects.
31143 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31144 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31145 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31146 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31147 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31148 (neg<mode>2_insn_set_overflow): Likewise.
31150 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31152 PR middle-end/79794
31153 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31154 maybe_expand_insn call, set ops[0].target. If still set after call,
31155 set alt_rtl. Add extra arg to recursive calls.
31156 (extract_bit_field): Add alt_rtl argument. Pass to
31158 * expmed.h (extract_bit_field): Fix prototype.
31159 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31160 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31161 to extract_bit_field_calls.
31162 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31163 Pass alt_rtl to extract_bit_field calls.
31164 * calls.c (store_unaligned_arguments_into_psuedos)
31165 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31166 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31168 * optabs.h (struct expand_operand): Add target bitfield.
31170 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31172 * compare-elim.c (try_eliminate_compare): Canonicalize
31173 operation with embedded compare to
31174 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31175 (set (reg) (operation)].
31177 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31179 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31182 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31183 cost of adding a carry flag for ADC instruction.
31184 [case MINUS]: Ignore the cost of subtracting a carry flag
31185 for SBB instruction.
31187 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31189 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31191 * config/rs6000/bmiintrin.h: New file.
31192 * config/rs6000/bmi2intrin.h: New file.
31193 * config/rs6000/x86intrin.h: New file.
31195 2017-05-12 Jeff Law <law@redhat.com>
31197 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31200 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31202 PR middle-end/80707
31203 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31205 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31207 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31208 early expansion of vector divide builtins.
31209 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31210 builtins identified as having unsigned arguments.
31212 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31214 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31215 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31216 expansion of vector logical operations (and, andc, or, xor,
31219 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31221 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31222 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31224 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31226 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31227 early GIMPLE expansion of vector multiplies.
31229 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31231 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31232 TARGET_HAVE_MOVT conditional.
31233 (movt splitter): Likewise.
31235 2017-05-12 Richard Biener <rguenther@suse.de>
31237 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31238 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31239 Fold all stmts not inplace.
31241 2017-05-12 Richard Biener <rguenther@suse.de>
31243 PR tree-optimization/80713
31244 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31245 inserted_exprs bit for not removed stmts.
31247 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31249 PR middle-end/69921
31250 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31251 parallelized" attribute for parallelized OpenACC kernels.
31252 * omp-offload.c (execute_oacc_device_lower): Use it.
31254 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31255 Set "oacc kernels" attribute.
31256 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31257 parameter. Adjust all users.
31258 (oacc_fn_attrib_kernels_p): Remove function.
31259 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31260 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31261 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31262 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31263 assert "oacc kernels" attribute is set.
31265 2017-05-11 Carl Love <cel@us.ibm.com>
31267 * config/rs6000/rs6000-c: Add support for built-in functions
31268 vector unsigned char vec_popcnt (vector signed char)
31269 vector unsigned char vec_popcnt (vector unsigned char)
31270 vector unsigned short vec_popcnt (vector signed short)
31271 vector unsigned short vec_popcnt (vector unsigned short)
31272 vector unsigned int vec_popcnt (vector signed int)
31273 vector unsigned int vec_popcnt (vector unsigned int)
31274 vector unsigned long long vec_popcnt (vector signed long long)
31275 vector unsigned long long vec_popcnt (vector unsigned long long)
31276 vector signed long long vec_slo (vector signed long long,
31277 vector signed char)
31278 vector signed long long vec_slo (vector signed long long,
31279 vector unsigned char)
31280 vector unsigned long long vec_slo (vector unsigned long long,
31281 vector signed char)
31282 vector unsigned long long vec_slo (vector unsigned long long,
31283 vector unsigned char)
31284 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31285 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31286 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31287 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31288 * doc/extend.texi: Update the built-in documentation file for the
31289 new built-in functions.
31291 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31293 * attribs.h (sorted_attr_string): Move machine independent
31294 functions for target clone support from the i386 port to common
31295 code. Rename ix86_function_versions to common_function_versions.
31296 Rename make_name to make_unique_name.
31297 (common_function_versions): Likewise.
31298 (make_unique_name): Likewise.
31299 (make_dispatcher_decl): Likewise.
31300 (is_function_default_version): Likewise.
31301 * attribs.c (attr_strcmp): Likewise.
31302 (sorted_attr_string): Likewise.
31303 (common_function_versions): Likewise.
31304 (make_unique_name): Likewise.
31305 (make_dispatcher_decl): Likewise.
31306 (is_function_default_version): Likewise.
31307 * config/i386/i386.c (attr_strcmp): Likewise.
31308 (sorted_attr_string): Likewise.
31309 (ix86_function_versions): Likewise.
31310 (make_name): Likewise.
31311 (make_dispatcher_decl): Likewise.
31312 (is_function_default_version): Likewise.
31313 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31315 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31318 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31319 Account for direct move costs for vec_construct of integer
31322 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31325 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31326 (UNSPEC_STX_ATOMIC): Ditto.
31327 (loaddi_via_sse): New insn.
31328 (storedi_via_sse): Ditto.
31329 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31330 Update corresponding peephole2 patterns.
31331 (atomic_storedi_fpu): Ditto.
31333 2017-05-11 Julia Koval <julia.koval@intel.com>
31335 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31336 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31338 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31339 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31340 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31342 2017-05-11 Nathan Sidwell <nathan@acm.org>
31344 * graphite-poly.c: Include dumpfile.h.
31346 * dumpfle.h (dump_function): Declare here ...
31347 * tree-dump.h (dump_function): ... not here.
31348 * dumpfile.c: #include tree-cfg.h.
31349 (dump_function): Move here from ...
31350 * tree-dump.c (dump_function): ... here.
31351 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31352 * graphite-poly.c: Don't include tree-dump.h.
31353 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31354 * print-tree.c: Likewise.
31355 * stor-layout.c: Likewise.
31356 * tree-nested.c: Likewise.
31358 * dumpfile.c (dump_start): Use TDF_FLAGS.
31359 (dump_enable_all): Fix TDF_KIND check thinko.
31361 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31363 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31364 array entries to represent two legal parameterizations of the
31365 overloaded __builtin_cmpb function, as represented by the
31366 P6_OV_BUILTIN_CMPB constant.
31367 (altivec_resolve_overloaded_builtin): Add special case handling
31368 for the __builtin_cmpb function, as represented by the
31369 P6_OV_BUILTIN_CMPB constant.
31370 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31371 (BU_P6_64BIT_2): New macro.
31372 (BU_P6_OVERLOAD_2): New macro
31373 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31374 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31375 (CMPB): Add overload support to represent both 32-bit and 64-bit
31376 compare-bytes function.
31377 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31378 support for TARGET_CMPB.
31379 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31380 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31381 documentation of the __builtin_cmpb overloaded built-in function.
31383 2017-05-11 Richard Biener <rguenther@suse.de>
31385 PR tree-optimization/80705
31386 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31387 bases are not vectorizable.
31389 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31391 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31392 when counting register pressure.
31394 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31396 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31397 (iv_ca_more_deps): Renamed to ...
31398 (iv_ca_compare_deps): ... this.
31399 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31401 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31403 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31405 (determine_group_iv_costs): ... here.
31406 (find_inv_vars_cb): Record inv var if it's not recorded before.
31408 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31410 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31411 (get_shiftadd_cost): Ditto.
31413 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31415 * tree-ssa-address.c: Include header file.
31416 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31418 (add_to_parts): Refactor.
31419 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31420 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31423 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31425 PR tree-optimization/53090
31426 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31428 (extract_cond_operands): Detect condition with IV on both sides
31429 and return COMP_IV_EXPR_2.
31430 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31431 (rewrite_use_compare): Simplify by removing call to function
31432 extract_cond_operands.
31434 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31436 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31437 (extract_cond_operands): Detect condition comparing against non-
31438 invariant bound and return appropriate enum value.
31439 (find_interesting_uses_cond): Update use of extract_cond_operands.
31440 Handle its return value accordingly.
31441 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31443 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31445 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31446 nonlinear iv_use computation in loop invariant sensitive way.
31448 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31450 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31451 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31453 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31455 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31456 (dump_cand): Support iv_cand.inv_exprs.
31457 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31459 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31462 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31464 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31466 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31467 as local function. Include necessary header files.
31468 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31470 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31472 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31474 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31476 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31477 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31478 RSHIFT_EXPR and BIT_NOT_EXPR.
31480 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31482 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31483 (adjust_setup_cost): New parameter supporting round up adjustment.
31484 (struct address_cost_data): Delete.
31485 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31486 (split_address_cost, ptr_difference_cost): Delete.
31487 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31488 (struct ainc_cost_data): New struct.
31489 (get_address_cost_ainc): New function.
31490 (get_address_cost, get_computation_cost): Reimplement.
31491 (determine_group_iv_cost_address): Record inv_expr for all uses of
31493 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31494 (iv_ca_has_deps): Reimplemented to ...
31495 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31497 (iv_ca_extend): Call iv_ca_more_deps.
31499 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31501 * tree-ssa-address.c (struct mem_address): Move to header file.
31502 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31503 * tree-ssa-address.h (struct mem_address): Move from C file.
31504 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31506 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31508 * tree-affine.h (aff_combination_type): New interface.
31509 (aff_combination_zero_p): Remove static.
31510 (aff_combination_const_p): New interface.
31511 (aff_combination_singleton_var_p): New interfaces.
31513 2017-05-11 Richard Biener <rguenther@suse.de>
31515 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31516 Skip unreachable blocks and destinations.
31517 (eliminate): Move stmt removal and fixup ...
31518 (fini_eliminate): ... here. Skip inserted exprs.
31519 (pass_pre::execute): Move fini_pre after fini_eliminate.
31520 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31521 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31522 PRE to get rid of dead code that has invalid SSA form and
31523 split critical edges again.
31525 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31527 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31529 2017-05-11 Richard Biener <rguenther@suse.de>
31531 * passes.c (execute_function_todo): Verify loops if they are
31532 said to be up-to-date.
31533 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31534 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31536 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31539 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31540 handle calling assemble_external ourself.
31543 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31544 modes with zero size. Enhance comment.
31546 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31548 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31549 built-ins for vec_xl and vec_xst with short and char pointer
31552 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31554 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31555 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31556 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31557 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31558 (_mm_maskz_min_round_ss): New intrinsics.
31559 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31560 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31561 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31562 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31563 (__builtin_ia32_minss_mask_round): New builtins.
31564 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31565 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31566 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31568 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31569 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31571 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31574 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31576 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31577 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31578 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31579 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31580 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31581 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31582 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31583 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31584 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31585 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31586 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31587 (__builtin_ia32_mulss_mask_round): New builtins.
31588 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31589 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31590 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31592 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31593 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31595 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31598 2017-05-10 Julia Koval <julia.koval@intel.com>
31600 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31601 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31602 (_mm256_setr_m128i): New intrinsics.
31604 2017-05-10 Julia Koval <julia.koval@intel.com>
31606 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31607 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31608 (_mm_maskz_rcp14_ss): New intrinsics.
31609 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31610 (__builtin_ia32_rcp14ss_mask): New builtins.
31611 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31613 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31615 PR tree-optimization/51513
31616 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31617 (assert_unreachable_fallthru_edge_p): Use it.
31618 (group_case_labels_stmt): Likewise.
31619 * tree-cfg.h: Prototype it.
31620 * stmt.c: Include cfghooks.h and tree-cfg.h.
31621 (emit_case_dispatch_table) <gap_label>: New local variable.
31622 Use it to fill dispatch table gaps.
31623 Test for default_label before updating probabilities.
31624 (expand_case) <default_label>: Remove unneeded initialization.
31625 Test for unreachable default case statement and remove its edge.
31626 Set default_label accordingly.
31627 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31629 2017-05-10 Carl Love <cel@us.ibm.com>
31631 * config/rs6000/rs6000-c: Add support for built-in functions
31632 vector signed char vec_neg (vector signed char)
31633 vector signed short int vec_neg (vector short int)
31634 vector signed int vec_neg (vector signed int)
31635 vector signed long long vec_neg (vector signed long long)
31636 vector float vec_neg (vector float)
31637 vector double vec_neg (vector double)
31638 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31640 * config/rs6000/altivec.h: Add define for vec_neg
31641 * doc/extend.texi: Update the built-in documentation for the
31642 new built-in functions.
31644 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31646 PR tree-optimization/77644
31647 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31649 2017-05-10 Nathan Sidwell <nathan@acm.org>
31651 * dumpfile.h (TDI_lang_all): New.
31652 (TDF_KIND): New. Renumber others
31653 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31655 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31657 (get_dump_file_name): Adjust suffix generation.
31658 (dump_enable_all): Use TDF_KIND.
31659 * doc/invoke.texi (-fdump-lang-all): Document.
31661 * dumpfile.h: Tabify.
31663 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31666 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31667 Move member access before delete.
31669 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31671 * tree-inline.c (expand_call_inline): Split block at stmt
31674 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31677 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31678 are now unused after splitting mov{sf,sd}_hardfloat.
31679 (f32_lr2): Likewise.
31680 (f32_lm): Likewise.
31681 (f32_lm2): Likewise.
31682 (f32_li): Likewise.
31683 (f32_li2): Likewise.
31684 (f32_lv): Likewise.
31685 (f32_sr): Likewise.
31686 (f32_sr2): Likewise.
31687 (f32_sm): Likewise.
31688 (f32_sm2): Likewise.
31689 (f32_si): Likewise.
31690 (f32_si2): Likewise.
31691 (f32_sv): Likewise.
31692 (f32_dm): Likewise.
31693 (f32_vsx): Likewise.
31694 (f32_av): Likewise.
31695 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31696 For movsf, order stores so the VSX stores occur before the GPR
31697 store which encourages the register allocator to use a traditional
31698 FPR instead of a GPR. For movsd, order the stores so that the GPR
31699 store comes before the VSX stores to allow the power6 to work.
31700 This is due to the power6 not having a 32-bit integer store
31701 instruction from a FPR.
31702 (movsf_hardfloat): Likewise.
31703 (movsd_hardfloat): Likewise.
31705 2017-05-09 Martin Sebor <msebor@redhat.com>
31707 PR translation/80280
31708 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31711 PR translation/80280
31712 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31713 data member added in r247778.
31714 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31716 2017-05-09 Nathan Sidwell <nathan@acm.org>
31718 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31720 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31723 2017-05-09 Marek Polacek <polacek@redhat.com>
31725 * doc/invoke.texi: Fix typo.
31727 2017-05-09 Richard Biener <rguenther@suse.de>
31729 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31730 (vrp_val_is_min): Likewise.
31731 (set_value_range_to_value): Likewise.
31732 (set_value_range_to_nonnegative): Likewise.
31733 (gimple_assign_nonzero_p): Likewise.
31734 (gimple_stmt_nonzero_p): Likewise.
31735 (vrp_int_const_binop): Likewise. Remove unreachable case.
31736 (adjust_range_with_scev): Adjust comments.
31737 (compare_range_with_value): Likewise.
31738 (extract_range_from_phi_node): Likewise.
31739 (test_for_singularity): Likewise.
31741 2017-05-09 Richard Biener <rguenther@suse.de>
31743 * tree-vrp.c (get_single_symbol): Add assert that we don't
31744 get overflowed constants as invariant part.
31745 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31746 checks. Use wi::cmp instead of recursing for integer constants.
31747 (compare_values): Just ignore whether we assumed undefined
31748 overflow instead of failing the compare.
31749 (extract_range_for_var_from_comparison_expr): Add comment before the
31750 TREE_NO_WARNING sets.
31751 (test_for_singularity): Likewise.
31752 (extract_range_from_comparison): Do not disable optimization
31753 when we assumed undefined overflow.
31754 (extract_range_basic): Remove init of unused var.
31756 2017-05-09 Richard Biener <rguenther@suse.de>
31758 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31759 (extract_range_from_multiplicative_op_1): Adjust.
31760 (extract_range_from_binary_expr_1): Use int_const_binop.
31762 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31765 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31766 rs6000_store_data_bypass_p in seven define_bypass directives and
31767 in several comments.
31768 * config/rs6000/rs6000-protos.h: Add prototype for
31769 rs6000_store_data_bypass_p function.
31770 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31771 function implements slightly different (rs6000-specific) semantics
31772 than store_data_bypass_p, returning false rather than aborting
31773 with assertion error when arguments do not satisfy the
31774 requirements of store data bypass.
31775 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31776 rs6000_store_data_bypass_p.
31778 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31780 * config/xtensa/xtensa-protos.h
31781 (xtensa_initial_elimination_offset): New declaration.
31782 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31783 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31784 macro definition, add case for FRAME_POINTER_REGNUM when
31785 FRAME_GROWS_DOWNWARD.
31786 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31787 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31788 xtensa_initial_elimination_offset.
31790 2017-05-08 Nathan Sidwell <nathan@acm.org>
31792 * doc/invoke.texi: Alphabetize -fdump options.
31794 2017-05-08 Martin Sebor <msebor@redhat.com>
31796 PR translation/80280
31797 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31799 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31801 * target.def (compute_frame_layout): New optional target hook.
31802 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31803 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31804 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31806 * reload1.c (verify_initial_elim_offsets): Likewise.
31807 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31808 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31809 (arm_get_frame_offsets): Split up into this ...
31810 (arm_compute_frame_layout): ... and this function.
31812 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31814 * config/aarch64/constraints.md (Usa): New constraint.
31815 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31817 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31819 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31820 with_multilib_list after it has been checked.
31822 2017-05-08 Richard Biener <rguenther@suse.de>
31824 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31825 (bitmap_set_subtract_values): Likewise.
31827 2017-05-08 Richard Biener <rguenther@suse.de>
31829 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31830 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31832 (gimple_stmt_nonzero_warnv_p): Rename to ...
31833 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31835 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31836 (extract_range_basic): Adjust, do not disable propagation on
31837 strict overflow sensitive simplification.
31838 (vrp_visit_cond_stmt): Likewise.
31840 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31842 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31843 body size unconditionally.
31845 2017-05-07 Jeff Law <law@redhat.com>
31848 2017-05-06 Jeff Law <law@redhat.com>
31849 PR tree-optimization/78496
31850 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31853 PR tree-optimization/78496
31854 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31855 (simplify_stmt_using_ranges): Call it.
31856 (vrp_dom_walker::before_dom_children): Extract equivalences
31857 from an ASSERT_EXPR with an equality comparison against a
31860 2017-05-06 Jeff Law <law@redhat.com>
31862 PR tree-optimization/78496
31863 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31866 PR tree-optimization/78496
31867 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31868 (simplify_stmt_using_ranges): Call it.
31869 (vrp_dom_walker::before_dom_children): Extract equivalences
31870 from an ASSERT_EXPR with an equality comparison against a
31873 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31875 * lra-constraints.c (lra_copy_reg_equiv): New function.
31876 (split_reg): Use it to copy equivalence information from the
31877 original register to the spill register.
31879 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31881 PR rtl-optimization/75964
31882 * simplify-rtx.c (simplify_const_relational_operation): Remove
31883 invalid handling of comparisons of integer ABS.
31885 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31887 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31888 initialize to zero.
31889 (init_regs): Remove declaration.
31890 (function_arg_advance_32): Initialize error_p as boolean variable.
31892 2017-05-05 Nathan Sidwell <nathan@acm.org>
31894 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31895 lines. Use for (;;).
31897 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31899 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31900 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31901 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31902 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31903 VF=2 that require versioning.
31905 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31907 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31910 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31912 * diagnostic.h (diagnostic_override_option_index): Convert from
31913 macro to inline function.
31915 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31917 * diagnostic.c (last_module_changed_p): New function.
31918 (set_last_module): New function.
31919 (diagnostic_report_current_module): Convert macro usage to
31920 the above functions.
31921 * diagnostic.h (diagnostic_context::last_module): Strengthen
31922 from const line_map * to const line_map_ordinary *.
31923 (diagnostic_last_module_changed): Delete macro.
31924 (diagnostic_set_last_module): Delete macro.
31926 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31928 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31929 with diagnostic_report_diagnostic.
31930 (diagnostic_n_impl_richloc): Likewise.
31931 * diagnostic.h (report_diagnostic): Delete macro.
31932 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31933 with diagnostic_report_diagnostic.
31934 * substring-locations.c (format_warning_va): Likewise.
31936 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31938 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31939 save/restor of format_spec. Move option-printing code to...
31940 (print_option_information): ...this new function, and
31941 reimplement by simply printing to the pretty_printer,
31942 rather than appending to the format string.
31944 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31946 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31947 handling logic into...
31948 (update_effective_level_from_pragmas): ...this new function.
31950 2017-05-04 Andrew Waterman <andrew@sifive.com>
31952 * config/riscv/riscv.opt (mstrict-align): New option.
31953 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31954 (SLOW_UNALIGNED_ACCESS): Define.
31955 (riscv_slow_unaligned_access): Declare.
31956 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31958 (riscv_slow_unaligned_access): New variable.
31959 (rocket_tune_info): Set slow_unaligned_access to true.
31960 (optimize_size_tune_info): Set slow_unaligned_access to false.
31961 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31962 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31963 (riscv_option_override): Set riscv_slow_unaligned_access.
31964 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31966 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31968 * config/riscv/riscv.md: Unify indentation.
31970 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31975 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31977 (extendsi<mode>2): Add support for doing sign extension via
31978 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31979 don't have ISA 3.0 instructions.
31980 (extendsi<mode>2 splitter): Likewise.
31981 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31982 generate the normal insns since SImode can now go in vector
31983 registers. Disallow the special UNSPECs needed for previous
31984 machines to hide SImode being used. Add new insns
31985 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31986 (fix_trunc<mode>si2_stfiwx): Likewise.
31987 (fix_trunc<mode>si2_internal): Likewise.
31988 (fixuns_trunc<mode>si2): Likewise.
31989 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31990 (fctiw<u>z_<mode>_smallint): Likewise.
31991 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31992 of floating point to 32-bit integer from doing a direct move to
31993 the GPR registers to do a store.
31994 (fctiwz_<mode>): Break long line.
31996 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31998 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31999 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32000 (addr_list, addr_offset_valid_p): New.
32001 (split_address_groups): Check offset validity with above function.
32002 (gt-tree-ssa-loop-ivopts.h): Include header file.
32004 2017-05-05 Nathan Sidwell <nathan@acm.org>
32006 * config.gcc (arm*-*-*): Add missing 'fi'.
32008 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32010 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32011 included in -fopt-info does not matter.
32012 * doc/optinfo.texi (-fopt-info): Fix description of default
32013 behavour. Explicitly say order of options included in -fopt-info
32016 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32018 * config.gcc: Allow combinations of aprofile and rmprofile values for
32019 --with-multilib-list.
32020 * config/arm/t-multilib: New file.
32021 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32022 variables. Remove setting of ISA and floating-point ABI in
32023 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32024 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32025 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32026 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32028 * config/arm/t-rmprofile: Likewise except for the matches changes.
32029 * doc/install.texi (--with-multilib-list): Document the combination of
32030 aprofile and rmprofile values and warn about pitfalls in doing that.
32032 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32034 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32035 (movdi_aarch64): Likewise.
32037 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32039 PR tree-optimization/80632
32040 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32042 (build_arrays): Initialize it for virtual phis.
32043 (fix_phi_nodes): Use it for virtual phis.
32045 PR tree-optimization/80558
32046 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32047 [x, y] op z into [x op, y op z] for op & or | if conditions
32050 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32051 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32054 * config/arm/arm.md (use_literal_pool): Remove.
32055 (64-bit immediate split): No longer takes cost into consideration
32056 if arm_disable_literal_pool is enabled.
32057 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32058 used when arm_disable_literal_pool is enabled.
32059 (arm_max_const_double_inline_cost): Remove use of
32060 arm_disable_literal_pool.
32061 (push_minipool_fix): Add assert.
32062 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32063 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32064 (no_literal_pool_sf_immediate): New.
32066 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32068 PR tree-optimization/80613
32069 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32070 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32072 2017-05-05 Richard Biener <rguenther@suse.de>
32074 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32076 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32078 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32079 of this flag from insn conditions due to removal from r247495.
32081 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32083 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32085 (arm_early_store_addr_dep_ptr): Likewise.
32086 * config/arm/aarch-common-protos.h
32087 (arm_early_load_addr_dep_ptr): Add prototype.
32088 (arm_early_store_addr_dep_ptr): Likewise.
32089 * config/arm/cortex-a53.md: Add new bypasses.
32091 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32093 * tree.c (next_type_uid): Change type to unsigned.
32094 (type_hash_canon): Decrement back next_type_uid if
32095 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32096 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32099 2017-05-04 Martin Sebor <msebor@redhat.com>
32101 * builtins.c: Fix a trivial typo in a comment.
32103 PR middle-end/79234
32104 * builtins.c (check_sizes): Adjust to handle reading past the end.
32105 Avoid printing excessive upper bound of ranges. Use %E to print
32106 tree nodes instead of converting them to %wu.
32107 (expand_builtin_memchr): New function.
32108 (compute_dest_size): Rename...
32109 (compute_objsize): ...to this.
32110 (expand_builtin_memcpy): Adjust.
32111 (expand_builtin_mempcpy): Adjust.
32112 (expand_builtin_strcat): Adjust.
32113 (expand_builtin_strcpy): Adjust.
32114 (check_strncat_sizes): Adjust.
32115 (expand_builtin_strncat): Adjust.
32116 (expand_builtin_strncpy): Adjust and simplify.
32117 (expand_builtin_memset): Adjust.
32118 (expand_builtin_bzero): Adjust.
32119 (expand_builtin_memcmp): Adjust.
32120 (expand_builtin): Handle memcmp.
32121 (maybe_emit_chk_warning): Check strncat just once.
32123 2017-05-04 Martin Sebor <msebor@redhat.com>
32125 PR preprocessor/79214
32126 PR middle-end/79222
32127 PR middle-end/79223
32128 * builtins.c (check_sizes): Add inlining context and issue
32129 warnings even when -Wno-system-headers is set.
32130 (check_strncat_sizes): Same.
32131 (expand_builtin_strncat): Same.
32132 (expand_builtin_memmove): New function.
32133 (expand_builtin_stpncpy): Same.
32134 (expand_builtin): Handle memmove and stpncpy.
32136 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32138 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32139 which is not used any more.
32141 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32143 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32145 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32147 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32148 (cortexa53_tunings): Likewise.
32149 (cortexa57_tunings): Likewise.
32150 (cortexa72_tunings): Likewise.
32151 (cortexa73_tunings): Likewise.
32153 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32155 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32156 Set loop alignment to 8.
32158 2017-05-04 Martin Sebor <msebor@redhat.com>
32160 PR translation/80280
32161 * builtins.c (expand_builtin_object_size): Add missing quoting to
32162 %D and like directives.
32163 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32164 (hsa_type_for_tree_type): Same.
32165 (verify_function_arguments): Same.
32166 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32167 * varasm.c (get_section): Same.
32170 2017-05-04 Martin Sebor <msebor@redhat.com>
32172 PR translation/80280
32173 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32175 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32177 * config/aarch64/aarch64.c (generic_addrcost_table):
32178 Change HI/TI mode setting.
32180 2017-05-04 Martin Jambor <mjambor@suse.cz>
32182 PR tree-optimization/80622
32183 * tree-sra.c (comes_initialized_p): New function.
32184 (build_accesses_from_assign): Only set write lazily when
32185 comes_initialized_p is false.
32186 (analyze_access_subtree): Use comes_initialized_p.
32187 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32188 instead of testing for PARM_DECL.
32190 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32192 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32193 constraint on operand 0 to allow more general addressing modes.
32194 Adjust output template.
32195 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32197 * config/aarch64/aarch64-protos.h
32198 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32199 * config/aarch64/constraints.md (Dp): New address constraint.
32200 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32203 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32205 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32206 update use of estimate_ipcp_clone_size_and_time.
32207 (estimate_local_effects): Update use of
32208 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32209 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32210 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32211 Return nonspecialized time.
32213 2017-05-04 Richard Biener <rguenther@suse.de>
32215 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32216 for the last VUSE which def dominates the PHI. Directly call
32218 (get_continuation_for_phi_1): Remove.
32220 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32222 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32223 to explain the use of truncating division. Cap the number of
32224 iterations to the maximum given by nb_iterations_upper_bound,
32227 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32229 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32230 * configure: Regenerate.
32231 * config.in: Regenerate.
32232 * config/i386/driver-mingw32.c: new file.
32233 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32234 * config.host: Link driver-mingw32.o on MinGW host.
32235 * doc/install.texi: Document new --enable-mingw-wildcard configure
32238 2017-05-04 Marek Polacek <polacek@redhat.com>
32240 PR tree-optimization/80612
32241 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32243 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32244 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32246 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32247 (movt splitter): Likewise.
32248 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32249 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32250 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32251 block for Thumb-1 with MOVT.
32252 (thumb2_legitimate_address_p): Move code block ...
32253 (can_avoid_literal_pool_for_label_p): ... into this new function.
32254 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32256 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32257 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32258 "M-profile targets with the MOVT instruction".
32260 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32262 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32263 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32264 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32266 2017-05-04 Martin Liska <mliska@suse.cz>
32268 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32269 variable cond_code.
32271 2017-05-04 Richard Biener <rguenther@suse.de>
32273 * tree.c (array_at_struct_end_p): Handle arrays at struct
32274 end with flexarrays more conservatively. Refactor and treat
32275 arrays of arrays or aggregates more strict. Fix
32276 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32277 * tree.c (array_at_struct_end_p): Adjust prototype.
32278 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32279 * gimple-fold.c (get_range_strlen): Likewise.
32280 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32282 2017-05-04 Richard Biener <rguenther@suse.de>
32284 PR tree-optimization/31130
32285 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32287 (supports_overflow_infinity): Likewise.
32288 (is_negative_overflow_infinity): Likewise.
32289 (is_positive_overflow_infinity): Likewise.
32290 (is_overflow_infinity): Likewise.
32291 (stmt_overflow_infinity): Likewise.
32292 (overflow_infinity_range_p): Likewise.
32293 (usable_range_p): Remove as always returning true.
32294 (make_overflow_infinity): Remove.
32295 (negative_overflow_infinity): Likewise.
32296 (positive_overflow_infinity): Likewise.
32297 (avoid_overflow_infinity): Likewise.
32298 (set_value_range): Adjust accordingly.
32299 (set_value_range_to_nonnegative): Likewise, remove now unused
32300 overflow_infinity arg.
32301 (vrp_operand_equal_p): Adjust.
32302 (update_value_range): Likewise.
32303 (range_int_cst_singleton_p): Likewise.
32304 (operand_less_p): Likewise.
32305 (compare_values_warnv): Likewise.
32306 (extract_range_for_var_from_comparison_expr): Likewise.
32307 (vrp_int_const_binop): Likewise.
32308 (zero_nonzero_bits_from_vr): Likewise.
32309 (extract_range_from_multiplicative_op_1): Likewise.
32310 (extract_range_from_binary_expr_1): Likewise.
32311 (extract_range_from_unary_expr): Likewise.
32312 (extract_range_from_comparison): Likewise.
32313 (extract_range_basic): Likewise.
32314 (adjust_range_with_scev): Likewise.
32315 (compare_ranges): Likewise.
32316 (compare_range_with_value): Likewise.
32317 (dump_value_range): Likewise.
32318 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32320 (simplify_cond_using_ranges): Adjust.
32322 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32324 * brig-builtins.def: Added a builtin for class_f64.
32325 * builtin-types.def: Added a builtin type needed by class_f64.
32327 2017-05-03 Jason Merrill <jason@redhat.com>
32329 * timevar.def: Add TV_CONSTEXPR.
32331 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32333 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32335 2017-05-03 Martin Jambor <mjambor@suse.cz>
32337 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32338 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32339 * ipa-cp.c (ipcp_propagate_stage): Do not call
32340 ipa_update_after_lto_read.
32341 * ipa-inline.c (ipa_inline): Likewise.
32343 2017-05-03 Martin Jambor <mjambor@suse.cz>
32345 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32346 tag. Added a default constructor and a destructor.
32347 (ipa_edge_args_sum_t): New class;
32348 (ipa_edge_args_sum): Declare.
32349 (ipa_edge_args_vector): Remove declaration.
32350 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32351 (ipa_free_edge_args_substructures): Remove declaration.
32352 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32353 (ipa_edge_args_info_available_for_edge_p): Likewise.
32354 * ipa-prop.c (ipa_edge_args_vector): Removed.
32355 (edge_removal_hook_holder): Likewise.
32356 (edge_duplication_hook_holder): Likewise.
32357 (ipa_edge_args_sum): New variable.
32358 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32359 ipa_edge_args_vector.
32360 (ipa_free_edge_args_substructures): Likewise.
32361 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32362 ipa_edge_args_vector.
32363 (ipa_edge_removal_hook): Turned into method
32364 ipa_edge_args_sum_t::remove.
32365 (ipa_edge_duplication_hook): Turned into method
32366 ipa_edge_args_sum_t::duplicate.
32367 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32368 registering edge hooks.
32369 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32370 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32371 ipa_edge_args_sum instead of ipa_edge_args_vector.
32372 * ipa-profile.c (ipa_profile): Likewise.
32374 2017-05-03 Martin Jambor <mjambor@suse.cz>
32376 * symbol-summary.h (function_summary): New method exists.
32377 (function_summary::symtab_removal): Deallocate through release.
32378 (call_summary): New class.
32379 (gt_ggc_mx): New overload.
32380 (gt_pch_nx): Likewise.
32381 (gt_pch_nx): Likewise.
32383 2017-05-03 Jeff Law <law@redhat.com>
32385 PR tree-optimization/78496
32386 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32387 from simplify_cond_using_ranges. Split off code to walk
32388 backwards through casts into ...
32389 (simplify_cond_using_ranges_2): New function.
32390 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32391 (execute_vrp): After identifying jump threads, call
32392 simplify_cond_using_ranges_2.
32394 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32397 * ipa-inline.h (inline_summary): Add ctor.
32398 (create_ggc): Do not use ggc_cleared_alloc.
32400 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32401 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32403 * gcc.c (handle_braces): Support escaping in switch matching
32405 * doc/invoke.texi (Spec Files): Document it.
32406 Remove superfluous @code markup in items.
32408 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32410 * diagnostic-show-locus.c (struct column_range): New struct.
32411 (get_affected_columns): New function.
32412 (get_printed_columns): New function.
32413 (struct correction): New struct.
32414 (correction::ensure_capacity): New function.
32415 (correction::ensure_terminated): New function.
32416 (struct line_corrections): New struct.
32417 (line_corrections::~line_corrections): New dtor.
32418 (line_corrections::add_hint): New function.
32419 (layout::print_trailing_fixits): Reimplement in terms of the new
32421 (selftest::test_overlapped_fixit_printing): New function.
32422 (selftest::diagnostic_show_locus_c_tests): Call it.
32424 2017-05-03 Nathan Sidwell <nathan@acm.org>
32426 Canonicalize canonical type hashing
32427 * tree.h (type_hash_canon_hash): Declare.
32428 * tree.c (type_hash_list, attribute_hash_list): Move into
32429 type_hash_canon_hash.
32430 (build_type_attribute_qual_variant): Break out hash code calc into
32431 type_hash_canon_hash.
32432 (type_hash_canon_hash): New. Generic type hash computation.
32433 (build_range_type_1, build_array_type_1, build_function_type,
32434 build_method_type_directly, build_offset_type, build_complex_type,
32435 make_vector_type): Call it.
32437 2017-05-03 Richard Biener <rguenther@suse.de>
32439 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32440 When all DRs have unknown misaligned do not always peel
32441 when there is a store but apply the same costing model as if
32442 there were only loads.
32444 2017-05-03 Richard Biener <rguenther@suse.de>
32447 PR tree-optimization/80492
32448 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32449 compare_base_decls returning dont-know properly.
32451 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32453 * config/arm/iterators.md (CCSI): New mode iterator.
32454 (arch): New mode attribute.
32455 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32456 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32457 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32458 code iterator for success result mode.
32459 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32460 the corresponding new insn generators.
32462 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32465 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32466 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32468 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32470 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32471 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32472 (DDR_A): Wrap DDR argument in brackets.
32473 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32474 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32475 (DDR_REVERSED_P): Likewise.
32477 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32479 PR tree-optimization/79472
32480 * tree-switch-conversion.c (struct switch_conv_info): Add
32481 contiguous_range and default_case_nonstandard fields.
32482 (collect_switch_conv_info): Compute contiguous_range and
32483 default_case_nonstandard fields, don't clear final_bb if
32484 contiguous_range and only the default case doesn't have the required
32486 (check_all_empty_except_final): Set default_case_nonstandard instead
32487 of failing if contiguous_range and the default case doesn't have empty
32489 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32490 and only the default case doesn't have the required constants. Skip
32492 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32493 if default_case_nonstandard.
32494 (build_constructors): Build constant 1 just once. Assert that default
32495 values aren't inserted in between cases if contiguous_range. Skip
32497 (build_arrays): Skip virtual phis.
32498 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32499 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32500 Handle virtual phis.
32501 (gen_inbound_check): Handle default_case_nonstandard case.
32502 (process_switch): Adjust check_final_bb caller. Call
32503 gather_default_values with the first non-default case instead of
32504 default case if default_case_nonstandard.
32506 2017-05-02 Nathan Sidwell <nathan@acm.org>
32508 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32509 check. Fix formatting.
32511 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32513 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32514 errors when comparing specialized and unspecialized times.
32516 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32518 * diagnostic-show-locus.c
32519 (layout::should_print_annotation_line_p): Make private.
32520 (layout::print_annotation_line): Make private.
32521 (layout::annotation_line_showed_range_p): Make private.
32522 (layout::show_ruler): Make private.
32523 (layout::print_source_line): Make private. Pass in line and
32524 line_width, rather than calling location_get_source_line. Drop
32526 (layout::print_leading_fixits): New method.
32527 (layout::print_any_fixits): Rename to...
32528 (layout::print_trailing_fixits): ...this, and make private.
32529 Don't print newline fixits.
32530 (diagnostic_show_locus): Move logic for printing one row into...
32531 (layout::print_line): ...this new function. Move the
32532 location_get_source_line call and error-handling from
32533 print_source_line to here. Call print_leading_fixits, and rename
32534 print_any_fixits to print_trailing_fixits.
32535 (selftest::test_fixit_insert_containing_newline): Update now that
32536 newlines are partially supported.
32537 (selftest::test_fixit_insert_containing_newline_2): New test.
32538 (selftest::test_fixit_replace_containing_newline): Update comments.
32539 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32540 * edit-context.c (class added_line): New class.
32541 (class edited_line): Describe newline handling in comment.
32542 (edited_line::actually_edited_p): New method.
32543 (edited_line::print_content): Delete redundant decl.
32544 (edited_line::m_predecessors): New field.
32545 (edited_file::print_content): Call edited_line::print_content.
32546 (edited_file::print_diff): Update to support newlines.
32547 (edited_file::print_diff_hunk): Likewise.
32548 (edited_file::print_run_of_changed_lines): New function.
32549 (edited_file::print_diff_line): Convert to...
32550 (print_diff_line): ...this.
32551 (edited_file::get_effective_line_count): New function.
32552 (edited_line::edited_line): Initialize new field m_predecessors.
32553 (edited_line::~edited_line): Clean up m_predecessors.
32554 (edited_line::apply_fixit): Handle newlines.
32555 (edited_line::get_effective_line_count): New function.
32556 (edited_line::print_content): New function.
32557 (edited_line::print_diff_lines): New function.
32558 (selftest::test_applying_fixits_insert_containing_newline): New
32560 (selftest::test_applying_fixits_replace_containing_newline): New
32562 (selftest::insert_line): New function.
32563 (selftest::test_applying_fixits_multiple_lines): Add example of
32565 (selftest::edit_context_c_tests): Call the new tests.
32567 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32569 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32570 parameter cand. Update dump information.
32571 (get_computation_cost): Update uses.
32573 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32575 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32576 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32577 (get_computation_at, rewrite_use_address): Update use of
32578 get_computation_aff.
32580 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32582 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32583 (get_computation): Delete.
32584 (get_computation_cost): Implement like get_computation_cost_at.
32585 Use get_computation_at.
32586 (get_computation_cost_at): Delete.
32587 (rewrite_use_nonlinear_expr): Use get_computation_at.
32588 (rewrite_use_compare, remove_unused_ivs): Ditto.
32590 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32592 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32594 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32596 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32597 (ivopts_global_cost_for_size): Rename parameter and update uses.
32598 (iv_ca_recount_cost): Update uses.
32599 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32600 candidates seperately in n_invs and n_cands.
32601 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32603 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32605 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32606 (find_inv_vars_cb): New.
32607 (find_depends): Renamed to ...
32608 (find_inv_vars): ... this.
32609 (add_candidate_1, force_var_cost): Call find_inv_vars.
32610 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32612 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32614 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32615 inv_vars. Add inv_exprs.
32616 (struct iv_cand): Rename depends_on to inv_vars.
32617 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32618 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32619 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32620 (dump_cand): Dump inv_vars.
32621 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32622 (record_invariant, find_depends, add_candidate_1): Ditto.
32623 (set_group_iv_cost, force_var_cost): Ditto.
32624 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32625 (get_computation_cost_at, get_computation_cost): Ditto.
32626 (determine_group_iv_cost_generic): Ditto.
32627 (determine_group_iv_cost_address): Ditto.
32628 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32629 (determine_group_iv_costs): Ditto.
32630 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32631 (iv_ca_set_remove_invariants): Renamed to ...
32632 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32633 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32634 (iv_ca_set_add_invariants): Renamed to ...
32635 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32636 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32637 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32638 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32639 (create_new_ivs): Remove useless dump.
32641 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32643 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32645 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32646 (iv_ca_set_no_cp, create_new_iv): Ditto.
32648 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32650 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32652 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32654 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32655 function tree_check2.
32657 2017-05-02 Martin Liska <mliska@suse.cz>
32659 * doc/gcov.texi: Add missing preposition.
32660 * gcov.c (function_info::function_info): Properly fill up
32661 all member variables.
32663 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32665 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32667 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32669 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32671 2017-05-02 Martin Liska <mliska@suse.cz>
32674 * lto-streamer-in.c (lto_read_tree_1): Remove
32675 LTO_STREAMER_DEBUG.
32676 * lto-streamer.c (struct tree_hash_entry): Likewise.
32677 (struct tree_entry_hasher): Likewise.
32678 (tree_entry_hasher::hash): Likewise.
32679 (tree_entry_hasher::equal): Likewise.
32680 (lto_streamer_init): Likewise.
32681 (lto_orig_address_map): Likewise.
32682 (lto_orig_address_get): Likewise.
32683 (lto_orig_address_remove): Likewise.
32684 * lto-streamer.h: Likewise.
32685 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32686 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32688 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32690 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32691 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32692 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32693 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32694 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32695 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32696 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32697 (mm_maskz_sub_ss): New intrinsics.
32698 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32699 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32700 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32701 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32702 (__builtin_ia32_subss_mask_round): New builtins.
32703 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32704 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32705 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32707 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32708 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32710 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32713 2017-05-02 Martin Jambor <mjambor@suse.cz>
32715 PR tree-optimization/78687
32716 * tree-sra.c (access): New field parent.
32717 (process_subtree_disqualification): New function.
32718 (disqualify_candidate): Call it.
32719 (build_accesses_from_assign): Reset write flag if creating an
32721 (build_access_subtree): Fill in parent field and also prpagate
32722 down grp_write flag.
32723 (create_artificial_child_access): New parameter set_grp_write, set
32724 grp_write to its value.
32725 (propagate_subaccesses_across_link): Also propagate grp_write flag
32727 (propagate_all_subaccesses): Push the closest parent back to work
32728 queue if add_access_to_work_queue returned true.
32730 2017-05-02 Richard Biener <rguenther@suse.de>
32732 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32733 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32734 -fstrict-overflow documentation.
32735 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32736 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32737 flag_strict_overflow.
32738 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32739 * lto-opts.c (lto_write_options): Do not stream it.
32740 * lto-wrapper.c (merge_and_complain): Do not handle it.
32741 * opts.c (default_options_table): Do not set -fstrict-overflow.
32742 (finish_options): Likewise do not clear it when sanitizing.
32743 * simplify-rtx.c (simplify_const_relational_operation): Do not
32744 test flag_strict_overflow.
32746 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32748 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32749 using enabled attribute.
32750 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32751 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32752 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32753 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32754 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32755 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32756 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32757 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32758 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32759 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32761 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32763 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32765 2017-05-02 Richard Biener <rguenther@suse.de>
32767 PR tree-optimization/80591
32769 2017-04-10 Richard Biener <rguenther@suse.de>
32771 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32774 2017-05-02 Richard Biener <rguenther@suse.de>
32776 PR tree-optimization/80549
32777 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32778 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32779 headers if they do not have a preheader.
32781 2017-05-02 Martin Liska <mliska@suse.cz>
32784 * common.opt: Fix typo.
32785 * doc/invoke.texi: Likewise.
32787 2017-05-01 Jan Beulich <jbeulich@suse.com>
32789 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32790 swapping, add (x,x,m,x,n) alternative.
32792 2017-05-01 Nathan Sidwell <nathan@acm.org>
32794 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32795 unnecessary unadjusted_alignment check.
32797 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32800 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32802 * gimplify.c (gimplify_cilk_detach): New function.
32803 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32804 * tree-core.h: Document EXPR_CILK_SPAWN.
32805 * tree.h (EXPR_CILK_SPAWN): Define.
32807 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32809 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32810 to use new fixit_hint representation, using the "replace" logic.
32811 (get_line_span_for_fixit_hint): Likewise.
32812 (layout::print_any_fixits): Likewise.
32813 (selftest::test_one_liner_many_fixits): Rename to...
32814 (selftest::test_one_liner_many_fixits_1): ...this, and update
32815 comment and expected output to reflect that the multiple fix-it
32816 hints are now consolidated into one insertion.
32817 (selftest::test_one_liner_many_fixits_2): New test.
32818 (selftest::test_diagnostic_show_locus_one_liner): Update for
32820 (selftest::test_fixit_consolidation): Update for fix-it API
32822 * diagnostic.c (print_parseable_fixits): Likewise.
32823 * edit-context.c (edited_line::m_line_events): Convert from
32824 auto_vec <line_event *> to auto_vec <line_event>.
32825 (class line_event): Convert from abstract base class to a concrete
32826 class, taking over the role of replace_event.
32827 (class insert_event): Delete.
32828 (class replace_event): Rename to class line_event. Convert to
32830 (edit_context::add_fixits): Reimplement.
32831 (edit_context::apply_insert): Delete.
32832 (edit_context::apply_replace): Rename to...
32833 (edit_context::apply_fixit): ...this. Convert to half-open range.
32834 (edited_file::apply_insert): Delete.
32835 (edited_file::apply_replace): Rename to...
32836 (edited_file::apply_fixit): ...this.
32837 (edited_line::~edited_line): Drop deletion of events.
32838 (edited_line::apply_insert): Delete.
32839 (edited_line::apply_replace): Rename to...
32840 (edited_line::apply_fixit): ...this. Convert to half-open range.
32841 Update for change to type of m_line_events.
32842 * edit-context.h (edit_context::apply_insert): Delete.
32843 (edit_context::apply_replace): Rename to...
32844 (edit_context::apply_fixit): ...this.
32846 2017-05-01 Martin Sebor <msebor@redhat.com>
32848 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32851 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32854 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32855 __get_cpuid_max returns 0.
32856 (__get_cpuid_count): Ditto.
32858 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32860 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32861 replacement expression is another instance of one of its arguments.
32863 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32866 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32867 check for stack push/pop autoinc.
32868 * config/i386/i386.c (ix86_agi_dependent): Return false
32869 if the only reason why modified_in_p returned true is that
32870 addr is SP based and set_insn is a push or pop.
32872 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32874 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32877 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32880 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32881 (account_size_time): Use two predicates - exec_pred and
32883 (evaluate_conditions_for_known_args): Compute both clause and
32885 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32886 (inline_summary_t::duplicate): Update.
32887 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32889 (compute_inline_parameters): Likewise.
32890 (estimate_edge_size_and_time): Update caluclation of time.
32891 (estimate_node_size_and_time): Compute both time and nonspecialized
32893 (estimate_ipcp_clone_size_and_time): Update.
32894 (inline_merge_summary): Update.
32895 (do_estimate_edge_time): Update.
32896 (do_estimate_edge_size): Update.
32897 (do_estimate_edge_hints): Update.
32898 (inline_read_section, inline_write_summary): Stream both new predicates.
32899 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32901 (compute_inlined_call_time): Cleanup.
32902 (big_speedup_p): Update.
32903 (edge_badness): Update.
32904 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32905 (size_time_entry): Replace predicate by exec_predicate and
32906 nonconst_predicate.
32907 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32908 (estimate_edge_time): Return also nonspec_time.
32909 (reset_edge_growth_cache): Update.
32911 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32913 PR rtl-optimization/80491
32914 * ifcvt.c (noce_process_if_block): When looking for x setter
32915 with missing else_bb, don't check only the insn right before
32916 cond_earliest, but look for the last insn that x is modified in
32917 within the same bb.
32919 PR rtl-optimization/80491
32920 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32922 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32924 PR tree-optimization/80487
32925 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32927 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32929 PR tree-optimization/79697
32930 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32931 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32932 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32934 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32935 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32937 2017-04-28 Martin Sebor <msebor@redhat.com>
32939 PR tree-optimization/80523
32940 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32941 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32943 (maybe_warn, format_directive, parse_directive): Use new functions.
32944 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32946 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32948 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32950 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32952 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32953 target_header_dir): Set correctly.
32954 * configure: Regenerated.
32955 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32956 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32957 instead of SYSTEM_HEADER_DIR.
32959 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32961 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32962 (estimate_local_effects): Likewise.
32963 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32964 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32965 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32966 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32967 do_estimate_edge_time, estimate_edge_time): Likewise.
32968 * ipa-inline-analysis.c (estimate_node_size_and_time,
32969 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32970 (estimate_time_after_inlining): Remove.
32972 2017-04-28 Martin Liska <mliska@suse.cz>
32974 * doc/gcov.texi: Enhance documentation of gcov.
32976 2017-04-28 Martin Liska <mliska@suse.cz>
32978 * doc/gcov.texi: Sort options in alphabetic order.
32979 * doc/gcov-dump.texi: Likewise.
32980 * doc/gcov-tool.texi: Likewise.
32981 * gcov.c (print_usage): Likewise.
32982 * gcov-dump.c (print_usage): Likewise.
32983 * gcov-tool.c (print_merge_usage_message): Likewise.
32984 (print_rewrite_usage_message): Likewise.
32985 (print_overlap_usage_message): Likewise.
32987 2017-04-28 Martin Liska <mliska@suse.cz>
32989 PR gcov-profile/53915
32990 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32992 2017-04-28 Martin Liska <mliska@suse.cz>
32994 PR gcov-profile/79891
32995 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32996 is marked by compiler as living on a line.
32997 (get_cycles_count): Remove usage of the union.
32998 (output_intermediate_file): Likewise.
32999 (find_source): Fix GNU coding style.
33000 (accumulate_line_counts): Remove old non-all block mode.
33001 (output_lines): Remove usage of the union.
33002 * profile.c (output_location): Include all BBs, even if
33003 belonging to a same line (and file) as a previous BB.
33005 2017-04-28 Martin Liska <mliska@suse.cz>
33007 * gcov.c (process_args): Handle new argument 'w'.
33008 (read_graph_file): Assign ID to BBs.
33009 (output_branch_count): Display BB # if verbose flag is set.
33010 (output_lines): Likewise for arcs.
33011 (print_usage): Add '--verbose' option help.
33012 * doc/gcov.texi: Document --verbose (-w) option.
33014 2017-04-28 Martin Liska <mliska@suse.cz>
33016 * gcov.c (struct block_location_info): New struct.
33017 (process_file): Fill up the new structure.
33018 (read_graph_file): Replace usage of encoding by the newly added
33020 (add_line_counts): Likewise.
33021 (accumulate_line_counts): Remove usage of the union.
33022 (function_info::function_info): New function.
33023 (function_info::~function_info): Likewise.
33024 (process_file): Call delete instead of release_function.
33025 (release_function): Release the function.
33026 (release_structures): Call delete instead of release_function.
33027 (solve_flow_graph): Replace usage of num_blocks.
33028 (find_exception_blocks): Likewise.
33029 (output_lines): Fix GNU coding style.
33031 2017-04-28 Martin Liska <mliska@suse.cz>
33034 * coverage.c (coverage_remove_note_file): New function.
33035 * coverage.h: Declare the function.
33036 * toplev.c (finalize): Clean if an error has been seen.
33038 2017-04-28 Martin Liska <mliska@suse.cz>
33040 PR gcov-profile/80031
33041 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33042 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33043 * gcov.c (read_graph_file): Read just number of blocks.
33044 * profile.c (branch_prob): Do not stream 0 flags per a basic
33047 2017-04-28 Martin Liska <mliska@suse.cz>
33049 * gcov-dump.c (tag_*): Add new argument to declarations.
33050 (dump_gcov_file): Likewise.
33051 (tag_blocks): Add and use new argument depth.
33052 (tag_arcs): Likewise.
33053 (tag_lines): Likewise.
33054 (tag_counters): Likewise.
33055 (tag_summary): Likewise.
33056 (dump_working_sets): Use depth to do a proper indentation.
33058 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33061 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33062 * symtab.c (symtab_node::debug_symtab): Move definition here.
33064 2017-04-28 Richard Biener <rguenther@suse.de>
33066 * lto-streamer.h (LTO_major_version): Bump to 7.
33068 2017-04-28 Richard Biener <rguenther@suse.de>
33070 * tree-vrp.c (assert_info): New struct.
33071 (add_assert_info): New helper.
33072 (register_edge_assert_for_2): Refactor to add asserts to a vector
33074 (register_edge_assert_for_1): Likewise.
33075 (register_edge_assert_for): Likewise.
33076 (finish_register_edge_assert_for): New helper actually registering
33077 asserts where live on edge.
33078 (find_conditional_asserts): Adjust.
33079 (find_switch_asserts): Likewise.
33080 (evrp_dom_walker::try_find_new_range): Generalize.
33081 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33083 2017-04-27 Marek Polacek <polacek@redhat.com>
33086 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33087 arg10 and arg11 to itype.
33089 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33091 * doc/extend.texi (Object Size Checking): Improve grammar.
33093 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33096 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33097 that the logic for permitting reciprocal estimates matches that
33100 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33103 * tree.c (type_cache_hasher::equal): Only compare
33104 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33105 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33106 non-aggregate element types.
33107 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33108 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33110 2017-04-27 Richard Biener <rguenther@suse.de>
33112 PR middle-end/80533
33113 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33114 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33115 keeping a reference to a trailing array.
33117 2017-04-27 Richard Biener <rguenther@suse.de>
33119 PR middle-end/80539
33120 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33121 being in loop-closed SSA form conservatively.
33122 (chrec_fold_multiply_poly_poly): Likewise.
33124 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33126 PR middle-end/79665
33127 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33128 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33130 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33133 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33134 (aarch64_function_arg_alignment): Return unsigned int again, but still
33135 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33136 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33137 Don't emit -Wpsabi note.
33138 (aarch64_function_arg_boundary): Likewise.
33139 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33142 2017-04-26 Nathan Sidwell <nathan@acm.org>
33144 * tree.h (crc32_unsigned_n): Declare.
33145 (crc32_unsigned, crc32_unsigned): Make inline.
33146 * tree.c (crc32_unsigned_bits): Replace with ...
33147 (crc32_unsigned_n): ... this.
33148 (crc32_unsigned, crc32_byte): Remove.
33149 (crc32_string): Remove unnecessary braces.
33151 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33153 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33154 * ipa-inline-analysis.c (MAX_TIME): Remove.
33155 (account_size_time): Use sreal for time.
33156 (dump_inline_summary): Update.
33157 (estimate_function_body_sizes): Update.
33158 (estimate_edge_size_and_time): Update.
33159 (estimate_calls_size_and_time): Update.
33160 (estimate_node_size_and_time): Update.
33161 (inline_merge_summary): Update.
33162 (inline_update_overall_summary): Update.
33163 (estimate_time_after_inlining): Update.
33164 (inline_read_section): Update.
33165 (inline_write_summary): Update.
33166 * ipa-inline.c (compute_uninlined_call_time): Update.
33167 (compute_inlined_call_time): Update.
33168 (recursive_inlining): Update.
33169 (inline_small_functions): Update.
33170 (dump_overall_stats): Update.
33171 * ipa-inline.h: Include sreal.h.
33172 (size_time_entry): Turn time to sreal.
33173 (inline_summary): Turn self_time nad time to sreal.
33175 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33177 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33179 (sreal::stream_out, sreal::stream_in): New.
33180 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33182 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33184 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33187 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33190 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33192 (dimode_scalar_chain::compute_convert_gain): Ditto.
33193 (dimode_scalar_chain::make_vector_copies): Ditto.
33194 (dimode_scalar_chain::convert_reg): Ditto.
33195 (dimode_scalar_chain::convert_insn): Ditto.
33196 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33197 (VI248_AVX512BW_1): New mode iterator.
33198 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33199 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33202 2017-04-25 Martin Sebor <msebor@redhat.com>
33204 PR tree-optimization/80497
33205 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33206 constants are representable in HOST_WIDE_INT.
33207 (parse_directive): Ditto.
33209 2017-04-25 Martin Sebor <msebor@redhat.com>
33212 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33213 (new_zero_array): Adjust signature.
33214 (dom_info::dom_init): Used unsigned rather that size_t.
33215 (dom_info::dom_info): Same.
33217 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33218 Jakub Jelinek <jakub@redhat.com>
33221 * config/arm/arm.c: Include gimple.h.
33222 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33223 returns negative, increment ncrn only if it returned positive.
33224 (arm_needs_doubleword_align): Return int instead of bool,
33225 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33226 members, but if there is any such non-FIELD_DECL
33227 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33228 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33229 returns negative, increment nregs only if it returned positive.
33230 (arm_setup_incoming_varargs): Likewise.
33231 (arm_function_arg_boundary): Emit -Wpsabi note if
33232 arm_needs_doubleword_align returns negative, return
33233 DOUBLEWORD_ALIGNMENT only if it returned positive.
33235 2017-04-25 Marek Polacek <polacek@redhat.com>
33238 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33239 first argument to type.
33241 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33244 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33245 type checks to test for compatibility instead of equality.
33247 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33248 Jakub Jelinek <jakub@redhat.com>
33251 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33253 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33254 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33255 the alignment computation, but return their maximum in warn_alignment.
33256 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33257 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33259 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33260 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33263 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33265 * config/arc/simdext.md (dmpyh): Fix typo.
33267 2017-04-25 Richard Biener <rguenther@suse.de>
33269 PR tree-optimization/80492
33270 * alias.c (compare_base_decls): Handle registers with asm
33271 specification conservatively.
33272 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33273 compare_base_decls returning dont-know properly.
33275 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33277 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33278 (legitimate_offset_address_p): New function.
33279 (arc_legitimate_address_p): Use above function.
33281 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33283 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33285 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33287 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33288 ACCH registers whenever they are available.
33290 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33292 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33293 double regs fix when not used.
33295 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33297 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33299 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33300 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33302 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33304 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33305 when emitting short ADD instructions.
33307 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33309 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33311 (cmpsi_cc_c_insn): Likewise.
33312 (cbranchsi4_scratch): Compute proper instruction length using
33313 compact_hreg_operand.
33314 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33316 2017-04-25 Richard Biener <rguenther@suse.de>
33318 PR middle-end/80509
33319 * passes.c (pass_manager::pass_manager): Initialize
33320 m_name_to_pass_map.
33322 2017-04-25 Richard Biener <rguenther@suse.de>
33324 PR tree-optimization/79201
33325 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33327 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33330 * config/s390/vector.md: Split MEM->GPR vector moves for
33331 non-s_operand addresses.
33333 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33336 * config/s390/predicates.md (reload_const_wide_int_operand): New
33338 * config/s390/s390.md ("movti"): Remove d/P alternative.
33339 ("movti_bigconst"): New pattern definition.
33341 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33344 * s390-protos.h (s390_expand_cs_hqi): Removed.
33345 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33346 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33347 modes as well as CCZ1mode and CCZmode.
33348 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33349 signature of s390_emit_compare_and_swap.
33350 (s390_expand_cs_hqi): Likewise, make static.
33351 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33352 compare-and-swap, in some cases.
33353 (s390_expand_cs): Wrapper function.
33354 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33356 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33357 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33358 patterns for small and large integers. Forbid symref memory operands.
33359 Move expander to s390.c. Require cc register.
33360 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33361 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33362 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33363 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33364 symref memory operands. Remove CC mode and call s390_match_ccmode
33366 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33368 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33370 * config/s390/s390.md (define_peephole2): New peephole to help
33371 combining the load-and-test pattern with volatile memory.
33373 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33375 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33376 with CCZmode for TARGET_Z196.
33378 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33380 PR rtl-optimization/80501
33381 * combine.c (make_compound_operation_int): Set subreg_code to SET
33382 even for AND with mask of the sign bit of mode.
33384 PR rtl-optimization/80500
33385 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33386 sum's initial value.
33388 2017-04-25 Julian Brown <julian@codesourcery.com>
33389 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33391 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33393 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33395 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33397 2017-04-25 Julian Brown <julian@codesourcery.com>
33398 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33400 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33401 (thunderx2t99_sha): New Reservation.
33403 2017-04-25 Julian Brown <julian@codesourcery.com>
33404 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33406 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33407 type for 1-element load.
33409 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33411 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33413 2017-04-24 Martin Jambor <mjambor@suse.cz>
33415 PR tree-optimization/80293
33416 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33417 char arrays not totally scalarizable if it is false.
33418 (analyze_all_variable_accesses): Pass correct value in the new
33419 parameter. Add a statistics counter.
33421 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33423 PR middle-end/79931
33424 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33426 2017-04-24 Richard Biener <rguenther@suse.de>
33428 PR tree-optimization/80494
33429 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33430 out for complex types.
33432 2017-04-24 Richard Biener <rguenther@suse.de>
33434 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33435 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33436 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33437 (DFS): Adjust and never fail.
33438 (sccvn_dom_walker::fail): Remove.
33439 (sccvn_dom_walker::before_dom_children): Adjust.
33440 (run_scc_vn): Likewise and never fail.
33441 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33442 (pass_fre::execute): Likewise.
33444 2017-04-24 Richard Biener <rguenther@suse.de>
33446 PR tree-optimization/79725
33447 * tree-ssa-sink.c (statement_sink_location): Return whether
33448 failure reason was zero uses. Move that check later.
33449 (sink_code_in_bb): Deal with zero uses by removing the stmt
33452 2017-04-24 Richard Biener <rguenther@suse.de>
33455 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33456 pointer-based references.
33458 2017-04-24 Richard Biener <rguenther@suse.de>
33461 * pass_manager.h (pass_manager::operator new): Remove.
33462 (pass_manager::operator delete): Likewise.
33463 * passes.c (pass_manager::operator new): Remove.
33464 (pass_manager::operator delete): Likewise.
33465 (pass_manager::pass_manager): Zero individual pass members.
33467 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33470 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33471 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33472 Check "XEXP (src, 1)" operand here.
33473 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33474 Check "XEXP (src, 1)" operand here.
33475 (dimode_scalar_chain::make_vector_copies): Detect count register
33476 of a shift instruction. Zero extend count register from QImode
33477 to DImode to satisfy vector shift pattern count operand predicate.
33478 Substitute vector shift count operand with a DImode copy.
33479 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33482 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33484 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33485 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33486 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33487 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33488 (UNSPEC_NOREX_MEM): Remove definition.
33490 2017-04-21 Richard Biener <rguenther@suse.de>
33492 PR tree-optimization/79547
33493 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33494 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33495 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33496 without any constraints.
33498 2017-04-21 Richard Biener <rguenther@suse.de>
33500 PR tree-optimization/78847
33501 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33503 2017-04-21 Richard Biener <rguenther@suse.de>
33505 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33506 (build_distinct_type_copy): Likewise.
33507 (build_variant_type_copy): Likewise.
33508 * tree.c (build_qualified_type): Pass down mem-stat info.
33509 (build_distinct_type_copy): Likewise.
33510 (build_variant_type_copy): Likewise.
33512 2017-04-21 Richard Biener <rguenther@suse.de>
33514 PR tree-optimization/80237
33515 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33517 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33518 for a simplified result.
33520 2016-04-21 Richard Biener <rguenther@suse.de>
33522 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33523 sth as strict as a simple_iv but a chrec without symbols and an
33524 operand defined in the loop we are peeling (and not some subloop).
33525 (propagate_constants_for_unrolling): Propagate all constants.
33527 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33530 * config/i386/i386.c (print_reg): Remove assert for disalowed
33531 regno values, call output_operand_lossage instead.
33533 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33536 * config/i386/constraints.md (Yc): New register constraint.
33537 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33538 Use Yc constraint for alternative 2 of operand 0. Remove
33539 preferred_for_speed attribute.
33541 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33543 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33544 lastprivate clauses in SIMT case.
33546 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33548 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33550 2017-04-20 Richard Biener <rguenther@suse.de>
33552 PR tree-optimization/57796
33553 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33554 as N scalar stores.
33555 (vect_model_load_cost): Cost gathers as N scalar loads.
33557 2017-04-20 Richard Biener <rguenther@suse.de>
33559 * ggc-page.c (ggc_allocated_p): Rename to ...
33560 (safe_lookup_page_table_entry): ... this and return the lookup
33562 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33564 2017-04-20 Richard Biener <rguenther@suse.de>
33566 PR tree-optimization/80453
33567 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33568 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33569 from the conditions.
33570 (vn_phi_eq): Pass them down.
33571 (vn_phi_lookup): Record them.
33572 (vn_phi_insert): Likewise.
33574 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33576 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33577 uninitialized variable warning to avoid buffer overrun.
33579 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33582 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33583 is suppressed for '{ 0 }' in C.
33585 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33587 * BASE-VER: Set to 8.0.0.
33589 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33591 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33592 priority .init_array and .fini_array section with SECTION_NOTYPE
33595 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33597 PR middle-end/80423
33598 * tree.h (build_array_type): Add typeless_storage default argument.
33599 * tree.c (type_cache_hasher::equal): Also compare
33600 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33601 (build_array_type): Add typeless_storage argument, set
33602 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33604 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33605 (build_array_type): Likewise. Add typeless_storage argument.
33607 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33608 Jakub Jelinek <jakub@redhat.com>
33610 PR tree-optimization/80426
33611 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33612 operation on symbolic operands, also compute the overflow for the
33613 invariant part when the operation degenerates into a negation.
33615 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33618 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33619 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33622 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33624 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33627 * config/avr/avr.c (tree.h): Include it.
33628 (cgraph.h): Include it.
33629 (avr_encode_section_info): Don't warn for uninitialized progmem
33630 variable if it's just an alias.
33632 2017-04-19 Richard Biener <rguenther@suse.de>
33635 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33636 when needed by AutoPGO.
33638 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33641 * doc/lto.texi: Remove an extra 'that'.
33643 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33645 PR rtl-optimization/80429
33646 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33647 are only used in debug insns.
33649 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33650 Vladimir Makarov <vmakarov@redhat.com>
33652 * config/sparc/predicates.md (input_operand): Add comment. Return
33653 true for any memory operand when LRA is in progress.
33654 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33656 2017-04-18 Jeff Law <law@redhat.com>
33659 * mips.md ({return,simple_return}_internal): Do not overwrite
33662 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33664 PR tree-optimization/80443
33665 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33666 instead of adding 1, subtract -1 and similarly instead of subtracting
33669 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33671 PR rtl-optimization/80357
33672 * haifa-sched.c (tmp_bitmap): New variable.
33673 (model_recompute): Handle duplicate use records.
33674 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33675 (free_global_sched_pressure_data): Free it.
33677 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33680 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33681 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33682 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33683 instead of SYSTEM_HEADER_DIR.
33685 2017-04-18 Jeff Law <law@redhat.com>
33687 PR middle-end/80422
33688 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33689 predecessors after walking up the insn chain.
33691 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33694 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33695 sizetype type into debug info.
33697 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33700 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33701 unneeded test for TARGET_UPPER_REGS_SF.
33702 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33704 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33707 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33708 instead of gsi_after_labels.
33710 2017-04-18 Jeff Law <law@redhat.com>
33712 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33716 2017-04-13 Jeff Law <law@redhat.com>
33717 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33718 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33720 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33723 * config/avr/avr.c (intl.h): Include it.
33724 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33726 2017-04-18 Martin Liska <mliska@suse.cz>
33728 PR gcov-profile/78783
33729 * gcov-tool.c (gcov_output_files): Validate that destination
33730 file is either removed by the tool or by a user.
33732 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33733 Guy Benyei <guybe@mellanox.com>
33735 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33736 block, and do not negate it, the stored id is already negative.
33738 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33740 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33742 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33745 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33746 masks of options that should be turned off if the VSX vector
33747 options are turned off.
33748 (OTHER_P8_VECTOR_MASKS): Likewise.
33749 (OTHER_VSX_VECTOR_MASKS): Likewise.
33750 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33751 rs6000_disable_incompatible_switches to validate no type switches
33753 (rs6000_incompatible_switch): New function to disallow turning on
33754 other vector options if -mno-vsx, -mno-power8-vector, or
33755 -mno-power9-vector are specified.
33757 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33759 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33761 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33763 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33764 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33765 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33766 (ARG_POINTER_CFA_OFFSET): Likewise.
33768 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33770 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33771 conditions to take advantage of various optimizations.
33773 2017-04-13 Jeff Law <law@redhat.com>
33775 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33776 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33777 (zero_extendsidi2_dext): Likewise.
33779 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33782 * fold-const.c (fold_ternary_loc): Revert
33783 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33786 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33788 PR rtl-optimization/80343
33789 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33790 new scratch pseudo.
33792 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33795 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33796 to ubsan_encode_value.
33798 2017-04-13 Jeff Law <law@redhat.com>
33800 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33801 appearing in DEBUG_INSNs.
33803 2017-04-13 Martin Liska <mliska@suse.cz>
33805 PR gcov-profile/80413
33806 * gcov-io.c (gcov_write_string): Copy to buffer just when
33807 allocated size is greater than zero.
33809 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33812 * dwarf2out.c (decls_for_scope): Ignore declarations of
33813 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33815 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33818 * ipa-visibility.c (non_local_p): Fix typos.
33819 (localize_node): When localizing symbol in same comdat group,
33820 dissolve the group only when we know external symbols are going
33822 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33824 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33826 PR tree-optimization/79390
33827 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33828 order does not result in usable sequence, retry with reversed operand
33834 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33835 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33836 op0 instead of fold_convert_loc (loc, type, arg0).
33838 2017-04-12 Jeff Law <law@redhat.com>
33840 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33841 has a delay slot in the generated code.
33843 * config/cris/cris.md (cris_preferred_reload_class): Return
33844 GENNONACR_REGS rather than GENERAL_REGS.
33846 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33849 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33850 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33851 signedness of the result type.
33853 2017-04-12 Richard Biener <rguenther@suse.de>
33854 Jeff Law <law@redhat.com>
33856 PR tree-optimization/80359
33857 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33858 trim stores to TARGET_MEM_REFs.
33860 2017-04-12 Richard Biener <rguenther@suse.de>
33862 PR tree-optimization/79390
33863 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33864 threading case even more.
33866 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33869 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33870 for quad_address_p for TImode, instead of just not indexed_address.
33872 2017-04-12 Richard Biener <rguenther@suse.de>
33873 Bernd Edlinger <bernd.edlinger@hotmail.de>
33875 PR middle-end/79671
33876 * alias.c (component_uses_parent_alias_set_from): Handle
33877 TYPE_TYPELESS_STORAGE.
33878 (get_alias_set): Likewise.
33879 * tree-core.h (tree_type_common): Add typeless_storage flag.
33880 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33881 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33882 for types containing members with TYPE_TYPELESS_STORAGE.
33883 (place_field): Likewise.
33884 (layout_type): Likewise for ARRAY_TYPE.
33885 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33886 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33887 TYPE_TYPELESS_STORAGE.
33888 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33890 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33893 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33894 first argument to type.
33896 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33900 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33901 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33902 (rs6000_expand_binop_builtin): Likewise.
33903 (rs6000_expand_ternop_builtin): Likewise; also add missing
33904 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33905 vshasigma built-ins.
33906 * doc/extend.texi: Document that vec_xxpermdi's third argument
33907 must be a constant.
33909 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33911 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33912 Use shift_const cost parameter when calculating gain of STV shifts.
33914 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33916 PR rtl-optimization/70478
33917 * lra-constraints.c (process_alt_operands): Check memory for
33918 disfavoring memory insn operand.
33920 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33922 PR middle-end/80100
33923 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33924 left shift in unsigned HOST_WIDE_INT type.
33926 PR rtl-optimization/80385
33927 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33928 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33931 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33932 if they have any depend clauses.
33934 2017-04-11 Martin Liska <mliska@suse.cz>
33937 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33938 * ipa-split.c (split_function): Create a local comdat symbol
33939 if caller is in a comdat group.
33941 2017-04-11 Martin Liska <mliska@suse.cz>
33944 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33947 2017-04-11 Martin Sebor <msebor@redhat.com>
33949 PR middle-end/80364
33950 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33951 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33953 (directive::set_width, directive::set_precision, format_character):
33955 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33958 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33961 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33962 conflict, set target->arch_name instead of target->cpu_name.
33964 2017-04-11 Richard Biener <rguenther@suse.de>
33966 PR tree-optimization/80374
33967 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33968 build_zero_cst, remove fold_convertible_p check again.
33970 2017-04-11 Martin Liska <mliska@suse.cz>
33973 * ubsan.c (instrument_object_size): Do not instrument register
33976 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33979 * config/i386/i386-builtin-types.def
33980 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33981 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33982 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33983 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33984 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33985 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33986 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33987 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33988 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33989 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33990 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33991 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33992 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33993 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33994 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33995 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33996 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33997 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33998 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33999 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34000 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34001 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34002 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34003 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34004 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34005 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34006 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34007 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34008 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34009 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34010 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34011 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34012 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34013 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34014 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34015 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34016 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34017 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34018 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34019 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34020 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34021 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34022 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34023 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34024 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34025 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34027 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34028 flag to second_arg_count, handle 4 argument function type _COUNT
34029 aliases, handle second_arg_count on second argument rather than last.
34031 2017-04-10 Jeff Law <law@redhat.com>
34033 PR tree-optimization/80374
34034 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34035 record anything if we can not convert integer_zero_node to the
34038 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34041 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34042 Enhance special handling given to the TARGET_P9_MINMAX option in
34043 relation to certain other options.
34045 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34047 PR tree-optimization/80153
34048 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34049 remove POINTER_PLUS_EXPR's base part directly, rather than through
34052 2017-04-10 Richard Biener <rguenther@suse.de>
34053 Bin Cheng <bin.cheng@arm.com>
34055 PR tree-optimization/80153
34056 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34057 the first element of pointer type aff_tree. Build result expr in
34059 (add_elt_to_tree): Convert to type unconditionally. Remove other
34060 fold_convert calls.
34061 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34062 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34064 2017-04-10 Richard Biener <rguenther@suse.de>
34066 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34069 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34071 PR rtl-optimization/70478
34072 * lra-constraints.c (curr_small_class_check): New.
34073 (update_and_check_small_class_inputs): New.
34074 (process_alt_operands): Update curr_small_class_check. Disfavor
34075 alternative insn memory operands. Check available regs for small
34078 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34081 * config/mips/mips.opt (-mvirt): Update description.
34082 * doc/invoke.texi (-mvirt): Likewise.
34084 2017-04-10 Richard Biener <rguenther@suse.de>
34086 PR middle-end/80362
34087 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34088 looking for NEGATE_EXPR in -A / -B to A / B folding.
34090 2017-04-10 Martin Liska <mliska@suse.cz>
34092 PR gcov-profile/80224
34093 * gcov.c (print_usage): Fix usage string.
34094 (get_gcov_intermediate_filename): Remove.
34095 (output_gcov_file): Use both for normal and intermediate format.
34096 (generate_results): Do not initialize special file for
34097 intermediate format.
34099 2017-04-10 Richard Biener <rguenther@suse.de>
34101 PR tree-optimization/80304
34102 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34105 2017-04-10 Nathan Sidwell <nathan@acm.org>
34108 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34109 (rs6000_init_builtins): Use it.
34111 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34113 * config/arm/arm.md (<mrc>): Add mode to SET source.
34114 (<mrrc>): Likewise.
34116 2017-04-10 Richard Biener <rguenther@suse.de>
34118 PR middle-end/80344
34119 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34121 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34124 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34125 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34126 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34127 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34128 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34129 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34130 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34131 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34132 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34133 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34134 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34135 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34136 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34137 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34138 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34139 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34140 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34141 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34142 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34143 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34144 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34145 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34146 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34148 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34150 PR rtl-optimization/70478
34151 * lra-constraints.c: Reverse the last patch.
34153 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34155 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34156 Add comment for WCHAR_T.
34158 2017-04-08 Martin Liska <mliska@suse.cz>
34161 2017-04-07 Martin Liska <mliska@suse.cz>
34164 * ipa-split.c (split_function): Add function part to a same comdat
34167 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34170 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34172 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34174 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34175 (vec_pairing): ...to this.
34176 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34177 (rs6000_sched_init): Adjust for name change.
34178 (struct rs6000_sched_context): Likewise.
34179 (rs6000_init_sched_context): Likewise.
34180 (rs6000_set_sched_context): Likewise.
34182 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34188 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34190 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34191 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34192 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34194 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34196 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34198 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34200 PR rtl-optimization/70703
34201 * ira-color.c (update_conflict_hard_regno_costs): Use
34202 int64_t instead of HOST_WIDE_INT.
34204 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34206 PR rtl-optimization/70478
34207 * lra-constraints.c (process_alt_operands): Disfavor alternative
34208 insn memory operands.
34210 2017-04-07 Jeff Law <law@redhat.com>
34212 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34213 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34215 2017-04-07 Martin Liska <mliska@suse.cz>
34218 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34219 Show error message instead of an ICE.
34221 2017-04-07 Martin Liska <mliska@suse.cz>
34224 * ipa-split.c (split_function): Add function part to a same comdat
34227 2017-04-07 Richard Biener <rguenther@suse.de>
34229 PR middle-end/80341
34230 * tree.c (get_unwidened): Also handle ! for_type case for
34232 * convert.c (do_narrow): Split out from ...
34233 (convert_to_integer_1): ... here. Do not pass final truncation
34234 type to get_unwidened for TRUNC_DIV_EXPR.
34236 2017-04-07 Richard Biener <rguenther@suse.de>
34238 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34240 (aff_combination_const): Adjust.
34241 (aff_combination_scale): Likewise.
34242 (aff_combination_add_elt): Likewise.
34243 (aff_combination_add_cst): Likewise.
34244 (aff_combination_convert): Likewise.
34245 (add_elt_to_tree): Likewise. Remove unused argument.
34246 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34248 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34250 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34252 * config/arm/arm.c (arm_default_short_enums): Use
34253 ARM_DEFAULT_SHORT_ENUMS.
34254 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34256 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34259 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34260 members with redundant out-of-class redeclaration.
34262 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34265 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34266 * config/i386/i386.md (*zero_extendsidi2):
34267 Add (?*x,*x) and (?*v,*v) alternatives.
34269 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34272 * config/i386/i386.c (ix86_expand_builtin)
34273 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34274 mode from insn data. Convert operands to insn operand mode.
34275 Copy operands that don't satisfy insn predicate to a register.
34277 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34279 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34282 2017-04-06 Richard Biener <rguenther@suse.de>
34284 PR tree-optimization/80334
34285 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34286 preserve alignment of accesses.
34288 2017-04-06 Richard Biener <rguenther@suse.de>
34290 PR tree-optimization/80262
34291 * tree-sra.c (build_ref_for_offset): Preserve address-space
34293 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34294 Drop useless address-space information on MEM_REF offsets.
34296 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34298 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34300 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34302 PR rtl-optimization/70703
34303 * ira-color.c (update_conflict_hard_regno_costs): Use
34304 HOST_WIDE_INT instead of long.
34306 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34309 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34310 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34312 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34313 for x86_64 target. Handle -m3dnowa option.
34315 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34317 PR rtl-optimization/70703
34318 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34319 (update_conflict_hard_regno_costs): Use long instead of unsigned
34320 arithmetic for cost calculation.
34322 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34323 Bernd Edlinger <bernd.edlinger@hotmail.de>
34326 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34329 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34332 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34333 ptr_mode with Pmode throughout.
34334 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34335 into probe_stack_range and use DImode.
34337 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34340 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34341 call_eh_return is true.
34343 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34345 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34346 Initialize last_match_fntype_index.
34348 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34351 * tree-nvr.c: Include internal-fn.h.
34352 (pass_return_slot::execute): Ignore internal calls without
34355 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34356 Richard Biener <rguenther@suse.de>
34359 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34360 captures used multiple times, except for the last use.
34361 * generic-match-head.c: Include gimplify.h.
34363 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34365 PR tree-optimization/79390
34366 * target.h (struct noce_if_info): Declare.
34367 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34368 * target.def (noce_conversion_profitable_p): New target hook.
34369 * ifcvt.h (struct noce_if_info): New type, moved from ...
34370 * ifcvt.c (struct noce_if_info): ... here.
34371 (noce_conversion_profitable_p): Renamed to ...
34372 (default_noce_conversion_profitable_p): ... this. No longer
34374 (noce_try_store_flag_constants, noce_try_addcc,
34375 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34376 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34377 instead of noce_conversion_profitable_p.
34378 * config/i386/i386.c: Include ifcvt.h.
34379 (ix86_option_override_internal): Don't override
34380 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34381 (ix86_noce_conversion_profitable_p): New function.
34382 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34383 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34384 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34385 * doc/tm.texi: Regenerated.
34387 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34389 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34392 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34395 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34396 instructions for small multiply cores.
34398 2017-04-04 Jeff Law <law@redhat.com>
34400 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34402 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34403 that are not set by the loop over the elements.
34405 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34408 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34409 int mode, convert_modes it to mode as unsigned, otherwise use
34410 lowpart_subreg to mode rather than SImode.
34411 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34412 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34413 Use DImode instead of SImode for the shift count operand.
34414 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34417 2017-04-04 Richard Biener <rguenther@suse.de>
34419 PR middle-end/80281
34420 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34421 arithmetic done for the negate or the plus. Simplify.
34422 (A - (-B) -> A + B): Likewise.
34423 * fold-const.c (split_tree): Make sure to not negate pointers.
34425 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34427 PR rtl-optimization/60818
34428 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34429 a compare of comparisons with the thing compared if this results
34430 in a different machine mode.
34432 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34434 * alias.c (base_alias_check): Fix typo in comment.
34435 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34436 * cgraphunit.c (symbol_table::compile): Likewise.
34437 * collect2.c (maybe_run_lto_and_relink): Likewise.
34438 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34439 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34440 * config/avr/avr.c (avr_map_op_t): Likewise.
34441 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34442 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34443 * config/epiphany/epiphany.md (movcc): Likewise.
34444 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34445 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34447 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34448 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34449 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34450 * config/sh/sh.c (sh_rtx_costs): Likewise.
34451 * fold-const.c (fold_truth_andor): Likewise.
34452 * genautomata.c (collapse_flag): Likewise.
34453 * gengtype.h (struct type::u::s): Likewise.
34454 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34455 * input.c (FORMAT_AMOUNT): Likewise.
34456 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34457 (known_aggs_to_agg_replacement_list): Likewise.
34458 * ipa-inline-analysis.c: Likewise.
34459 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34460 * ipa-polymorphic-call.c
34461 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34462 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34463 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34465 * modulo-sched.c (apply_reg_moves): Likewise.
34466 * omp-expand.c (build_omp_regions_1): Likewise.
34467 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34468 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34469 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34470 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34471 * value-prof.c: Likewise.
34472 * var-tracking.c (val_reset): Likewise.
34474 2017-04-03 Richard Biener <rguenther@suse.de>
34476 PR tree-optimization/80275
34477 * fold-const.c (split_address_to_core_and_offset): Handle
34480 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34482 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34483 descriptors is at least equal to that of functions.
34485 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34487 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34489 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34492 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34493 (mov<IMOD4:mode>): New expander.
34494 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34496 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34498 PR rtl-optimization/79405
34499 * fwprop.c (propagations_left): New variable.
34500 (forward_propagate_into): Decrement it.
34501 (fwprop_init): Initialize it.
34502 (fw_prop): If the variable has reached zero, stop propagating.
34503 (fwprop_addr): Ditto.
34505 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34508 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34509 a FUNCTION_DECL, pass it as decl instead of origin to
34512 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34514 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34517 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34520 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34521 TARGET_VSX_SMALL_INTEGER.
34523 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34525 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34526 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34528 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34530 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34531 extraction from odd-numbered MSA register.
34533 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34535 PR middle-end/80173
34536 * expmed.c (store_bit_field_1): Don't attempt to create
34537 a word subreg out of hard registers wider than word if they
34538 have HARD_REGNO_NREGS of 1 for their mode.
34540 PR middle-end/80163
34541 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34542 conversions to integer types wider than word and pointer.
34545 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34546 (rtx_equal_for_cselib_p): Pass 0 to it.
34547 * cselib.c (cselib_hasher::equal): Likewise.
34548 (rtx_equal_for_cselib_1): Add depth argument. If depth
34549 is 128, don't look up VALUE locs and punt. Increment
34550 depth in recursive calls when walking VALUE locs.
34552 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34554 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34555 (make_gcov_file_name): Use the canonical path name for generating
34557 (read_line): Fix handling of files with ascii null bytes.
34559 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34561 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34562 to initialise a vector register instead
34563 of using a const_int.
34565 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34567 PR translation/80189
34568 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34569 diagnostic messages.
34571 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34574 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34575 (dfp_diex_<mode>): Update mode of operand 1.
34576 * doc/extend.texi (dxex, dxexq): Document change to return type.
34577 (diex, diexq): Document change to argument type.
34579 2017-03-30 Martin Jambor <mjambor@suse.cz>
34582 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34583 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34584 it reflects the signature changes performed at the callee side.
34585 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34586 to cgraph_build_function_type_skip_args.
34587 (build_function_decl_skip_args): Adjust call to the above function.
34589 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34592 * config/i386/sse.md
34593 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34594 register as dest whenever it is a MEM not rtx_equal_p to the
34595 corresponding dup operand, and when forcing into reg move the
34596 reg into the memory afterwards.
34597 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34598 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34599 for the force_reg mode.
34600 (avx512vl_vextractf128<mode>): Use register as dest either
34601 always when a MEM, or when it is a MEM not rtx_equal_p to the
34602 corresponding dup operand, or even not when it is a CONST_VECTOR
34603 depending on the mode and lo vs. hi.
34604 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34606 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34607 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34608 Likewise. Require that operands[2] is even.
34609 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34610 Remove extraneous parens. Require that operands[2] is a multiple
34612 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34613 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34614 disallow memory then.
34616 2017-03-30 Richard Biener <rguenther@suse.de>
34618 PR tree-optimization/77498
34619 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34620 to non-constants over backedges.
34622 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34624 PR rtl-optimization/80233
34625 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34626 as last_combined_insn. Do not test for BARRIER_P separately.
34628 2017-03-29 Andreas Schwab <schwab@suse.de>
34631 * calls.c (prepare_call_address): Convert funexp to Pmode before
34632 copying to temp reg.
34634 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34636 PR tree-optimization/80158
34637 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34638 Handle possible future case of more than one alternate
34640 (replace_rhs_if_not_dup): Likewise.
34641 (replace_one_candidate): Likewise.
34643 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34645 PR rtl-optimization/80193
34646 * ira.c (ira): Do not check allocation for LRA.
34648 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34650 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34651 (nvptx_output_simt_exit): Declare.
34652 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34653 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34654 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34655 (nvptx_declare_function_name): ...here. Emit declaration of local
34656 memory space buffer for omp_simt_enter insn.
34657 (nvptx_output_unisimt_switch): New.
34658 (nvptx_output_softstack_switch): New.
34659 (nvptx_output_simt_enter): New.
34660 (nvptx_output_simt_exit): New.
34661 * config/nvptx/nvptx.h (struct machine_function): New fields
34662 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34663 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34664 (UNSPECV_SIMT_EXIT): Ditto.
34665 (omp_simt_enter_insn): New insn.
34666 (omp_simt_enter): New expansion.
34667 (omp_simt_exit): New insn.
34668 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34670 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34671 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34672 (expand_GOMP_SIMT_EXIT): New.
34673 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34674 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34675 (GOMP_SIMT_EXIT): Ditto.
34676 * target-insns.def (omp_simt_enter): New insn.
34677 (omp_simt_exit): Ditto.
34678 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34680 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34681 (lower_rec_input_clauses): Likewise.
34682 (lower_lastprivate_clauses): Handle SIMT privatization.
34684 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34685 (ompdevlow_adjust_simt_enter): New.
34686 (find_simtpriv_var_op): New.
34687 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34688 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34690 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34691 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34692 (copy_decl_for_dup_finish): Ditto.
34694 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34696 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34699 * config/i386/i386.c (ix86_option_override_internal): Always
34700 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34702 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34704 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34706 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34708 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34709 mark new edge's irreducible flag accordign to it.
34710 (vect_do_peeling): Check loop preheader edge's irreducible flag
34711 and pass it to function slpeel_add_loop_guard.
34713 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34715 PR tree-optimization/80218
34716 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34717 Update block frequencies and counts.
34719 2017-03-28 Richard Biener <rguenther@suse.de>
34721 PR tree-optimization/78644
34722 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34723 of a simplification result we may not use it at all.
34725 2017-03-28 Richard Biener <rguenther@suse.de>
34728 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34729 without arguments, generate default definition of a SSA name.
34731 2017-03-28 Richard Biener <rguenther@suse.de>
34733 PR middle-end/80222
34734 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34735 TYPE_REF_CAN_ALIAS_ALL references.
34736 * fold-const.c (fold_indirect_ref_1): Likewise.
34738 2017-03-28 Martin Liska <mliska@suse.cz>
34741 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34742 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34744 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34745 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34747 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34748 (EXTRA_SPECS): Define.
34749 (SUBTARGET_EXTRA_SPECS): Likewise.
34750 (SUBTARGET_CPP_SPEC): Likewise.
34751 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34752 SUBTARGET_EXTRA_SPECS.
34753 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34755 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34757 * config/arc/simdext.md (vst64_insn): Update pattern.
34758 (vld32wh_insn): Likewise.
34759 (vld32wl_insn): Likewise.
34760 (vld64_insn): Likewise.
34761 (vld32_insn): Likewise.
34763 2017-03-28 Marek Polacek <polacek@redhat.com>
34766 * fold-const.c (fold_comparison): Use protected_set_expr_location
34767 instead of SET_EXPR_LOCATION.
34769 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34771 * tree.c (add_expr): Avoid name lookup warning.
34773 2017-03-27 Jeff Law <law@redhat.com>
34775 PR tree-optimization/80216
34776 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34777 function name. Limit recursion depth.
34778 (record_temporary_equivalences): Corresponding changes.
34780 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34782 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34785 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34788 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34790 * cfgcleanup.c (reg_note_cfa_p): New array.
34791 (insns_have_identical_cfa_notes): New function.
34792 (old_insns_match_p): Don't cross-jump in between /f
34793 and non-/f instructions. If both i1 and i2 are frame related,
34794 verify all CFA notes, their order and content.
34796 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34799 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34800 HImode and SImode with zero extend to DImode to one insn.
34801 (bswap<mode>2_extenddi): Likewise.
34802 (bswapsi2_extenddi): Likewise.
34803 (bswaphi2_extendsi): Likewise.
34804 (bswaphi2): Combine bswap HImode and SImode into one insn.
34805 Separate memory insns from swapping register.
34806 (bswapsi2): Likewise.
34807 (bswap<mode>2): Likewise.
34808 (bswaphi2_internal): Delete, no longer used.
34809 (bswapsi2_internal): Likewise.
34810 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34811 store, and gpr<-gpr swap insns.
34812 (bswap<mode>2_store): Likewise.
34813 (bswaphi2_reg): Register only splitter, combine with the splitter.
34814 (bswaphi2 splitter): Likewise.
34815 (bswapsi2_reg): Likewise.
34816 (bswapsi2 splitter): Likewise.
34817 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34818 the insns into load, store, and register/register insns.
34819 (bswapdi2_ldbrx): Likewise.
34820 (bswapdi2_load): Likewise.
34821 (bswapdi2_store): Likewise.
34822 (bswapdi2_reg): Likewise.
34824 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34826 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34827 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34829 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34832 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34834 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34835 special handling for target option conflicts between dform
34836 options (-mpower9-dform, -mpower9-dform-vector,
34837 -mpower9-dform-scalar) and -mno-direct-move.
34839 2017-03-27 Richard Biener <rguenther@suse.de>
34841 PR tree-optimization/80181
34842 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34844 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34846 * config/arc/predicates.md (move_double_src_operand): Replace the
34847 call to move_double_src_operand with a call to address_operand.
34849 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34851 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34852 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34853 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34855 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34857 * config/arc/predicates.md (long_immediate_loadstore_operand):
34858 Consider scaled addresses cases.
34860 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34862 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34863 restored when in interrupt.
34864 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34865 doesn't have delay slot.
34867 2017-03-27 Richard Biener <rguenther@suse.de>
34870 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34871 inlined thunk clones.
34873 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34876 * asan.c (instrument_derefs): Copy over last operand from
34877 original COMPONENT_REF to the new COMPONENT_REF with
34878 DECL_BIT_FIELD_REPRESENTATIVE.
34879 * ubsan.c (instrument_object_size): Likewise.
34881 2017-03-27 Richard Biener <rguenther@suse.de>
34883 PR tree-optimization/80170
34884 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34885 sure DR/SCEV didnt fold in constants we do not see when looking
34886 at the reference base alignment.
34888 2017-03-27 Richard Biener <rguenther@suse.de>
34890 PR middle-end/80171
34891 * gimple-fold.c (fold_ctor_reference): Properly guard against
34892 NULL return value from canonicalize_constructor_val.
34894 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34897 * config/i386/i386.c (ix86_expand_builtin)
34898 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34899 flags reg setting and flags reg using instructions.
34900 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34901 clobbering instructions to zero extend op2.
34903 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34905 * doc/install.texi (Configuration) <--with-aix-soname>:
34906 Update link to AIX ld.
34908 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34910 PR rtl-optimization/80160
34911 PR rtl-optimization/80159
34912 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34913 reg_alternate_class into account.
34915 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34918 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34919 to consider in curr_insn_transform.
34921 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34923 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34924 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34925 and emit_mode_inner.
34927 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34929 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34930 argument to the overloaded builtin variants. Use the new flag to
34931 deprecate certain builtin variants.
34932 * config/s390/s390-builtin-types.def: Add new builtin types.
34933 * config/s390/s390-builtins.h: Support new flags field for
34934 overloaded builtins.
34935 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34936 (s390_macro_to_expand): Enable vector float data type.
34937 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34938 builtins by incrementing the __VEC__ version number.
34939 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34941 (s390_resolve_overloaded_builtin): Emit error messages depending
34942 on the builtin flags.
34943 * config/s390/s390.c (s390_expand_builtin): Support additional
34944 flags argument. Change error message to match the messages
34945 emitted in s390-c.c.
34946 * config/s390/s390.md: New UNSPEC_* constants.
34947 (op_type): Add new instruction types.
34948 * config/s390/vecintrin.h: Add new builtins and test data class
34950 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34951 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34952 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34953 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34954 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34955 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34957 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34958 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34959 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34960 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34962 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34963 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34964 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34965 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34967 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34968 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34969 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34970 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34971 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34972 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34973 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34975 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34976 ("vec_scatter_element<V_HW_4:mode>_DI")
34977 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34978 ("vec_fpint<mode>", "vflls")
34979 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34980 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34981 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34982 ("*vec_cmphe<mode>_cc"): ... these.
34984 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34985 mode constant instead of magic value.
34987 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34989 * config/s390/s390.c (s390_expand_vec_compare): Support other
34990 vector floating point modes than just V2DF.
34991 (s390_expand_vcond): Likewise.
34992 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34993 (s390_cannot_change_mode_class): Prevent mode changes between TF
34994 and V1TF in vector registers.
34995 * config/s390/s390.md (DF, SF): New mode attributes.
34996 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34997 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34998 SFmode support for VRs.
34999 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35001 (VFT, VF_HW): New mode iterators.
35002 (vw, sdx): New mode attributes.
35003 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35004 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35005 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35006 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35007 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35008 also the new vector floating point modes. Renaming to ...
35010 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35011 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35012 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35013 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35014 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35015 ("vec_unordered<mode>"): ... these.
35017 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35018 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35019 ("*vec_extendv2df"): New insn definitions.
35021 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35023 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35024 ("mulditi3_2", "*muldi3_sign"): New patterns.
35025 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35026 rename the pattern definition.
35028 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35030 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35032 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35034 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35036 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35037 instruction if possible.
35038 * config/s390/vector.md (vec_halfnumelts): New mode
35040 ("*vec_vllezlf<mode>"): New pattern.
35042 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35044 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35045 ("popcountv4si2", "popcountv2di2"): Rename to ...
35046 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35047 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35049 ("popcount<mode>2_vxe"): New pattern.
35051 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35053 * common/config/s390/s390-common.c (processor_flags_table): Add
35055 * config.gcc: Add arch12.
35056 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35057 Default to arch12 for unknown CPU model numbers.
35058 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35059 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35060 PROCESSOR_max sanity check.
35061 * config/s390/s390-opts.h (enum processor_type): Add
35063 * config/s390/s390.c (processor_table): Add arch12.
35064 (s390_expand_builtin): Add check for B_VXE flag.
35065 (s390_issue_rate): Add PROCESSOR_ARCH12.
35066 (s390_get_sched_attrmask): Likewise.
35067 (s390_get_unit_mask): Likewise.
35068 (s390_sched_score): Enable z13 scheduling for arch12.
35069 (s390_sched_reorder): Likewise.
35070 (s390_sched_variable_issue): Likewise.
35071 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35073 (s390_tune_attr): Use z13 scheduling also for arch12.
35074 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35075 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35076 (TARGET_VXE_P): New macros.
35077 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35078 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35079 * config/s390/s390.opt: Add arch12 as processor_type.
35081 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35083 * config/s390/s390.md
35084 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35085 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35086 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35088 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35089 Rename expanders to ...
35091 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35092 ("fixuns_truncdddi2_emu"): ... these.
35094 ("fixuns_trunc<mode>si2_emu"): New expander.
35096 ("*fixuns_truncdfdi2_z13"): Rename to ...
35097 ("*fixuns_truncdfdi2_vx"): ... this.
35099 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35101 * config/s390/2964.md: Remove the single element vector compare
35102 instructions which are no longer used.
35103 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35105 (s390_canonicalize_comparison): Remove handling of DFmode
35107 (s390_expand_vec_compare_scalar): Remove function.
35108 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35109 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35111 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35113 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35115 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35117 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35118 will anyway by matched by mov<mode>_64dfp.
35120 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35122 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35123 vlef/vstef. Add missing operand to vleif.
35125 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35127 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35128 pair for all vector types with 64 bit elements.
35129 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35130 * config/s390/vector.md (V_HW_64): ... here.
35131 (V_128_NOSINGLE): New mode iterator.
35132 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35133 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35134 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35135 ("*vec_load_pairv2di"): Change to ...
35136 ("*vec_load_pair<mode>"): ... this one.
35138 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35140 * config/s390/constraints.md: Add comments.
35141 (jKK): Reject element sizes > 8 bytes.
35142 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35144 * config/s390/s390.md: Add the s_operand checks formerly in
35145 s390_split_ok_p to various splitters where they are still
35147 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35148 for 128 bit vectors. Plus two splitters.
35150 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35152 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35155 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35158 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35159 error if the boundary argument is not constant.
35161 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35163 PR rtl-optimization/80112
35164 * loop-doloop.c (doloop_condition_get): Don't check condition
35165 if cmp isn't SET with IF_THEN_ELSE src.
35167 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35169 PR tree-optimization/80158
35170 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35171 replacing a candidate statement, also replace it for the
35172 candidate's alternate interpretation.
35173 (replace_rhs_if_not_dup): Likewise.
35174 (replace_one_candidate): Likewise.
35176 2017-03-24 Richard Biener <rguenther@suse.de>
35178 PR tree-optimization/80167
35179 * graphite-isl-ast-to-gimple.c
35180 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35182 (translate_isl_ast_to_gimple::get_rename): Likewise.
35184 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35186 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35187 handling of certain combinations of target options, including the
35188 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35189 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35191 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35194 * config/arm/arm.md (*load_multiple): Add reload_completed to
35195 matching condition.
35197 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35198 Richard Biener <rguenth@suse.de>
35200 PR tree-optimization/79908
35201 PR tree-optimization/80136
35202 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35203 been cast away, gimplify_and_add suffices.
35205 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35207 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35209 2017-03-23 Richard Biener <rguenther@suse.de>
35211 PR tree-optimization/80032
35212 * gimplify.c (gimple_push_cleanup): Forced unconditional
35213 cleanups still have to go to the conditional_cleanups
35216 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35218 PR tree-optimization/80072
35219 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35221 (next_operand_entry_id): Change type to unsigned int.
35222 (sort_by_operand_rank): Make sure to return the right return value
35223 even if unsigned fields are bigger than INT_MAX.
35224 (struct oecount): Change cnt and id type to unsigned int.
35225 (oecount_hasher::equal): Formatting fix.
35226 (oecount_cmp): Make sure to return the right return value
35227 even if unsigned fields are bigger than INT_MAX.
35228 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35231 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35232 TREE_READONLY on result if writing it more than once.
35235 * doc/invoke.texi (-fsanitize=thread): Document that with
35236 -fnon-call-exceptions atomics are not able to throw
35240 * tsan.c: Include tree-eh.h.
35241 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35242 maybe_clean_or_replace_eh_stmt where needed.
35243 (instrument_memory_accesses): Add cfg_changed argument.
35244 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35245 if it returned true.
35246 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35248 PR rtl-optimization/63191
35249 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35250 wrapper function, moved the whole old content into ...
35251 (ix86_delegitimize_address_1): ... this. New inline function.
35252 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35253 true as last argument instead of ix86_delegitimize_address.
35255 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35257 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35258 cortexa57_branch_cost.
35260 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35262 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35264 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35267 * doc/md.texi (Constraints): Document wA constraint.
35268 * config/rs6000/constraints.md (wA): New.
35269 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35270 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35271 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35272 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35274 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35277 * gimplify.c (is_oacc_declared): New function.
35278 (oacc_default_clause): Use it to set default flags for acc declared
35279 variables inside parallel regions.
35280 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35281 declared variables.
35282 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35283 declare attribute to any decl as necessary.
35285 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35288 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35289 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35290 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35291 (arm_arch_lpae): This.
35292 * config/arm/arm.c (arm_arch7ve): Rename into ...
35293 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35294 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35297 2017-03-22 Martin Liska <mliska@suse.cz>
35300 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35301 error message instead of an ICE.
35303 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35305 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35307 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35309 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35311 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35314 2017-03-21 Martin Sebor <msebor@redhat.com>
35316 * doc/extend.texi: Use "cannot" instead of "can't."
35317 * doc/hostconfig.texi: Same.
35318 * doc/install.texi: Same.
35319 * doc/invoke.texi: Same.
35320 * doc/loop.texi: Same.
35321 * doc/md.texi: Same.
35322 * doc/objc.texi: Same.
35323 * doc/rtl.texi: Same.
35324 * doc/tm.texi: Same.
35325 * doc/tm.texi.in: Same.
35326 * doc/trouble.texi: Same.
35328 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35331 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35332 (collect_checksum_attributes): Set it.
35333 (die_checksum_ordered): Use it.
35335 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35337 PR tree-optimization/79908
35338 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35339 change: For a VA_ARG whose LHS has been cast away, use
35340 force_gimple_operand to construct the side effects.
35342 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35344 PR translation/80001
35345 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35346 more amenable to translation.
35347 (oacc_loop_auto_partitions): Likewise.
35349 2017-03-21 Marek Polacek <polacek@redhat.com>
35350 Martin Sebor <msebor@redhat.com>
35352 PR tree-optimization/80109
35353 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35354 on INTEGRAL_TYPE_P.
35356 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35357 Segher Boessenkool <segher@kernel.crashing.org>
35360 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35361 check reg_used_between_p between insn and one of succ or succ2
35362 depending on if succ is artificial insn not inserted into insn
35365 2017-03-21 Martin Liska <mliska@suse.cz>
35367 PR gcov-profile/80081
35368 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35369 * doc/gcc.texi: Include gcov-dump stuff.
35370 * doc/gcov-dump.texi: New file.
35372 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35374 PR rtl-optimization/79150
35375 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35376 conditional jump, if the jump is the last insn of the loop.
35378 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35379 Richard Biener <rguenth@suse.de>
35381 PR tree-optimization/79908
35382 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35383 been cast away, use force_gimple_operand to construct the side
35386 2017-03-21 Martin Liska <mliska@suse.cz>
35388 PR libfortran/79956
35389 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35392 2017-03-21 Brad Spengler <spender@grsecurity.net>
35395 * plugin.c (htab_hash_plugin): New function.
35396 (add_new_plugin): Use it and adjust.
35397 (parse_plugin_arg_opt): Adjust.
35398 (init_one_plugin): Likewise.
35400 2017-03-21 Richard Biener <rguenther@suse.de>
35402 PR tree-optimization/80032
35403 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35404 if set force the cleanup to happen unconditionally.
35405 (gimplify_target_expr): Push inserted clobbers with force_uncond
35406 to avoid them being removed by control-dependent DCE.
35408 2017-03-21 Richard Biener <rguenther@suse.de>
35410 PR tree-optimization/80122
35411 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35412 va_arg_pack_len when the inlined call stmt requires pack
35414 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35416 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35419 * tsan.c (instrument_builtin_call): If the memory model argument
35420 is not a constant, assume it is valid.
35423 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35426 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35428 PR rtl-optimization/79910
35429 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35430 if its dest is used by an insn before I2 (other than the combined
35431 insns themselves, which are properly handled already).
35433 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35436 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35438 * combine.c (record_used_regs): New static function.
35439 (try_combine): Handle situations where there is an additional
35440 instruction between I2 and I3 which needs to have a LOG_LINK
35444 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35446 * combine.c (try_combine): Delete redundant i1 test. Call
35447 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35449 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35452 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35453 alternatives 13/14.
35455 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35457 PR tree-optimization/80054
35458 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35459 the optimization if a PHI or any of its arguments is not dominated
35460 by the candidate's basis. Use gphi* rather than gimple* as
35462 (replace_profitable_candidates): Clean up a gimple* variable that
35463 should be a gphi* variable.
35465 2017-03-20 Martin Sebor <msebor@redhat.com>
35468 * doc/extend.texi (attribute constructor): Document present limitation.
35470 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35473 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35474 __POWER9_VECTOR__ #ifdef control, change template definition to
35475 use Power9-specific built-in function.
35476 (vec_any_eq): Likewise.
35477 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35478 to control outcomes from this test.
35479 (vector_ae_<mode>p): For VEC_F modes, likewise.
35481 2017-03-20 Ian Lance Taylor <iant@google.com>
35483 * config/i386/i386.c (ix86_function_regparm): Save an extra
35484 register for -fsplit-stack with DECL_STATIC_CHAIN.
35486 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35489 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35490 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35492 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35494 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35496 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35499 2017-03-20 Marek Polacek <polacek@redhat.com>
35502 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35504 2017-03-20 Richard Biener <rguenther@suse.de>
35506 PR tree-optimization/80113
35507 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35508 allocate extra SSA name for PHI def.
35509 (add_close_phis_to_outer_loops): Likewise.
35510 (add_close_phis_to_merge_points): Likewise.
35511 (copy_loop_close_phi_args): Likewise.
35512 (copy_cond_phi_nodes): Likewise.
35514 2017-03-20 Martin Liska <mliska@suse.cz>
35516 PR middle-end/79753
35517 * tree-chkp.c (chkp_build_returned_bound): Do not build
35518 returned bounds for a LHS that's not a BOUNDED_P type.
35520 2017-03-20 Martin Liska <mliska@suse.cz>
35524 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35525 COMPLEX_CST and VECTOR_CST.
35527 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35530 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35531 target operand. A new splitter adds the clobber statement in case
35532 the target operand is dead anyway.
35534 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35536 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35537 to age-old versions of binutils and glibc.
35539 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35541 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35543 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35545 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35547 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35549 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35550 requirement for binutils 2.13.
35552 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35554 * combine.c (try_combine): Delete redundant i1 test. Call
35555 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35557 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35559 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35560 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35562 <riscv64-*-elf>: Re-arrange section
35563 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35564 <riscv32-*-linux>: Likewise.
35565 <riscv64-*-elf>: Likewise
35566 <riscv64-*-linux>: Likewise.
35568 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35571 * aarch64.opt(verbose-cost-dump): Fix typo.
35573 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35576 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35577 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35579 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35581 * reload.c (find_reloads): When reloading a nonoffsettable address,
35582 use RELOAD_OTHER for it and its address reloads.
35584 PR rtl-optimization/79910
35585 * combine.c (record_used_regs): New static function.
35586 (try_combine): Handle situations where there is an additional
35587 instruction between I2 and I3 which needs to have a LOG_LINK
35590 2017-03-17 Jeff Law <law@redhat.com>
35592 PR tree-optimization/71437
35593 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35594 conditional in the hash table first.
35595 (vrp_dom_walker::before_dom_children): Extract condition from
35596 ASSERT_EXPR. Record condition, its inverion and any implied
35597 conditions as well.
35599 2017-03-17 Marek Polacek <polacek@redhat.com>
35600 Markus Trippelsdorf <markus@trippelsdorf.de>
35602 PR tree-optimization/80079
35603 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35606 2017-03-17 Richard Biener <rguenther@suse.de>
35608 PR middle-end/80075
35609 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35610 Properly verify the LHS before the RHS possibly claims to be
35612 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35615 2017-03-17 Martin Jambor <mjambor@suse.cz>
35617 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35618 (List of -O2 options): Likewise.
35619 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35622 2017-03-17 Tom de Vries <tom@codesourcery.com>
35624 * gcov-dump.c (print_usage): Print bug_report_url.
35626 2017-03-17 Richard Biener <rguenther@suse.de>
35628 PR middle-end/80050
35629 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35630 (parser::peek): Likewise.
35632 2017-03-17 Richard Biener <rguenther@suse.de>
35634 PR tree-optimization/80048
35635 * sese.c (free_sese_info): Properly release rename_map and
35636 copied_bb_map elements.
35638 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35640 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35641 Add linked-list forward and backlinks. Insert on
35642 construction, remove on destruction.
35643 (class pass_store_merging): Add m_stores_head field.
35644 (pass_store_merging::terminate_and_process_all_chains):
35645 Iterate over m_stores_head list.
35646 (pass_store_merging::terminate_all_aliasing_chains):
35648 (pass_store_merging::execute): Check for debug stmts first.
35649 Push new chains onto the m_stores_head stack.
35651 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35654 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35655 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35656 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35658 2017-03-16 Jeff Law <law@redhat.com>
35660 PR tree-optimization/71437
35661 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35662 member function. Implementation moved into after_dom_children
35663 member function and into the threader's thread_outgoing_edges
35665 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35666 some code into new thread_outgoing_edges.
35667 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35668 definition. Simplify marker handling (do it here). Assume we always
35669 have the available expression and the const/copies tables.
35670 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35672 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35673 * tree-vrp.c (equiv_stack): No longer file scoped.
35674 (vrp_dom_walker): New class.
35675 (vrp_dom_walker::before_dom_children): New member function.
35676 (vrp_dom_walker::after_dom_children): Likewise.
35677 (identify_jump_threads): Setup domwalker. Use it rather than
35678 walking edges in a random order by hand. Simplify setup/finalization.
35679 (finalize_jump_threads): Remove.
35680 (vrp_finalize): Do not call identify_jump_threads here.
35681 (execute_vrp): Do it here instead and call thread_through_all_blocks
35684 PR tree-optimization/71437
35685 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35687 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35689 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35690 (dom_opt_dom_walker::thread_across_edge): Remove
35691 handle_dominating_asserts argument. All callers changed.
35692 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35693 changes. Remove calls to lhs_of_dominating_assert. Other
35694 uses of handle_dominating_asserts turn into unconditional code
35695 (simplify_control_stmt_condition_1): Likewise.
35696 (simplify_control_stmt_condition): Likewise.
35697 (thread_through_normal_block, thread_across_edge): Likewise.
35698 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35699 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35700 object if it is not an SSA_NAME.
35701 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35702 before calling into the VRP specific simplifiers.
35703 (identify_jump_threads): Remove handle_dominating_asserts
35706 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35709 * tree-diagnostic.c (default_tree_printer): No longer static.
35710 * tree-diagnostic.h (default_tree_printer): New prototype.
35712 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35714 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35715 Change ins into fmov.
35717 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35719 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35720 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35721 Use h_con constraint for operand 1.
35722 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35723 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35725 2017-03-15 Jeff Law <law@redhat.com>
35727 PR tree-optimization/71437
35728 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35729 (record_temporary_equivalences): Use it.
35731 PR tree-optimization/71437
35732 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35733 tree-ssa-scopedtables.
35734 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35735 (record_conditions, record_cond, vuse_eq): Likewise.
35736 (record_edge_info): Adjust to API tweak of record_conditions.
35737 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35738 (record_temporary_equivalences, optimize_stmt): Likewise.
35739 (eliminate_redundant_computations): Likewise.
35740 (record_equivalences_from_stmt): Likewise.
35741 * tree-ssa-scopedtables.c: Include options.h and params.h.
35742 (vuse_eq): New function, moved from tree-ssa-dom.c
35743 (build_and_record_new_cond): Likewise.
35744 (record_conditions): Likewise. Accept vector of conditions rather
35745 than edge_equivalence structure for first argument.
35746 for the first argument.
35747 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35748 from tree-ssa-dom.c.
35749 (avail_exprs_stack::record_cond): Likewise.
35750 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35751 from tree-ssa-dom.c.
35752 (avail_exprs_stack): Add new member functions lookup_avail_expr
35754 (record_conditions): Declare.
35756 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35759 * lra-constraints.c (process_alt_operands): Increase reject for
35760 reloading an input/output operand.
35762 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35765 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35766 insns to convert from signed/unsigned char/short to IEEE 128-bit
35768 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35770 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35773 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35774 subreg of inner mode for values already in registers.
35776 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35778 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35779 iteration reg is used after the loop.
35781 2017-03-14 Martin Sebor <msebor@redhat.com>
35783 PR tree-optimization/79800
35784 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35785 precision in negative-positive range.
35786 (format_floating): Call non-const overload with adjusted precision.
35788 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35791 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35794 2017-03-14 Martin Sebor <msebor@redhat.com>
35796 PR middle-end/80020
35797 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35798 * builtins.def (aligned_alloc): Use it.
35801 * Makefile.in (GTFILES): Add calls.c.
35802 * calls.c: Include "gt-calls.h".
35804 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35806 PR rtl-optimization/79728
35807 * regs.h (struct target_regs): New field
35808 x_contains_allocatable_regs_of_mode.
35809 (contains_allocatable_regs_of_mode): New macro.
35810 * reginfo.c (init_reg_sets_1): Initialize it, and change
35811 contains_reg_of_mode so it includes global regs as well.
35812 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35813 rather than contains_regs_of_mode.
35815 2017-03-14 Martin Liska <mliska@suse.cz>
35817 * doc/invoke.texi: Document options that can't be combined with
35818 -fcheck-pointer-bounds.
35820 2017-03-14 Martin Liska <mliska@suse.cz>
35822 PR middle-end/79831
35823 * doc/invoke.texi (-Wchkp): Document the option.
35825 2017-03-14 Martin Liska <mliska@suse.cz>
35827 * Makefile.in: Install gcov-dump.
35829 2017-03-14 Martin Liska <mliska@suse.cz>
35831 * multiple_target.c (expand_target_clones): Bail out for
35832 an invalid attribute.
35834 2017-03-14 Richard Biener <rguenther@suse.de>
35836 * alias.c (struct alias_set_entry): Pack properly.
35837 * cfgloop.h (struct loop): Likewise.
35838 * cse.c (struct set): Likewise.
35839 * ipa-utils.c (struct searchc_env): Likewise.
35840 * loop-invariant.c (struct invariant): Likewise.
35841 * lra-remat.c (struct cand): Likewise.
35842 * recog.c (struct change_t): Likewise.
35843 * rtl.h (struct address_info): Likewise.
35844 * symbol-summary.h (function_summary): Likewise.
35845 * tree-loop-distribution.c (struct partition): Likewise.
35846 * tree-object-size.c (struct object_size_info): Likewise.
35847 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35848 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35849 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35850 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35851 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35852 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35853 (struct _stmt_vec_info): Likewise.
35855 2017-03-14 Martin Liska <mliska@suse.cz>
35858 * multiple_target.c (create_dispatcher_calls): Check that
35859 a target can create a function dispatcher.
35861 2017-03-14 Martin Liska <mliska@suse.cz>
35864 * multiple_target.c (expand_target_clones): Drop local.local
35865 flag for default implementation.
35867 2017-03-14 Richard Biener <rguenther@suse.de>
35869 PR tree-optimization/80030
35870 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35872 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35874 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35875 gcc_fallthrough() instead of __attribute__((fallthrough));
35877 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35879 * doc/gcc.texi: Remove "up" link to (DIR).
35880 * doc/gccint.texi: Ditto.
35882 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35884 * doc/install.texi (Specific) <avr>: Remove reference to
35887 2017-03-13 Jeff Law <law@redhat.com>
35889 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35890 attribute rather than comments.
35892 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35893 match_scratch operand is highest.
35895 2017-03-13 Martin Liska <mliska@suse.cz>
35897 PR middle-end/78339
35898 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35899 is a CHKP clone, use original declaration.
35901 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35903 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35904 (arc_conditional_register_usage): Use a different allocation order
35905 when optimizing for size.
35906 * common/config/arc/arc-common.c (arc_option_optimization_table):
35907 Section anchors default on when optimizing for size.
35909 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35911 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35913 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35915 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35916 * config/arc/arc.md (cpu_facility): Add cd variant.
35917 (*movqi_insn): Add code density variant.
35918 (*movhi_insn): Likewise.
35919 (*movqi_insn): Likewise.
35920 (*addsi3_mixed): Likewise.
35921 (subsi3_insn): Likewise.
35923 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35925 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35927 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35929 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35930 expressions with MINUS and UNARY ops.
35932 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35935 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35937 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35938 between vec_select and vector argument.
35939 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35940 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35941 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35942 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35943 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35944 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35946 2017-03-13 Richard Biener <rguenther@suse.de>
35949 * params.def (vect-max-peeling-for-alignment): Fix typo.
35951 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35953 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35954 issue that only occurred with binutils below 2.18.
35956 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35958 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35959 refer to binutils 2.11/2.12 minimum.
35961 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35963 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35964 ftp.kernel.org and simplify binutils requirement.
35966 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35968 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35970 (Optimize Options): Ditto. Also remove redundancy.
35972 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35974 PR translation/79848
35975 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35977 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35978 to G_ to avoid double translation.
35980 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35982 PR translation/79923
35983 * auto-profile.c (get_combined_location): Convert leading
35984 character of diagnostics to lower case and remove trailing period.
35985 (read_profile): Likewise for various diagnostics.
35986 * config/arm/arm.c (arm_option_override): Remove trailing period
35987 from various diagnostics.
35988 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35989 (msp430_expand_delay_cycles): Likewise.
35991 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35994 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35995 full command-line argument, rather than just "str".
35996 (aarch64_validate_march): Likewise.
35997 (aarch64_validate_mtune): Likewise.
35999 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36001 PR rtl-optimization/78911
36002 * lra-assigns.c (must_not_spill_p): New function.
36003 (spill_for): Use it.
36005 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36007 PR tree-optimization/79981
36008 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36009 ATOMIC_COMPARE_EXCHANGE ifn result.
36010 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36011 IFN_ATOMIC_COMPARE_EXCHANGE.
36013 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36016 * opts.c (parse_sanitizer_options): Add missing question mark to
36017 "did you mean" message.
36019 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36021 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36023 (VMULEUH_UNS): Likewise.
36024 (VMULOUB_UNS): Likewise.
36025 (VMULOUH_UNS): Likewise.
36026 * config/rs6000/rs6000.c (builtin_function_type): Remove
36027 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36029 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36032 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36033 x with result of extra_parsing_for_operand_code_0.
36034 (function_reader::extra_parsing_for_operand_code_0): Convert
36035 return type from void to rtx, returning x. When reading
36036 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36037 larger size containing struct block_symbol.
36039 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36041 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36042 -mfloat128-hardware without -m64.
36044 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36047 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36048 entries to the case statement that marks unsigned arguments to
36049 overloaded functions.
36051 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36053 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36054 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36056 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36059 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36060 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36062 2017-03-10 Martin Liska <mliska@suse.cz>
36066 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36067 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36070 2017-03-10 Olivier Hainque <hainque@adacore.com>
36072 * tree-switch-conversion (array_value_type): Start by resetting
36073 candidate type to it's main variant.
36075 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36077 PR rtl-optimization/79909
36078 * combine.c (try_combine): Use simplify_replace_rtx on individual
36079 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36080 of the whole CALL_INSN_FUNCTION_USAGE.
36082 PR tree-optimization/79972
36083 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36084 get_range_info on SSA_NAMEs. Formatting fixes.
36086 2017-03-10 Richard Biener <rguenther@suse.de>
36087 Jakub Jelinek <jakub@redhat.com>
36089 PR tree-optimization/77975
36090 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36091 edge to be constant.
36092 (get_val_for): For constant x return it. Formatting fix.
36093 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36094 would use the same bases as the current one.
36096 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36098 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36099 instead of vec_select for V1TImode.
36100 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36102 (VSX_LE_128): Add V1TI to this mode iterator.
36103 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36104 (*vsx_le_perm_store_<mode>): Likewise.
36105 (pre-reload splitter for VSX stores): Likewise.
36106 (post-reload splitter for VSX stores): Likewise.
36107 (*vsx_xxpermdi2_le_<mode>): Likewise.
36108 (*vsx_lxvd2x2_le_<mode>): Likewise.
36109 (*vsx_stxvd2x2_le_<mode>): Likewise.
36111 2017-03-09 Michael Eager <eager@eagercon.com>
36113 Correct failures with --enable-checking=yes,rtl.
36115 * config/microblaze/microblaze.c (microblaze_expand_shift):
36116 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36117 test for const0_rtx.
36118 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36119 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36121 2017-03-09 Richard Biener <rguenther@suse.de>
36123 PR tree-optimization/79977
36124 * graphite-scop-detection.c (scop_detection::merge_sese):
36125 Handle the case of extra exits to blocks dominating the entry.
36127 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36129 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36132 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36134 PR rtl-optimization/79949
36135 * lra-constraints.c (process_alt_operands): Check memory when
36136 trying to predict a cycle. Print about the overall increase.
36138 2017-03-09 Richard Biener <rguenther@suse.de>
36140 PR middle-end/79971
36141 * gimple-expr.c (useless_type_conversion_p): Preserve
36142 TYPE_SATURATING for fixed-point types.
36144 2017-03-09 Richard Biener <rguenther@suse.de>
36147 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36148 alignment of BLKmode params.
36150 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36153 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36154 (VALL_NO_V2Q): Likewise.
36156 * config/aarch64/aarch64-simd.md
36157 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36159 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36160 VALL_NO_V2Q mode iterator.
36161 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36163 2017-03-09 Martin Liska <mliska@suse.cz>
36165 PR tree-optimization/79631
36166 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36167 tree_int_cst_sign_bit just for INTEGER constants.
36169 2017-03-09 Martin Liska <mliska@suse.cz>
36173 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36176 2017-03-09 Marek Polacek <polacek@redhat.com>
36179 * tree.c (inchash::add_expr): Handle TREE_VEC.
36181 2017-03-09 Martin Liska <mliska@suse.cz>
36184 (chkp_narrow_size_and_offset): New function.
36185 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36186 (void chkp_parse_bit_field_ref): New function.
36187 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36188 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36190 2017-03-09 Martin Liska <mliska@suse.cz>
36193 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36194 (chkp_find_bounds_1): Remove gcc_unreachable.
36196 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36199 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36200 BUILT_IN_SYNC*, determine the access type from the size suffix and
36201 always build a MEM_REF with that type. Handle forgotten
36202 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36205 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36206 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36207 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36208 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36209 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36210 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36211 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36212 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36213 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36214 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36215 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36216 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36217 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36218 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36219 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36220 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36221 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36222 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36223 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36224 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36225 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36226 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36227 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36228 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36229 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36230 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36231 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36232 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36233 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36234 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36235 definitions outside of __OPTIMIZE__ guarded section.
36238 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36239 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36240 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36241 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36244 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36246 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36247 ("vfenez<mode>"): Add missing constraints.
36249 2017-03-08 Martin Sebor <msebor@redhat.com>
36252 * config/nds32/nds32.c (nds32_option_override):
36253 Fix misspelled diagnostic.
36255 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36258 * gimplify.c (gimplify_omp_for): Replace index var in outer
36259 taskloop statement with an artificial variable and add
36260 OMP_CLAUSE_PRIVATE clause for it.
36262 2017-03-08 Richard Biener <rguenther@suse.de>
36264 PR tree-optimization/79955
36265 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36266 for accesses that are completely outside of the variable.
36268 2017-03-08 Andrew Haley <aph@redhat.com>
36270 PR tree-optimization/79943
36271 * tree-ssa-loop-split.c (compute_new_first_bound): When
36272 calculating the new upper bound, (END-BEG) should be added, not
36275 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36277 * config/avr/avr.md (setmemhi): Make sure match_dup
36278 operand number comes before match_scratch.
36280 2017-03-08 Richard Biener <rguenther@suse.de>
36282 PR tree-optimization/79920
36283 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36284 with ncopies == 1 to ...
36285 (vect_transform_slp_perm_load): ... here. Properly compute
36286 all element loads by iterating VF times over the group. Do
36287 not handle ncopies (computed in a broken way) in
36288 vect_create_mask_and_perm.
36290 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36293 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36294 is a uniform vector, use uniform_vector_p return value instead of
36295 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36297 2017-03-07 Marek Polacek <polacek@redhat.com>
36299 PR middle-end/79809
36300 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36301 (alloca_call_type): Likewise.
36303 2017-03-07 Martin Liska <mliska@suse.cz>
36305 * gcov.c (process_args): Put comment to correct location.
36307 2017-03-07 Martin Liska <mliska@suse.cz>
36309 PR middle-end/68270
36310 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36311 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36312 (chkp_narrow_bounds_for_field): Likewise.
36313 (chkp_parse_array_and_component_ref): Pass one more argument to
36316 2017-03-07 Richard Biener <rguenther@suse.de>
36318 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36321 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36323 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36324 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36326 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36329 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36330 to end of description.
36331 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36333 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36335 PR rtl-optimization/79901
36336 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36338 (*avx512f_<code><mode>3<mask_name>): ... this.
36339 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36340 iterator instead of VI8_AVX2_AVX512BW.
36342 PR rtl-optimization/79901
36343 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36344 min/max expander, expand it using expand_vec_cond_expr.
36347 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36350 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36353 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36354 to void * for PCH reasons.
36355 * dwarf2out.c (output_loc_operands, output_die): Cast
36356 v.val_vec.array to unsigned char *.
36358 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36361 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36364 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36366 PR rtl-optimization/79571
36367 * lra-constraints.c (process_alt_operands): Calculate static
36368 reject and subtract it from overall when only addresses will be
36371 2017-03-06 Julia Koval <julia.koval@intel.com>
36374 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36375 incoming stack boundary to 128 for 64-bit targets.
36377 2017-03-06 Richard Biener <rguenther@suse.de>
36379 PR tree-optimization/79894
36380 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36381 to NULL after folding it.
36383 2017-03-06 Richard Biener <rguenther@suse.de>
36385 PR tree-optimization/79824
36386 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36387 check disabling peeling for gaps.
36389 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36391 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36392 attributes): Document gettimeofday.
36394 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36396 * config/s390/s390.c (s390_option_override_internal): Set
36397 PARAM_MIN_VECT_LOOP_BOUND
36399 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36401 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36402 * config/s390/s390.md: Likewise.
36404 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36407 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36408 (<avx2_avx512>_perm<mode>): Rename to ...
36409 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36411 (<avx512>_perm<mode>_mask): Rename to ...
36412 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36414 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36415 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36416 instead of VI8F_256_512.
36417 (avx512f_perm<mode>): New define_expand.
36418 (avx512f_perm<mode>_mask): Likewise.
36419 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36420 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36422 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36424 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36425 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36427 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36429 2017-03-06 Martin Liska <mliska@suse.cz>
36432 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36433 when having a SSA NAME w/o VAR_DECL assigned to it.
36435 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36437 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36438 msa_dpsub_<su>_d): Fix MODE for vec_select.
36440 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36442 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36444 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36446 2017-03-06 Richard Biener <rguenther@suse.de>
36448 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36449 * plugin.c (register_plugin_info): Likewise.
36450 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36452 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36454 * config/i386/sse.md (sse_storehps, sse_storelps,
36455 avx_<castmode><avxsizesuffix>_<castmode>,
36456 avx512f_<castmode><avxsizesuffix>_<castmode>,
36457 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36458 in condition that at least one operand is not a MEM.
36460 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36462 PR middle-end/79805
36463 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36464 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36466 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36467 gimple_call_nothrow_p flag based on whether original builtin can throw.
36468 If it can, emit following stmts on the fallthrough edge.
36469 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36470 don't create new bb if inserting just debug stmts on the edge, try to
36471 insert them on the fallthru bb or just reset debug stmts.
36473 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36476 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36477 restore recog_data (including the operand rtxes inside it) around
36478 the call to get_insn_template.
36480 2017-03-03 Martin Sebor <msebor@redhat.com>
36482 PR tree-optimization/79699
36483 * context.c (context::~context): Free MPFR caches to avoid
36484 a memory leak on program exit.
36486 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36488 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36489 Use wide_int::ulow () instead of .elt (0).
36491 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36493 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36494 (*pushxf): Limit oF constraint to 32bit targets and add oC
36495 constraint for 64bit targets.
36496 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36497 (*pushdf): Change rmF constraint to rmC.
36499 2017-03-03 Martin Liska <mliska@suse.cz>
36501 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36502 Remove unused variable.
36504 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36507 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36508 is a memory operand, increase num_memory.
36509 (ix86_expand_args_builtin): Likewise.
36511 2017-03-03 Jan Hubicka <jh@suse.cz>
36514 * ipa-devirt.c (maybe_record_node): Properly handle
36515 __cxa_pure_virtual visibility.
36517 2017-03-03 Martin Liska <mliska@suse.cz>
36519 PR tree-optimization/79803
36520 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36522 (pass_loop_prefetch::execute): Disabled optimization if an
36523 assumption about L1 cache size is not met.
36525 2017-03-03 Martin Liska <mliska@suse.cz>
36527 PR rtl-optimization/79574
36528 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36529 (hash_scan_set): Likewise.
36530 (dump_hash_table): Likewise.
36531 (hoist_code): Likewise.
36533 2017-03-03 Richard Biener <rguenther@suse.de>
36535 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36536 in place of uhigh ().
36537 (fixed_convert_from_real): Likewise.
36539 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36542 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36544 2017-03-03 Richard Biener <rguenther@suse.de>
36546 PR middle-end/79818
36547 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36548 TYPE_OVERFLOW_UNDEFINED check.
36550 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36552 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36554 (vector_ae_<mode>_p): Likewise.
36555 (vector_nez_<mode>_p): Likewise.
36556 (vector_ne_v2di_p): Likewise.
36557 (vector_ae_v2di_p): Likewise.
36558 (vector_ne_<mode>_p): Likewise.
36559 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36561 (vsx_tsqrt<mode>2_fe): Likewise.
36563 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36566 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36568 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36570 PR rtl-optimization/79780
36571 * cprop.c (one_cprop_pass): When second and further conditional trap
36572 in a single basic block is turned into an unconditional trap, turn it
36573 into a deleted note to avoid RTL verification failures.
36575 2017-03-02 Richard Biener <rguenther@suse.de>
36577 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36579 2017-03-02 Richard Biener <rguenther@suse.de>
36581 PR tree-optimization/79345
36583 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36584 param and abort the walk, returning -1 if it is hit.
36585 (walk_aliased_vdefs): Take a limit param and pass it on.
36586 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36587 defaulting to 0 and return a signed int.
36588 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36589 (check_defs): New helper.
36590 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36591 about uninitialized memory.
36592 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36593 bogus uninitialized warning.
36594 (fixed_convert_from_real): Likewise.
36596 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36598 PR tree-optimization/66768
36599 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36600 iv_use if base object can't be determined.
36602 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36604 PR tree-optimization/79345
36605 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36606 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36607 (get_pattern_stats): Initialize it.
36608 * genemit.c (gen_expand): Verify match_scratch numbers come after
36609 match_operand/match_dup numbers.
36610 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36611 match_scratch numbers.
36612 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36614 * config/s390/s390.md (trunctdsd2): Likewise.
36616 2017-03-02 Richard Biener <rguenther@suse.de>
36618 * wide-int.h (wide_int_storage::operator=): Implement in terms
36621 2017-03-02 Richard Biener <rguenther@suse.de>
36623 PR tree-optimization/79777
36624 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36625 the to insert expression to sth existing.
36627 2017-03-01 Martin Sebor <msebor@redhat.com>
36629 PR middle-end/79692
36630 * gimple-ssa-sprintf.c
36631 (directive::known_width_and_precision): New function.
36632 (format_integer): Use it.
36633 (get_mpfr_format_length): Consider the full range of precision
36634 when computing %g output with the # flag. Set the likely byte
36635 count to 3 rather than 1 when precision is indeterminate.
36636 (format_floating): Correct the lower bound of precision.
36638 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36640 * doc/invoke.texi: Document default code model for 64-bit Linux.
36642 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36645 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36646 udiv rather than div since input pattern is unsigned.
36648 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36650 * config/i386/i386.c (print_reg): Warn for values of
36651 unsupported size in integer register.
36653 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36656 * config/rs6000/predicates.md (current_file_function_operand): Do
36657 not allow self calls to be local if the function is replaceable.
36659 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36662 * config/rs6000/altivec.h (vec_ctz and others): Change the
36663 preprocessor macro that controls conditional compilation from
36664 _ARCH_PWR9 to __POWER9_VECTOR__.
36665 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36666 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36667 control (instead of _ARCH_PWR9 control) so that template
36668 definition uses power9-specific function.
36669 (vec_any_eq): Likewise.
36670 (vec_all_ne): Change macro definition to use a power9-specific
36671 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36672 _ARCH_PWR9 control).
36673 (vec_any_eq) Likewise.
36674 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36675 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36676 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36677 support for xvcmpnedp instruction.
36678 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36679 macro expansion so that Power9 implementation of vec_all_ne does
36680 not use the AltiVec predicate framework.
36681 (VCMPNEH_P): Likewise.
36682 (VCMPNEW_P): Likewise.
36683 (VCMPNED_P): Likewise.
36684 (VCMPNEFP_P): Likewise.
36685 (VCMPNEDP_P): Likewise.
36686 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36687 implementation of vec_any_eq to not use AltiVec predicate
36689 (VCMPAEH_P): Likewise.
36690 (VCMPAEW_P): Likewise.
36691 (VCMPAED_P): Likewise.
36692 (VCMPAEFP_P): Likewise.
36693 (VCMPAEDP_P): Likewise.
36694 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36695 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36696 not use the AltiVec predicate framework.
36697 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36698 of vec_any_eq to not use AltiVec predicate framework.
36699 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36700 support for predefined __POWER9_VECTOR__ macro to indicate that
36701 Power9 instruction selection is enabled.
36702 (altivec_overloaded_builtins): Remove extraneous
36703 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36704 function argument types RS6000_BTI_bool_V16QI and
36705 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36706 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36707 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36708 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36709 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36710 Power9 for implementations of vec_cmpne. Change the signature for
36711 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36712 (representing vec_all_ne) to remove the previously described first
36713 argument of type RS6000_BTI_INTSI, as this was an artifact of
36714 reliance on the AltiVec predicate framework, which is no longer
36715 used in the implementation of these functions. Add
36716 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36717 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36718 since, unlike the AltiVec predicate framework implementation, we
36719 do not share function descriptors between vec_alle and vec_anyeq.
36720 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36721 set of modes that receive special treatment even when
36722 TARGET_P9_VECTOR is true. The special treatment emits code that
36723 does not depend on Power9 instructions.
36724 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36725 define_expand to not rely on AltiVec predicate framework.
36726 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36728 (vector_ne_v2di_p): Change this define_expand to not rely on
36729 AltiVec predicate framework.
36730 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36732 (vector_ne_<mode>_p): Change this define_expand to not rely on
36733 AltiVec predicate framework.
36734 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36736 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36737 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36738 define_insn pattern.
36739 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36740 define_insn pattern because the xvcmpne<VSs>. instruction is not
36742 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36743 instruction is not supported.
36745 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36747 * config/nvptx/nvptx.c: Include intl.h.
36749 2017-03-01 Martin Jambor <mjambor@suse.cz>
36752 * ipa-prop.h (ipa_bits): Removed field known.
36753 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36754 to pointers. Adjusted their comments to warn about their sharing.
36755 (ipcp_transformation_summary): Change bits to a vector of pointers.
36756 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36757 (ipa_get_ipa_bits_for_value): Declare.
36758 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36759 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36760 (ipa_bits_hash_table): Likewise.
36761 (ipa_vr_ggc_hash_traits): Likewise.
36762 (ipa_vr_hash_table): Likewise.
36763 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36764 being pointers and vr_known being removed.
36765 (ipa_set_jf_unknown): Likewise.
36766 (ipa_get_ipa_bits_for_value): New function.
36767 (ipa_set_jfunc_bits): Likewise.
36768 (ipa_get_value_range): New overloaded functions.
36769 (ipa_set_jfunc_vr): Likewise.
36770 (ipa_compute_jump_functions_for_edge): Use the above functions to
36771 construct bits and vr parts of jump functions.
36772 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36773 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36775 (ipcp_grow_transformations_if_necessary): Also allocate
36776 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36778 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36779 them. Fix too long lines.
36780 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36781 vr_known being removed.
36782 (ipa_read_jump_function): Use new setter functions to construct bits
36783 and vr parts of jump functions or set them to NULL.
36784 (write_ipcp_transformation_info): Adjust for bits being pointers.
36785 (read_ipcp_transformation_info): Likewise.
36786 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36788 Include gt-ipa-prop.h.
36789 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36791 (ipcp_store_bits_results): Likewise.
36792 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36793 Do not write to existing jump functions but use a temporary instead.
36795 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36798 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36799 attempt to use its first operand as BIT_FIELD_REF base.
36801 2017-03-01 Richard Biener <rguenther@suse.de>
36803 PR middle-end/79721
36804 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36805 interpolating formula in wrapping arithmetic.
36806 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36808 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36810 PR tree-optimization/79734
36811 * tree-vect-generic.c (expand_vector_condition): Optimize
36812 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36813 Handle VEC_COND_EXPR where comparison has different inner width from
36814 type's inner width.
36816 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36818 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36819 markup, and similar issues. Remove @opindex entries for things
36820 that aren't options. Add missing -mmpy-option entries.
36822 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36824 PR tree-optimization/79737
36825 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36826 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36827 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36828 instead of byte_size. Formatting fix.
36829 (shift_bytes_in_array_right): Formatting fix.
36831 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36834 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36835 condition on optimize for the leaf function test.
36837 2017-02-28 Martin Liska <mliska@suse.cz>
36840 * read-rtl-function.c (function_reader::handle_unknown_directive):
36841 Bail out when one uses -flto.
36843 2017-02-28 Martin Liska <mliska@suse.cz>
36845 * common.opt: Replace space with tabular for options of <number>
36847 * config/i386/i386.opt: Show <number> value for
36848 -mlarge-data-threshold.
36849 * opts.c (print_filtered_help): Do not display number in hexadecimal
36852 2017-02-28 Martin Liska <mliska@suse.cz>
36854 * common.opt: Fix --help=option -Q for options which are of
36857 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36859 * config/i386/i386.c (print_reg): Error out for values
36860 of 8-bit size in invalid integer register.
36862 2017-02-28 Martin Sebor <msebor@redhat.com>
36864 PR tree-optimization/79691
36865 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36867 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36870 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36871 gcc_unreachable with output_operand_lossage.
36873 2017-02-28 Richard Biener <rguenther@suse.de>
36875 PR tree-optimization/79740
36876 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36878 (visit_nary_op): Insert the nary into the hashtable if we
36879 pattern-matched sth.
36880 * tree-ssa-pre.c (eliminate_insert): Robustify.
36882 2017-02-28 Richard Biener <rguenther@suse.de>
36884 PR middle-end/79731
36885 * fold-const.c (decode_field_reference): Reject out-of-bound
36888 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36890 * config/i386/i386.c: Include intl.h.
36891 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36892 instead of just cond ? "..." : "...".
36893 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36894 * coverage.c (read_counts_file): Likewise.
36895 * omp-offload.c: Include intl.h.
36896 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36897 of just cond ? "..." : "...".
36898 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36899 of just cond ? "..." : "...".
36901 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36904 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36906 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36907 'tune for' CPU name.
36908 * config/arm/arm-cpu-data.h: Regenerated.
36910 2017-02-28 Richard Biener <rguenther@suse.de>
36912 PR tree-optimization/79732
36913 * tree-inline.c (expand_call_inline): Do not shadow var.
36915 2017-02-28 Richard Biener <rguenther@suse.de>
36917 PR tree-optimization/79723
36918 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36919 address-space properly.
36921 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36923 * doc/optinfo.texi (Optimization groups): Fix option used for
36925 * doc/invoke.texi (-fopt-info): Document "omp".
36926 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36927 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36928 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36929 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36930 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36932 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36934 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36935 OPTGROUP_OMP with "omp".
36937 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36940 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36941 for arithmetic shift of unsigned V2DI.
36943 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36945 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36946 arc/linux.h headers.
36947 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36948 (LINK_SPEC): Likewise.
36949 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36950 (EXTRA_SPECS): Likewise.
36951 (STARTFILE_SPEC): Likewise.
36952 (ENDFILE_SPEC): Likewise.
36953 (LIB_SPEC): Likewise.
36954 (TARGET_SDATA_DEFAULT): Likewise.
36955 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36956 (MULTILIB_DEFAULTS): Likewise.
36957 (DWARF2_UNWIND_INFO): Likewise.
36958 * config/arc/big.h: New file.
36959 * config/arc/elf.h: Likewise.
36960 * config/arc/linux.h: Likewise.
36961 * config/arc/t-uClibc: Remove.
36963 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36965 PR tree-optimization/77536
36966 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36967 (tree_transform_and_unroll_loop): Use above function to compute the
36968 estimated niter of unrolled loop and use it when scaling profile.
36969 Also use count info rather than frequency if it's non-zero.
36970 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36971 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36972 (vect_transform_loop): Call above function.
36974 2017-02-27 Richard Biener <rguenther@suse.de>
36976 PR tree-optimization/45397
36977 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36978 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36979 (visit_nary_op): Add pattern matching for CSEing sign-changed
36980 or truncated operations with wider ones.
36982 2017-02-27 Richard Biener <rguenther@suse.de>
36984 PR tree-optimization/79690
36985 * tree-vect-stmts.c (vectorizable_store): Use vector type
36986 built from the DR with address-space.
36988 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36990 * doc/invoke.texi (Optimize Options): Refine the description
36991 of asan-use-after-return.
36993 2017-02-25 Alan Modra <amodra@gmail.com>
36995 PR rtl-optimization/79584
36996 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36997 base, not ad->base_term, the reg within base. Remove assertion
36998 that ad->base == ad->base_term. Replace gen_int_mode using
36999 bogus mode with const0_rtx.
37001 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37003 PR middle-end/79396
37004 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37005 FMA_EXPR like tcc_binary or tcc_unary.
37007 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37010 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37012 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37013 (output_loc_operands): Handle DW_OP_call_ref and
37014 DW_OP_GNU_variable_value.
37015 (struct variable_value_struct): New type.
37016 (struct variable_value_hasher): Likewise.
37017 (variable_value_hash): New variable.
37018 (string_types): Remove.
37019 (copy_loc_descr): New function.
37020 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37021 (prepend_loc_descr_to_each): New function.
37022 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37023 instead of add_loc_descr_to_each if the first argument is single
37024 location list and the second has multiple.
37025 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37026 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37027 when looking for variable value which doesn't have other location info.
37028 (loc_list_from_tree): Formatting fix.
37029 (gen_array_type_die): Simplify DW_AT_string_length handling.
37030 (adjust_string_types): Remove.
37031 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37032 string_types. Call resolve_variable_values.
37033 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37034 (resolve_addr_in_expr): Likewise. Add A argument.
37035 (copy_deref_exprloc): Remove deref argument. Adjust for the
37036 original expression being DW_OP_GNU_variable_value with optionally
37037 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37038 optionally after it.
37039 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37040 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37041 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37042 (variable_value_hasher::hash, variable_value_hasher::equal): New
37044 (resolve_variable_value_in_expr, resolve_variable_value,
37045 resolve_variable_values, note_variable_value_in_expr,
37046 note_variable_value): New functions.
37047 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37050 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37053 * opts.h (handle_generated_option): Add GENERATED_P argument.
37054 * opts-common.c (handle_option): Adjust function comment.
37055 (handle_generated_option): Add GENERATED_P argument, pass it to
37057 (control_warning_option): Pass false to handle_generated_option
37059 * opts.c (maybe_default_option): Pass true to handle_generated_option
37061 * optc-gen.awk: Likewise.
37063 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37065 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37066 a REG, look at the REG it is a SUBREG of.
37067 (splitter for cmpeqsi_t): Ditto.
37069 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37071 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37072 the special USEs with the pattern of the insn, not the insn itself.
37074 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37077 * doc/invoke.texi: Document -mload-store-pairs.
37079 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37080 Sandra Loosemore <sandra@codesourcery.com>
37082 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37083 argument isn't a CONST_INT.
37084 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37085 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37086 (nios2_validate_compare): Bypass alternate compare logic if *op2
37087 is not a CONST_INT.
37088 (ldstwm_operation_p): Return false if first_base is not a REG or
37089 if first_offset is not a CONST_INT.
37091 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37093 * config/cris/cris.md: Use correct operand in a define_peephole2.
37095 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37097 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37099 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37101 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37102 this_insn if it is an INSN or JUMP_INSN.
37103 (force_offsettable): Look at base, not at addr.
37104 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37105 on things that aren't necessarily CONST_INTs.
37107 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37109 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37110 -mfpmath=sse is the default also for x86-32 targets with SSE2
37111 instruction set when @option{-ffast-math} is enabled
37113 2017-02-24 Jeff Law <law@redhat.com>
37115 PR rtl-optimizatoin/79286
37116 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37119 2017-02-24 Richard Biener <rguenther@suse.de>
37121 PR tree-optimization/79389
37122 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37125 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37127 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37128 function comment to reflect reality.
37129 (loop_exits_before_overflow): Fix typo in function description.
37131 2017-02-24 Richard Biener <rguenther@suse.de>
37133 PR tree-optimization/79389
37134 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37135 properly that a threading opportunity exists. Detect conditional
37136 copy/constant propagation opportunities.
37138 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37140 * config/visium/visium.md (type): Add trap.
37141 (b): New mode attribute.
37142 (*btst): Rename into...
37143 (*btst<mode>): ...this and adjust.
37144 (*cbranchsi4_btst_insn): Rename into...
37145 (*cbranch<mode>4_btst_insn): ...this and adjust.
37146 (trap): New define_insn.
37148 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37150 PR tree-optimization/79389
37151 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37152 (noce_reversed_cond_code): New function.
37153 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37154 reversed_comparison_code. Formatting fix.
37155 (noce_try_store_flag): Test rev_cond != NULL in addition to
37156 reversed_comparison_code.
37157 (noce_try_store_flag_constants): Likewise.
37158 (noce_try_store_flag_mask): Likewise.
37159 (noce_try_addcc): Use rev_cond if non-NULL instead of
37160 reversed_comparison_code.
37161 (noce_try_cmove_arith): Likewise. Formatting fixes.
37162 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37163 (noce_find_if_block): Initialize rev_cond.
37164 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37165 instead of false as last argument never attempt to reverse it
37168 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37170 PR tree-optimization/79663
37171 * tree-predcom.c (combine_chains): Process refs in reverse order
37172 only for ZERO length chains, and add explaining comment.
37174 2017-02-23 Jeff Law <law@redhat.com>
37176 PR tree-optimization/79578
37177 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37178 in call to operand_equal_p.
37180 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37183 * config/i386/cpuid.h: Fix another undefined behavior.
37185 2017-02-23 Richard Biener <rguenther@suse.de>
37187 PR tree-optimization/79683
37188 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37189 vector types for data-refs.
37191 2017-02-23 Martin Liska <mliska@suse.cz>
37193 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37195 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37197 PR middle-end/79665
37198 * internal-fn.c (get_range_pos_neg): Moved to ...
37199 * tree.c (get_range_pos_neg): ... here. No longer static.
37200 * tree.h (get_range_pos_neg): New prototype.
37201 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37202 are known to be in between 0 and signed maximum inclusive, try to
37203 expand both unsigned and signed divmod and use the cheaper one from
37206 2017-02-22 Jeff Law <law@redhat.com>
37208 PR tree-optimization/79578
37209 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37210 to compare base operands.
37212 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37215 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37216 gpc_reg_operand instead of fpr_reg_operand.
37218 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37220 * config/mips/mips.c (mips_return_in_memory): Force FP
37221 vector types to be returned in memory for o32 ABI.
37223 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37225 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37226 instead of DW_TAG_member for static data member declarations and don't
37227 set no_linkage_name for static inline data members.
37228 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37231 2017-02-22 Martin Liska <mliska@suse.cz>
37233 * doc/invoke.texi: Replace inequality signs with square brackets
37236 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37239 * lra-constraints.c (simplify_operand_subreg): Handle
37240 WORD_REGISTER_OPERATIONS targets.
37242 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37245 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37246 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37247 elimination by swapping fld*.
37249 2017-02-22 Richard Biener <rguenther@suse.de>
37251 PR tree-optimization/79673
37252 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37253 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37254 irrelevant address-space qualifiers and avoiding a
37255 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37257 2017-02-22 Richard Biener <rguenther@suse.de>
37259 PR tree-optimization/79666
37260 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37261 to not symbolically negate if that may introduce undefined
37264 2017-02-22 Martin Liska <mliska@suse.cz>
37267 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37268 * data-streamer-out.c (streamer_write_gcov_count_stream):
37270 * value-prof.c (stream_out_histogram_value): Make assert more
37271 precise based on type of counter.
37273 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37276 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37277 Use nonimmediate_operand instead of memory_operand for operand 1.
37278 (float-extend standard_x87sse_constant_load splitter): Ditto.
37280 2017-02-21 Jeff Law <law@redhat.com>
37282 PR tree-optimization/79621
37283 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37284 blocks with edges to themselves.
37286 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37289 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37290 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37291 Use gimple_call_builtin_p.
37294 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37295 on temporarily removed DEBUG_INSNs.
37297 PR tree-optimization/79649
37298 * tree-loop-distribution.c (classify_partition): Give up on
37299 non-generic address space loads/stores.
37301 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37303 * doc/loop.texi (Loop manipulation): Remove nonexistent
37304 tree_ssa_loop_version from the documentation.
37305 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37307 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37310 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37311 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37312 * config/rs6000/rs6000.c: Include except.h.
37313 (rs6000_expand_split_stack_prologue): Call
37314 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37316 2017-02-21 Martin Jambor <mjambor@suse.cz>
37319 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37320 have been analyzed.
37322 2017-02-21 Martin Jambor <mjambor@suse.cz>
37324 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37325 for backward compatibility only.
37326 * doc/invoke.texi (Option Summary): Remove all references to
37327 -fipa-cp-alignment.
37329 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37333 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37335 * lra-constraints.c (curr_insn_transform): Handle
37336 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37338 2017-02-21 Martin Liska <mliska@suse.cz>
37340 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37342 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37344 PR translation/79638
37345 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37347 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37350 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37351 (arm_function_ok_for_sibcall): Return false for an indirect call by
37352 descriptor if all the argument registers are used.
37353 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37354 alignment of the function.
37356 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37358 PR tree-optimization/61441
37359 * simplify-rtx.c (simplify_const_unary_operation): For
37360 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37361 the sNaN unmodified.
37363 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37365 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37366 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37367 instead of SYSTEM_HEADER_DIR.
37369 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37370 Martin Liška <mliska@suse.cz>
37372 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37373 Fix typos and grammar, use active voice, and clarify.
37375 2017-02-20 Marek Polacek <polacek@redhat.com>
37377 PR middle-end/79537
37378 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37381 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37383 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37386 * config/i386/i386.c (ix86_expand_builtin): Handle
37387 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37388 ix86_builtins_isa[fcode].isa as a requirement of those
37389 flags and any other flag in the bitmask.
37390 (ix86_init_mmx_sse_builtins): Use 0 instead of
37391 ~OPTION_MASK_ISA_64BIT as mask.
37392 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37393 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37394 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37395 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37397 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37400 * lra-constraints.c (split_reg): Check requested split mode
37401 is supported by the register.
37403 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37405 * lra-constraints.c (simplify_operand_subreg): Remove early
37408 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37411 * lra-constraints.c (curr_insn_transform): Tighten condition
37412 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37414 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37417 * lra-constraints.c (curr_insn_transform): Handle
37418 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37420 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37423 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37425 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37427 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37430 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37433 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37435 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37438 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37440 * final.c (last_columnnum, override_columnnum): New variables.
37441 (final_start_function): Set last_columnnum, pass it to begin_prologue
37442 hook and pass 0 to dwarf2out_begin_prologue.
37443 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37444 to source_line debug hook.
37445 (notice_source_line): Compute last_columnnum and for debug_column_info
37446 return true on column changes.
37447 * debug.h (struct gcc_debug_hooks): Add column argument to
37448 source_line and begin_prologue hooks.
37449 (debug_nothing_int_charstar_int_bool): Remove prototype.
37450 (debug_nothing_int_int_charstar,
37451 debug_nothing_int_int_charstar_int_bool): New prototypes.
37452 (dwarf2out_begin_prologue): Add column argument.
37453 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37454 begin_prologue hooks.
37455 (debug_nothing_int_charstar_int_bool): Remove.
37456 (debug_nothing_int_int_charstar,
37457 debug_nothing_int_int_charstar_int_bool): New functions.
37458 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37459 through to dwarf2out_source_line.
37460 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37461 (dwarf2out_source_line): Add column argument, emit it if requested.
37462 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37464 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37465 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37466 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37467 through to dwarf2out_begin_prologue.
37468 (vmsdbgout_source_line): Add column argument, pass it through to
37469 dwarf2out_source_line.
37470 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37471 dbxout_source_line caller.
37472 (dbxout_source_line): Add column argument.
37474 * common.opt (gno-column-info, gcolumn-info): New options.
37475 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37476 (check_die): Also test for multiple DW_AT_decl_column attributes.
37477 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37478 DW_AT_decl_column if requested.
37479 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37481 (gen_variable_die): Likewise.
37482 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37483 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37486 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37487 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37488 (ix86_handle_option): Handle OPT_m3dnowa.
37489 * doc/invoke.texi (-m3dnowa): Document.
37490 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37491 -m3dnowa instead of -m3dnow -march=athlon.
37494 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37495 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37497 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37500 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37501 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37502 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37503 generator for vsx_xxpermdi_<mode>_be.
37504 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37505 force big-endian semantics.
37506 (vsx_xxpermdi_<mode>_be): New define_expand with same
37507 implementation as previous version of vsx_xxpermdi_<mode>.
37509 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37511 PR tree-optimization/79327
37512 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37513 variable, its initialization and use.
37515 2017-02-17 Julia Koval <julia.koval@intel.com>
37517 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37518 (OPTION_MASK_ISA_PKU_UNSET): New.
37519 (ix86_handle_option): Handle -mrdpid.
37520 * config/i386/cpuid.h (bit_RDPID): New.
37521 * config/i386/driver-i386.c (host_detect_local_cpu):
37522 Detect RDPID feature.
37523 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37524 * config/i386/i386-c.c (ix86_target_macros_internal):
37526 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37527 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37528 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37529 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37530 * config/i386/i386.md (define_insn "rdpid"): New.
37531 * config/i386/i386.opt Add -mrdpid.
37532 * config/i386/immintrin.h (_rdpid_u32): New.
37534 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37536 PR rtl-optimization/79541
37537 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37538 instead of transforming it into USE.
37540 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37542 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37543 If HONOR_SNANS (SFmode) force the input to a register.
37544 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37545 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37546 an frsp or similar insn.
37548 2017-02-17 Martin Liska <mliska@suse.cz>
37550 PR rtl-optimization/79577
37551 * params.def (selsched-max-sched-times): Increase minimum to 1.
37553 2017-02-17 Martin Liska <mliska@suse.cz>
37555 PR rtl-optimization/79574
37556 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37558 2017-02-17 Martin Liska <mliska@suse.cz>
37560 PR tree-optimization/79529
37561 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37562 ssa_defined_default_def_p to handle cases which are implicitly
37564 * tree-ssa.c (ssa_defined_default_def_p): New function.
37565 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37566 which are implicitly defined.
37567 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37569 2017-02-17 Richard Biener <rguenther@suse.de>
37571 PR middle-end/79576
37572 * params.def (max-ssa-name-query-depth): Limit to 10.
37574 2017-02-17 Richard Biener <rguenther@suse.de>
37576 PR tree-optimization/79552
37577 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37580 2017-02-17 Richard Biener <rguenther@suse.de>
37583 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37585 2017-02-17 Marek Polacek <polacek@redhat.com>
37587 PR middle-end/79536
37588 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37589 (fold_negate_expr): New wrapper.
37591 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37593 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37594 Correct terminology and de-emphasize pre-standard behavior.
37596 2017-02-16 Alan Modra <amodra@gmail.com>
37598 PR rtl-optimization/79286
37599 * ira.c (def_dominates_uses): New function.
37600 (update_equiv_regs): Don't create an equivalence for insns that
37601 may trap where the register def does not dominate the use.
37603 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37605 PR rtl-optimization/78127
37606 * lra.c (lra): Call lra_eliminate before finish the loop after
37609 2017-02-16 Richard Biener <rguenther@suse.de>
37611 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37613 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37614 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37615 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37616 (isl_val_int_from_wi): New function.
37617 (extract_affine_gmp): Rename to ...
37618 (extract_affine_wi): ... this, take a widest_int.
37619 (extract_affine_int): Just wrap extract_affine_wi.
37620 (add_param_constraints): Use isl_val_int_from_wi.
37621 (add_loop_constraints): Likewise, and extract_affine_wi.
37623 2017-02-15 Jeff Law <law@redhat.com>
37625 PR middle-end/79521
37626 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37627 ira_init_register_move_cost_if_necessary.
37629 2017-02-15 Martin Sebor <msebor@redhat.com>
37631 PR middle-end/32003
37632 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37633 removed in a prior commit.
37635 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37637 PR tree-optimization/79347
37638 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37639 counters during peeling.
37641 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37643 * Makefile.in (site.exp): Remove "set ISLVER".
37645 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37648 * real.c (real_from_integer): Call real_convert even for decimal.
37650 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37653 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37655 2017-02-14 Andrew Pinski <apinski@cavium.com>
37657 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37658 cores and change the partno/implementer to be correct.
37659 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37660 the 'B" as the implementer.
37661 * config/aarch64/aarch64-tune.md: Regenerate.
37663 2017-02-14 Carl Love <cel@us.ibm.com>
37665 * config/rs6000/rs6000.c: Add case statement entry to make the
37666 xvcvuxdsp built-in argument unsigned.
37667 * config/rs6000/vsx.md: Fix the source and return operand types so they
37668 match the instruction definitions from the ISA document. Fix typo
37669 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37672 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37675 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37676 member early_clobber_alts.
37677 * lra-lives.c (reg_early_clobber_p): New.
37678 (process_bb_lives): Use it.
37679 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37680 (debug_operand_data): Initialize early_clobber_alts.
37681 (setup_operand_alternative): Set up early_clobber_alts.
37682 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37683 alternatives to new_insn_reg.
37684 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37686 (lra_update_insn_regno_info): Pass the new arg.
37688 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37690 PR middle-end/79505
37691 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37692 (new_oacc_loop_raw): Don't clear already cleared fields.
37695 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37696 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37697 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37699 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37702 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37704 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37707 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37708 the extra instruction to the right place to store 128-bit constant
37711 2017-02-14 Martin Sebor <msebor@redhat.com>
37713 PR middle-end/79448
37714 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37715 warning for strings of unknown length.
37717 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37719 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37721 2017-02-14 Jeff Law <law@redhat.com>
37724 * ira-costs.c (scan_one_insn): Initialize register move costs
37725 for pseudos seen in USE/CLOBBER insns.
37727 PR tree-optimization/79095
37728 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37729 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37730 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37731 if the operands are known to be not equal, then the resulting range
37733 (intersect_ranges): If the new range is ~[0,0] and the old range is
37734 wide, then prefer ~[0,0].
37735 * tree-vrp.c (overflow_comparison_p_1): New function.
37736 (overflow_comparison_p): New function.
37737 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37738 if NAME is used in an overflow test.
37739 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37740 overflow check that can be expressed as an equality test, then adjust
37741 ops to be that equality test.
37743 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37745 * config/s390/s390-builtin-types.def: Remove flags argument.
37746 * config/s390/s390.c (s390_init_builtins): Likewise.
37748 2017-02-14 Martin Liska <mliska@suse.cz>
37750 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37751 vector. Fix trailing white spaces.
37753 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37755 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37758 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37760 PR rtl-optimization/68664
37761 * config/arm/arm.c (arm_sched_can_speculate_insn):
37762 New function. Declare prototype.
37763 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37765 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37767 PR rtl-optimization/68664
37768 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37770 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37772 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37774 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37775 max skip bytes for function, loop and jump.
37777 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37779 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37780 ABS_EXPR for gimple dump.
37782 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37785 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37787 PR tree-optimization/79408
37788 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37789 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37790 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37791 also if rhs1 is INTEGER_CST.
37793 2017-02-14 Richard Biener <rguenther@suse.de>
37795 PR middle-end/79432
37796 * tree-into-ssa.c (insert_phi_nodes): When the function can
37797 have abnormal edges rewrite SSA names with broken use-def
37798 dominance out of SSA and register them for PHI insertion.
37800 2017-02-13 Martin Sebor <msebor@redhat.com>
37802 PR middle-end/79496
37803 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37804 clearing info.nowrite flag when snprintf size argument is a range.
37806 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37808 * cprop.c (cprop_jump): Add missing space in string literal.
37809 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37810 (get_constraint_for_component_ref): Likewise.
37811 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37812 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37813 * lra-constraints.c (process_alt_operands): Likewise.
37814 * ipa-inline.c (inline_small_functions): Likewise.
37815 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37816 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37817 * trans-mem.c (diagnose_tm_1_op): Likewise.
37818 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37819 (grid_parallel_clauses_gridifiable): Likewise.
37821 * config/nvptx/mkoffload.c (process): Add space in between
37824 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37825 "MOD4_SSE_REGS" and "ALL_REGS".
37827 * spellcheck.c (test_data): Add , in between "foo" and "food".
37829 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37832 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37833 boundary crossing check and subsequent code generation agree.
37835 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37837 * config/aarch64/aarch64.c (has_memory_op): Delete.
37838 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37841 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37843 PR rtl-optimization/79388
37844 PR rtl-optimization/79450
37845 * combine.c (distribute_notes): When removing TEM_INSN for which
37846 corresponding dest has last value recorded, invalidate that last
37849 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37851 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37852 of explicit '@'. Add missing assembly comment marker on branch costs
37855 2017-02-13 Nathan Sidwell <nathan@acm.org>
37857 * gengtype-lex.l (<in_struct>): Add '/'.
37859 2017-02-13 Martin Liska <mliska@suse.cz>
37862 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37864 2017-02-13 Richard Biener <rguenther@suse.de>
37866 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37868 * configure: Re-generate.
37869 * config.in: Likewise.
37870 * graphite-dependences.c: Simplify as if
37871 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37872 * graphite-isl-ast-to-gimple.c: Likewise.
37873 * graphite-optimize-isl.c: Likewise.
37874 * graphite-poly.c: Likewise.
37875 * graphite-sese-to-poly.c: Likewise.
37876 * graphite.h: Likewise.
37877 * toplev.c: Include isl/version.h and use isl_version () for
37878 printing the ISL version.
37879 * doc/install.texi: Update ISL requirement.
37881 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37883 * doc/standards.texi (Standards): Update reference to
37886 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37888 * doc/extend.texi (Named Address Spaces): sourceware.org now
37890 * doc/install.texi (Binaries): Ditto.
37893 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37895 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37896 standard terminology "stringize"/"stringizing" throughout.
37897 * doc/cppinternals.texi: Likewise.
37899 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37901 * doc/extend.texi: Fix some spelling mistakes and typos.
37902 * doc/invoke.texi: Likewise.
37904 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37907 * params.def (inline-min-speedup) Change from 10 to 8.
37909 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37911 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37914 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37917 * ipa-inline-analysis.c (get_minimal_bb): New function.
37918 (record_modified): Use it.
37919 (remap_edge_change_prob): Handle also ancestor functions.
37921 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37923 * doc/contrib.texi (Contributors): Remove broken link into
37924 the Mauve CVS repository.
37926 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37928 PR middle-end/79454
37929 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37930 result computation whenever lhs doesn't have vector mode, not
37931 just when it has BLKmode.
37933 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37935 * doc/makefile.texi (profiledbootstrap): Refer to the
37936 installation instructions only in textual form.
37938 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37941 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37943 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37945 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37946 (Specific): Update mingw-w64 reference.
37948 (Specific): Remove broken link to Renesas RX processor.
37950 2017-02-10 Richard Biener <rguenther@suse.de>
37952 * toplev.c (process_options): Do not mention obsolete graphite
37953 options when printing sorry message about missing graphite support.
37954 Mention -floop-nest-optimize.
37956 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37958 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37959 (vtst_p16): Likewise.
37960 (vtstq_p8): Likewise.
37961 (vtstq_p16): Likewise.
37963 (vtstq_p64): Likewise.
37964 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37965 (vset_lane_p64): New.
37966 (vsetq_lane_p64): New.
37968 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37970 PR tree-optimization/79411
37971 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37972 stmt operands are SSA_NAMEs used in abnormal phis.
37973 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37976 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37979 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37982 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37984 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37986 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37988 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37992 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37993 not arbitrary TREE_CONSTANT.
37996 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37997 "omp declare target link" attribute unless is_global_var.
37998 * omp-offload.c (find_link_var_op): Likewise.
38000 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38001 Chung-Lin Tang <cltang@codesourcery.com>
38003 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38005 (gimplify_adjust_omp_clauses): Don't delete TILE.
38006 (gimplify_omp_for): Deal with TILE.
38007 * internal-fn.c (expand_GOACC_TILE): New function.
38008 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38010 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38011 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38013 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38014 avoid DIV for outermost collapse var.
38015 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38016 Remove out of date comments, fix whitespace.
38017 * omp-general.c (omp_extract_for_data): Deal with tiling.
38018 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38019 adjust OLF_DIM_BASE value.
38020 (struct omp_for_data): Add tiling field.
38021 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38022 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38023 for auto loops. Remove default auto determining, moved to
38024 oacc_loop_fixed_partitions.
38025 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38026 stmts, add e_mask field.
38027 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38028 (oacc_thread_numbers): Use oacc_dim_call.
38029 (oacc_xform_tile): New.
38030 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38031 (finish_oacc_loop): Adjust for ifns vector.
38032 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38033 add case for GOACC_TILE fns.
38034 (oacc_loop_xform_loop): Delete.
38035 (oacc_loop_process): Iterate over call list directly, and add
38036 handling for GOACC_TILE fns.
38037 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38039 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38040 vector partitioning to outer loops. Assign 2 partitions to loops
38041 when available. Add TILE handling.
38042 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38043 (execite_oacc_device_lower): Process GOACC_TILE fns,
38044 ignore unknown specs.
38045 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38046 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38047 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38049 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38051 * configure.ac (ACX_BUGURL): Update.
38052 * configure: Regenerate.
38054 2017-02-09 Richard Biener <rguenther@suse.de>
38056 PR tree-optimization/69823
38057 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38058 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38060 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38062 * config/arc/arc-c.def: Add __NPS400__ definition.
38063 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38064 (TARGET_NPS400): Define.
38066 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38068 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38070 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38071 pointer, arch_info.
38072 (arc_cpu_types): Fill the arch_info field with a pointer into the
38073 arc_arch_types table.
38074 (arc_selected_cpu): Declare.
38075 * config/arc/arc.c (arc_selected_cpu): Make global.
38076 (arc_selected_arch): Delete.
38077 (arc_base_cpu): Delete.
38078 (arc_override_options): Remove references to deleted variables,
38079 update access to arch information.
38080 (ARC_OPT): Update access to arch information.
38081 (ARC_OPTX): Likewise.
38082 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38083 (TARGET_ARC600): Update access to arch information.
38084 (TARGET_ARC601): Likewise.
38085 (TARGET_ARC700): Likewise.
38086 (TARGET_EM): Likewise.
38087 (TARGET_HS): Likewise.
38088 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38091 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38094 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38095 condition/operands for integer GE/LE/GEU/LEU operations.
38097 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38099 PR translation/79397
38100 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38103 2017-02-08 Martin Jambor <mjambor@suse.cz>
38106 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38107 whether allocation happened.
38108 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38109 nothing was allocated.
38111 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38113 PR tree-optimization/79408
38114 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38115 constant, but SSA_NAME with a known integer range, use the minimum
38116 of that range instead of op1 to determine if modulo can be replaced
38117 with its first operand.
38119 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38121 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38123 2017-02-08 Richard Biener <rguenther@suse.de>
38125 PR tree-optimization/71824
38126 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38127 Check all loops contained in the merged region.
38129 2017-02-07 Andrew Pinski <apinski@cavium.com>
38131 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38133 2017-02-07 Andrew Pinski <apinski@cavium.com>
38135 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38136 (thunderxt88): Likewise.
38137 (thunderxt81): Disable LSE and change v8.1 to v8.
38138 (thunderxt83): Likewise.
38140 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38141 Richard Biener <rguenther@suse.de>
38143 PR middle-end/79399
38144 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38145 type from int to size_t.
38146 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38148 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38150 PR rtl-optimization/79386
38151 * cprop.c (bypass_conditional_jumps): Initialize
38152 bypass_last_basic_block already before splitting bbs after
38153 unconditional traps...
38154 (bypass_conditional_jumps): ... rather than here.
38157 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38158 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38159 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38160 fix -masm=intel patterns.
38162 2017-02-07 Richard Biener <rguenther@suse.de>
38164 PR tree-optimization/79256
38165 PR middle-end/79278
38166 * builtins.c (get_object_alignment_2): Use min_align_of_type
38167 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38168 and ADJUST_FIELD_ALIGN.
38170 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38172 * doc/tm.texi: Regenerate.
38173 * stor-layout.c (layout_decl): Adjust.
38174 (update_alignment_for_field): Likewise.
38175 (place_field): Likewise.
38176 (min_align_of_type): Likewise.
38177 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38178 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38179 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38180 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38181 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38182 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38183 * config/i386/i386.c (x86_field_alignment): Likewise.
38184 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38185 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38186 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38187 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38188 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38189 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38193 2017-01-30 Richard Biener <rguenther@suse.de>
38195 PR tree-optimization/79256
38196 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38197 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38200 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38202 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38203 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38204 builtins to SImode and emit a zero-extend, if necessary.
38206 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38208 * docs/invoke.texi (RISC-V Options): Alphabetize.
38210 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38212 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38215 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38217 * config/riscv/riscv.c: New file.
38218 * common/config/riscv/riscv-common.c: Likewise.
38219 * config.gcc: Likewise.
38220 * config/riscv/constraints.md: Likewise.
38221 * config/riscv/elf.h: Likewise.
38222 * config/riscv/generic.md: Likewise.
38223 * config/riscv/linux.h: Likewise.
38224 * config/riscv/multilib-generator: Likewise.
38225 * config/riscv/peephole.md: Likewise.
38226 * config/riscv/pic.md: Likewise.
38227 * config/riscv/predicates.md: Likewise.
38228 * config/riscv/riscv-builtins.c: Likewise.
38229 * config/riscv/riscv-c.c: Likewise.
38230 * config/riscv/riscv-ftypes.def: Likewise.
38231 * config/riscv/riscv-modes.def: Likewise.
38232 * config/riscv/riscv-opts.h: Likewise.
38233 * config/riscv/riscv-protos.h: Likewise.
38234 * config/riscv/riscv.h: Likewise.
38235 * config/riscv/riscv.md: Likewise.
38236 * config/riscv/riscv.opt: Likewise.
38237 * config/riscv/sync.md: Likewise.
38238 * config/riscv/t-elf-multilib: Likewise.
38239 * config/riscv/t-linux: Likewise.
38240 * config/riscv/t-linux-multilib: Likewise.
38241 * config/riscv/t-riscv: Likewise.
38242 * configure.ac: Likewise.
38243 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38244 Waterman as RISC-V maintainers.
38245 * doc/install.texi: Add RISC-V entries.
38246 * doc/invoke.texi: Add RISC-V options section.
38247 * doc/md.texi: Add RISC-V constraints section.
38248 * configure: Regenerated.
38250 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38253 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38254 false values to be constant vectors with all 0 or all 1 bits set.
38255 (vcondu<mode><mode>): Likewise.
38256 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38258 (fpmask_comparison_operator): Update comment.
38259 (vecint_comparison_operator): New predicate.
38260 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38261 vector conditionals when the true and false values are constant
38262 vectors with all 0 bits or all 1 bits set.
38264 2017-02-06 Martin Sebor <msebor@redhat.com>
38266 PR tree-optimization/79376
38267 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38269 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38271 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38272 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38273 to simplify split condition.
38275 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38277 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38278 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38281 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38283 PR rtl-optimization/68664
38284 * target.def (can_speculate_insn): New hook.
38285 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38286 * doc/tm.texi: Regenerate.
38287 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38288 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38289 (rs6000_sched_can_speculate_insn): New function.
38291 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38293 PR tree-optimization/79284
38294 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38295 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38296 vectorizable_mask_load_store, vectorizable_operation,
38297 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38298 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38299 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38300 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38301 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38302 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38303 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38304 is_gimple_assign (stmt). Replace another such test with
38305 is_gimple_assign (stmt).
38307 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38310 * config/avr/avr.c (rtl-iter.h): Include it.
38311 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38312 (avr_legitimate_combined_insn): ...and implementation.
38314 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38316 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38317 * config/s390/s390.c (s390_const_operand_ok)
38318 (s390_canonicalize_comparison, s390_extract_part)
38319 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38320 (s390_contiguous_bitmask_p, s390_rtx_costs)
38321 (legitimize_pic_address): Likewise.
38322 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38323 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38324 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38325 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38326 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38328 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38330 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38331 REGNO($0) == REGNO($1).
38333 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38335 * config/s390/linux.h(SIZE_TYPE): Add comment.
38337 2017-02-06 Julian Brown <julian@codesourcery.com>
38338 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38339 Virendra Pathak <virendra.pathak@broadcom.com>
38341 * config/aarch64/aarch64-cores.def: Change the scheduler
38343 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38344 * config/aarch64/thunderx2t99.md: New file.
38346 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38348 * doc/standards.texi (Go Language): Update link to language
38351 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38353 * tree-eh.c (lower_resx): Sanitize profile.
38354 (cleanup_empty_eh_move_lp): Likewise.
38356 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38359 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38361 * cfgloopmanip.h (loop_version): Update prototype.
38362 * modulo-sched.c (sms_schedule): Update call of loop_version.
38363 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38364 * tree-parloops.c (gen_parallel_loop): Likewise.
38365 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38366 * tree-ssa-loop-split.c (split_loop): Likewise.
38367 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38368 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38370 2017-02-05 Martin Liska <mliska@suse.cz>
38373 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38375 (print_operand_address): Initialize a struct to zero.
38377 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38379 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38380 garbage collector only in textual form.
38382 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38384 * doc/extend.texi (x86 specific memory model extensions for
38385 transactional memory): Simplify a phrase.
38387 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38390 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38391 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38392 (atomic_storedi_1): Likewise.
38394 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38396 PR tree-optimization/79338
38397 * tree-parloops.c (gather_scalar_reductions): Don't call
38398 vect_analyze_loop_form for loop->inner before destroying loop's
38401 2017-02-03 Martin Sebor <msebor@redhat.com>
38403 PR tree-optimization/79327
38404 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38405 when precision has resulted in leading zeros.
38406 (format_integer): Adjust the likely counter to assume an unknown
38407 argument that may be zero is non-zero.
38409 2017-02-03 Jason Merrill <jason@redhat.com>
38412 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38413 avoid copying non-taken branch.
38415 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38417 PR tree-optimization/79340
38418 * tree-vect-loop.c (vectorizable_reduction): Release
38419 vec_defs elements after safe_splicing them into other vectors.
38422 PR tree-optimization/79327
38423 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38424 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38426 (format_integer): Use wide_int_to_tree instead of build_int_cst
38427 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38428 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38429 of shortest and longest sequence.
38431 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38433 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38434 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38436 2017-02-03 Walter Lee <walt@tilera.com>
38439 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38440 after initial stackframe link reg save.
38441 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38443 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38446 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38447 wu for stxssp alternative.
38449 2017-02-03 Martin Sebor <msebor@redhat.com>
38451 PR tree-optimization/79352
38452 * gimple-fold.c (get_range_strlen): Add argument.
38453 (get_range_strlen): Change return type to bool.
38454 (get_maxval_strlen): Pass in a dummy argument.
38455 * gimple-fold.h (get_range_strlen): Change return type to bool.
38456 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38457 * tree.h (array_at_struct_end_p): Add argument.
38458 * tree.c (array_at_struct_end_p): Handle it.
38460 2017-02-03 Martin Liska <mliska@suse.cz>
38463 * multiple_target.c (create_dispatcher_calls): Redirect edge
38464 from a caller of a dispatcher.
38465 (expand_target_clones): Make the clones local.
38466 (ipa_target_clone): Do both target clones and resolvers.
38467 (ipa_dispatcher_calls): Remove the pass.
38468 (pass_dispatcher_calls::gate): Likewise.
38469 (make_pass_dispatcher_calls): Likewise.
38470 * passes.def (pass_target_clone): Put as very first IPA early
38473 2017-02-03 Martin Liska <mliska@suse.cz>
38475 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38476 in case of a function with ifunc attribute.
38478 2017-02-03 Martin Liska <mliska@suse.cz>
38480 * cgraph.c (cgraph_node::dump): Dump function version info.
38481 * symtab.c (symtab_node::dump_base): Add missing new line.
38483 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38485 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38486 (ifcombine_ifandif): Use it.
38488 2017-02-03 Martin Liska <mliska@suse.cz>
38490 * doc/invoke.texi: Document default value for
38491 use-after-scope-direct-emission-threshold.
38493 2017-02-03 Martin Liska <mliska@suse.cz>
38495 PR tree-optimization/79339
38496 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38497 (format_floating): Likewise.
38499 2017-02-03 Martin Liska <mliska@suse.cz>
38502 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38504 (ipa_node_params_t::remove): Likewise.
38505 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38506 initialization from removed ipa_node_params_t::insert.
38507 (ipa_node_params::~ipa_node_params): Move from removed
38508 ipa_node_params_t::release.
38509 * symbol-summary.h (symbol_summary::m_released): New member.
38510 Do not release a summary twice. Do not allow to call finalizer
38511 for types of a summary that live in GGC memory.
38513 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38515 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38518 2017-02-02 Martin Sebor <msebor@redhat.com>
38520 PR middle-end/79275
38521 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38522 (format_string): Tighten up the range of output for non-constant
38523 strings and correct the expected range for wide non-constant strings.
38525 2017-02-02 Martin Sebor <msebor@redhat.com>
38527 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38529 PR middle-end/32003
38530 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38532 (-fdump-tree-@var): Add to index and document how to come up
38533 with pass-specific option and dump file names.
38534 (-fdump-passes): Clarify where to look for output.
38536 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38538 PR middle-end/77445
38539 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38540 statistics of the analyzed path; allow threading for speed when
38541 any of BBs along the path are optimized for speed.
38543 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38545 PR middle-end/78468
38546 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38547 settings of the virtual registers.
38550 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38552 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38553 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38556 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38558 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38559 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38561 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38563 * config/s390/s390.md: Add missing comments with the expanded
38565 * config/s390/vector.md: Likewise.
38566 * config/s390/vx-builtins.md: Likewise.
38568 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38571 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38572 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38573 conditions on a single line.
38575 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38577 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38578 __S390_VX__ to __VX__.
38580 2017-02-01 Andrew Pinski <apinski@cavium.com>
38582 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38583 stmt_info to record_stmt_cost.
38584 (vect_get_known_peeling_cost): Pass stmt_info if known to
38586 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38587 cpu_vector_cost field into
38588 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38589 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38590 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38591 splitting of scalar_stmt_cost and vec_stmt_cost.
38592 (thunderx_vector_cost): Likewise.
38593 (cortexa57_vector_cost): LIkewise.
38594 (exynosm1_vector_cost): Likewise.
38595 (xgene1_vector_cost): Likewise.
38596 (thunderx2t99_vector_cost): Improve after the splitting of the two
38598 (aarch64_builtin_vectorization_cost): Update for the splitting of
38599 scalar_stmt_cost and vec_stmt_cost.
38601 2017-02-01 Torvald Riegel <triegel@redhat.com>
38602 Richard Henderson <rth@redhat.com>
38604 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38605 conditional on existance of a fast atomic load.
38606 * optabs-query.c (can_atomic_load_p): New function.
38607 * optabs-query.h (can_atomic_load_p): Declare it.
38608 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38609 no fast atomic load is available for the particular size of access.
38610 (expand_atomic_compare_and_swap): Likewise.
38611 (expand_atomic_load): Likewise.
38612 (expand_atomic_store): Likewise.
38613 (expand_atomic_fetch_op): Likewise.
38614 * testsuite/lib/target-supports.exp
38615 (check_effective_target_sync_int_128): Remove x86 because it provides
38616 no fast atomic load.
38617 (check_effective_target_sync_int_128_runtime): Likewise.
38619 2017-02-01 Richard Biener <rguenther@suse.de>
38621 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38622 (graphite_transform_loops): Provide opt-info for optimized nests.
38623 * tree-parloop.c (parallelize_loops): Provide opt-info for
38624 parallelized loops.
38626 2017-02-01 Richard Biener <rguenther@suse.de>
38628 PR middle-end/79315
38629 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38630 was not set before.
38632 2017-02-01 Richard Biener <rguenther@suse.de>
38634 PR tree-optimization/71824
38635 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38636 Verify the loops are valid in the merged SESE region.
38637 (scop_detection::can_represent_loop_1): Check analyzing the
38638 evolution of the number of iterations in the region succeeds.
38640 2017-01-31 Ian Lance Taylor <iant@golang.org>
38642 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38643 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38645 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38647 PR preprocessor/79210
38648 * input.c (get_substring_ranges_for_loc): Replace line_width
38649 assertion with error-handling.
38651 2017-01-31 Richard Biener <rguenther@suse.de>
38653 PR tree-optimization/77318
38654 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38655 (create_pw_aff_from_tree): Take loop parameter.
38656 (add_condition_to_pbb): Pass loop of the condition to
38657 create_pw_aff_from_tree.
38659 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38661 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38662 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38664 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38668 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38670 (convert_int_to_float128): Likewise.
38671 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38672 (convert_int_to_float128): Likewise.
38673 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38674 (UNSPEC_IEEE128_CONVERT): Likewise.
38675 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38676 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38677 Use local variables for IBM extended format.
38678 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38679 (fix_trunc<mode>si2_fprs): Likewise.
38680 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38681 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38682 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38683 to know that we can now have integers of all sizes in vector
38685 (fix<uns>_<mode>di2_hw): Likewise.
38686 (float<uns>_<mode>si2_hw): Likewise.
38687 (fix_<mode>si2_hw): Likewise.
38688 (fixuns_<mode>si2_hw): Likewise.
38689 (float<uns>_<mode>di2_hw): Likewise.
38690 (float_<mode>di2_hw): Likewise.
38691 (float_<mode>si2_hw): Likewise.
38692 (floatuns_<mode>di2_hw): Likewise.
38693 (floatuns_<mode>si2_hw): Likewise.
38694 (xscvqp<su>wz_<mode>): Delete, no longer used.
38695 (xscvqp<su>dz_<mode>): Likewise.
38696 (xscv<su>dqp_<mode>): Likewise.
38697 (ieee128_mfvsrd_64bit): Likewise.
38698 (ieee128_mfvsrd_32bit): Likewise.
38699 (ieee128_mfvsrwz): Likewise.
38700 (ieee128_mtvsrw): Likewise.
38701 (ieee128_mtvsrd_64bit): Likewise.
38702 (ieee128_mtvsrd_32bit): Likewise.
38704 2017-01-31 Martin Liska <mliska@suse.cz>
38707 * ipa-prop.c (ipa_free_all_node_params): Call release method
38708 instead of ~sumbol_summary to not to trigger double times
38711 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38713 PR tree-optimization/71691
38714 * bitmap.h (class auto_bitmap): New.
38715 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38716 is_maybe_undefined instead of ssa_undefined_value_p.
38718 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38720 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38721 __S390_ARCH_LEVEL__ to __ARCH__.
38723 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38725 PR tree-optimization/79267
38726 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38727 if should_remove_lhs_p is true.
38729 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38732 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38733 (add_alignment_attribute): New.
38734 (base_type_die): Add alignment attribute.
38735 (subrange_type_die): Likewise.
38736 (modified_type_die): Likewise.
38737 (gen_array_type_die): Likewise.
38738 (gen_descr_array_type_die: Likewise.
38739 (gen_enumeration_type_die): Likewise.
38740 (gen_subprogram_die): Likewise.
38741 (gen_variable_die): Likewise.
38742 (gen_field_die): Likewise.
38743 (gen_ptr_to_mbr_type_die): Likewise.
38744 (gen_struct_or_union_type_die): Likewise.
38745 (gen_subroutine_type_die): Likewise.
38746 (gen_typedef_die): Likewise.
38747 (base_type_cmp): Compare alignment attribute.
38749 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38752 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38753 (setb_unsigned) New pattern for setb with CCUNS.
38754 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38755 subfc./subfe sequence to avoid overflow problems. Generate a
38756 shorter sequence with cmpld/setb for power9.
38757 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38758 for generating subfc. instruction.
38759 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38760 now uses this instruction.
38762 2017-01-30 Ian Lance Taylor <iant@google.com>
38765 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38766 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38768 2017-01-30 Martin Sebor <msebor@redhat.com>
38770 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38771 Move constant to the right of a relational operator.
38772 (get_mpfr_format_length, format_character, format_string): Ditto.
38773 (should_warn_p, maybe_warn): Same.
38775 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38777 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38780 * asan.c (get_translation_unit_decl): Remove function.
38781 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38783 2017-01-30 Martin Liska <mliska@suse.cz>
38785 PR gcov-profile/79259
38786 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38787 -fprofile-generate.
38789 2017-01-30 Martin Liska <mliska@suse.cz>
38792 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38793 Initialize variables with NULL value.
38795 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38798 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38800 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38802 2017-01-30 Richard Biener <rguenther@suse.de>
38804 PR tree-optimization/79276
38805 * tree-vrp.c (process_assert_insertions): Properly adjust common
38806 when removing a duplicate.
38808 2017-01-30 Richard Biener <rguenther@suse.de>
38810 PR tree-optimization/79256
38811 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38812 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38814 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38816 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38819 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38820 ("*r<noxa>sbg_<mode>_sll_bitmask")
38821 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38822 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38823 Use contiguous_bitmask_nowrap_operand.
38825 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38828 * config/rs6000/altivec.h (vec_xl): Revise #define.
38829 (vec_xst): Likewise.
38831 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38833 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38835 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38837 PR rtl-optimization/79194
38838 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38839 traps before call to bypass_conditional_jumps.
38841 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38843 PR tree-optimization/71374
38844 * lra-constraints.c (check_conflict_input_operands): New.
38845 (match_reload): Use it.
38847 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38850 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38851 account to calculate conflict_set.
38853 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38855 PR rtl-optimization/78559
38856 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38857 other_insn in combine.
38859 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38861 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38862 uint16_type_node for BT_UINT16.
38864 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38866 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38867 "RTL Tests" to menu.
38868 (GIMPLE Tests): New node.
38869 (RTL Tests): New node.
38871 2017-01-27 Richard Biener <rguenther@suse.de>
38873 PR tree-optimization/79245
38874 * tree-loop-distribution.c (distribute_loop): Apply cost
38875 modeling also to detected patterns.
38877 2017-01-27 Richard Biener <rguenther@suse.de>
38879 PR tree-optimization/71433
38880 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38881 (compare_assert_loc): New function.
38882 (process_assert_insertions): Sort and optimize assert locations
38883 to remove duplicates and push down identical assertions on
38884 edges to their destination block.
38886 2017-01-27 Richard Biener <rguenther@suse.de>
38888 PR tree-optimization/79244
38889 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38890 out SSA names even if abnormal.
38892 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38894 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38895 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38896 instead of MPFR_RNDN.
38898 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38901 * arm.c (arm_option_override): Don't call build_target_option_node
38902 until after doing all option overrides.
38903 (arm_valid_target_attribute_tree): Likewise.
38905 2017-01-27 Martin Liska <mliska@suse.cz>
38907 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38908 for {cd}tors and C++ {cd}tors.
38910 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38912 * config/s390/s390.md ("*setmem_long_and")
38913 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38915 2017-01-26 Martin Sebor <msebor@redhat.com>
38917 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38920 2017-01-26 Martin Sebor <msebor@redhat.com>
38922 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38923 HAVE_DFmode before using XFmode or DFmode.
38924 (parse_directive): Avoid using the z length modifier to avoid
38925 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38927 PR middle-end/78703
38928 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38929 to accept adjustment as an array.
38930 (get_int_range): New function.
38931 (struct directive): Make width and prec arrays.
38932 (directive::set_width, directive::set_precision): Call get_int_range.
38933 (format_integer, format_floating): Handle width and precision ranges.
38934 (format_string, parse_directive): Same.
38936 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38939 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38940 comdat_type_p set, just clone them, but keep the children in the
38944 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38945 which have direct callers with -fvar-tracking-assignments enabled
38947 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38948 inside of type units.
38950 2017-01-26 Martin Sebor <msebor@redhat.com>
38952 PR middle-end/78703
38953 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38955 (struct format_result): Replace number_chars, number_chars_min,
38956 and number_chars_max with a single member of struct result_range.
38958 (format_result::operator+=): Adjust.
38959 (struct fmtresult): Remove bounded. Handle likely and unlikely
38961 (fmtresult::adjust_for_width_or_precision): New function.
38962 (fmtresult:type_max_digits): New function.
38963 (bytes_remaining): Handle likely and unlikely counters.
38964 (min_bytes_remaining): Remove.
38965 (format_percent): Simplify.
38966 (format_integer, format_floating): Set likely and unlikely counters.
38967 (get_string_length, format_character, format_string): Same.
38968 (format_plain, should_warn_p): New function.
38969 (maybe_warn): Call should_warn_p. Update diagnostic messages
38970 and handle those for all directives, including plain strings.
38971 (format_directive): Handle likely and unlikely counters.
38972 Remove unnecessary quoting from diagnostics. Add an informational
38974 (add_bytes): Remove.
38975 (pass_sprintf_length::compute_format_length): Simplify.
38976 (try_substitute_return_value): Handle likely and unlikely counters.
38978 2017-01-26 Carl Love <cel@us.ibm.com>
38980 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38981 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38983 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38986 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38987 endianess for subregs into account.
38988 * lra-constraints.c (lra_constraints): Do risky transformations
38989 always on the first iteration.
38990 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38991 last_call_used_reg_set.
38992 (process_bb_lives): Define and use last_call_used_reg_set.
38993 * lra.c (lra): Always continue after lra_constraints on the first
38996 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38998 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39000 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39002 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39004 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39005 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39006 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39007 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39008 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39009 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39010 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39011 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39012 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39014 2017-01-26 Marek Polacek <polacek@redhat.com>
39017 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39018 for the third operand.
39020 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39022 PR middle-end/79236
39023 * omp-low.c (struct omp_context): Add simt_stmt field.
39024 (scan_omp_for): Return omp_context *.
39025 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39026 context to the _simt_ SIMD stmt.
39027 (lower_omp_for): For combined SIMD with sibling _simt_
39028 SIMD, make sure to use the same decls in _looptemp_
39029 clauses as in the sibling.
39031 2017-01-26 David Sherwood <david.sherwood@arm.com>
39033 PR middle-end/79212
39034 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39037 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39040 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39041 emit fld b; fld a; if possible.
39043 * brig-builtins.def: Update copyright years.
39044 * config/arm/arm_acle_builtins.def: Update copyright years.
39046 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39049 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39050 constraint instead of o for the stxsd instruction.
39052 2017-01-25 Carl Love <cel@us.ibm.com>
39054 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39055 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39057 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39059 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39061 2017-01-25 Richard Biener <rguenther@suse.de>
39063 PR tree-optimization/69264
39064 * target.def (vector_alignment_reachable): Improve documentation.
39065 * doc/tm.texi: Regenerate.
39066 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39068 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39069 earlier changes with respect to TYPE_USER_ALIGN.
39070 (vector_alignment_reachable_p): Likewise. Improve dumping.
39072 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39075 * config/arm/arm.md (xordi3): Force constant operand into a register
39078 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39080 * doc/invoke.texi (-fstore-merging): Correct default optimization
39081 levels at which it is enabled.
39082 (-O): Move -fstore-merging from list to...
39085 2017-01-25 Richard Biener <rguenther@suse.de>
39088 * omp-expand.c: Include debug.h.
39089 (expand_omp_taskreg): Make sure to generate early debug before
39090 outlining anything from a function.
39091 (expand_omp_target): Likewise.
39092 (grid_expand_target_grid_body): Likewise.
39094 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39097 * asan.c (get_translation_unit_decl): New function.
39098 (asan_add_global): Extract modules file name from globals
39099 TRANSLATION_UNIT_DECL name.
39101 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39104 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39105 for long calls with APCS frame and VFP.
39107 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39109 * cfg.c (original_copy_tables_initialized_p): New function.
39110 * cfg.h (original_copy_tables_initialized_p): New decl.
39111 * cfgrtl.c (relink_block_chain): Guard the call to
39112 free_original_copy_tables with a call to
39113 original_copy_tables_initialized_p.
39114 * cgraph.h (symtab_node::native_rtl_p): New decl.
39115 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39116 (symtab_node::needed_p): Don't assert for early assembly output
39117 for __RTL functions.
39118 (cgraph_node::finalize_function): Set "force_output" for __RTL
39120 (cgraph_node::analyze): Bail out early for __RTL functions.
39121 (analyze_functions): Update assertion to support __RTL functions.
39122 (cgraph_node::expand): Bail out early for __RTL functions.
39123 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39125 * function.h (struct function): Update comment for field
39127 * gimple-expr.c: Include "tree-pass.h".
39128 (gimple_has_body_p): Return false for __RTL functions.
39129 * Makefile.in (OBJS): Add run-rtl-passes.o.
39130 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39132 (gcc::pass_manager::get_clean_slate): New accessor.
39133 * passes.c: Include "insn-addr.h".
39134 (should_skip_pass_p): Add logging. Update logic for running
39135 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39136 property-provider override so it is only done for gimple passes.
39138 (skip_pass): New function.
39139 (execute_one_pass): Call skip_pass when skipping passes.
39140 * read-md.c (md_reader::read_char): Support filtering
39141 the input to a subset of line numbers.
39142 (md_reader::md_reader): Initialize fields
39143 m_first_line and m_last_line.
39144 (md_reader::read_file_fragment): New function.
39145 * read-md.h (md_reader::read_file_fragment): New decl.
39146 (md_reader::m_first_line): New field.
39147 (md_reader::m_last_line): New field.
39148 * read-rtl-function.c (function_reader::create_function): Only
39149 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39150 curr_properties. Set DECL_INITIAL to a dummy block.
39151 (read_rtl_function_body_from_file_range): New function.
39152 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39154 * run-rtl-passes.c: New file.
39155 * run-rtl-passes.h: New file.
39157 2017-01-24 Jeff Law <law@redhat.com>
39159 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39162 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39164 PR tree-optimization/79159
39165 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39166 (record_nonwrapping_iv): Improve boundary using above function if no
39167 value range information.
39169 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39170 Martin Jambor <mjambor@suse.cz>
39172 * brig-builtins.def: New file.
39173 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39174 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39175 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39176 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39177 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39178 * builtin-types.def (BT_INT8): New.
39179 (BT_INT16): Likewise.
39180 (BT_UINT8): Likewise.
39181 (BT_UINT16): Likewise.
39182 (BT_FN_ULONG): Likewise.
39183 (BT_FN_UINT_INT): Likewise.
39184 (BT_FN_UINT_ULONG): Likewise.
39185 (BT_FN_UINT_LONG): Likewise.
39186 (BT_FN_UINT_PTR): Likewise.
39187 (BT_FN_ULONG_PTR): Likewise.
39188 (BT_FN_INT8_FLOAT): Likewise.
39189 (BT_FN_INT16_FLOAT): Likewise.
39190 (BT_FN_UINT32_FLOAT): Likewise.
39191 (BT_FN_UINT16_FLOAT): Likewise.
39192 (BT_FN_UINT8_FLOAT): Likewise.
39193 (BT_FN_UINT64_FLOAT): Likewise.
39194 (BT_FN_UINT16_UINT32): Likewise.
39195 (BT_FN_UINT32_UINT16): Likewise.
39196 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39197 (BT_FN_INT_PTR_INT): Likewise.
39198 (BT_FN_UINT_PTR_UINT): Likewise.
39199 (BT_FN_LONG_PTR_LONG): Likewise.
39200 (BT_FN_ULONG_PTR_ULONG): Likewise.
39201 (BT_FN_VOID_UINT64_UINT64): Likewise.
39202 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39203 (BT_FN_INT8_INT8_INT8): Likewise.
39204 (BT_FN_INT16_INT16_INT16): Likewise.
39205 (BT_FN_INT_INT_INT): Likewise.
39206 (BT_FN_UINT_FLOAT_UINT): Likewise.
39207 (BT_FN_FLOAT_UINT_UINT): Likewise.
39208 (BT_FN_ULONG_UINT_UINT): Likewise.
39209 (BT_FN_ULONG_UINT_PTR): Likewise.
39210 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39211 (BT_FN_UINT_UINT_UINT): Likewise.
39212 (BT_FN_VOID_UINT_PTR): Likewise.
39213 (BT_FN_UINT_UINT_PTR: Likewise.
39214 (BT_FN_UINT32_UINT64_PTR): Likewise.
39215 (BT_FN_INT_INT_UINT_UINT): Likewise.
39216 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39217 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39218 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39219 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39220 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39221 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39222 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39223 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39224 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39225 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39226 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39227 * doc/frontends.texi: List BRIG FE.
39228 * doc/install.texi (Testing): Add BRIG tesring requirements.
39229 * doc/invoke.texi (Overall Options): Mention BRIG.
39230 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39232 2017-01-24 Richard Biener <rguenther@suse.de>
39234 PR translation/79208
39235 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39237 2017-01-24 Martin Jambor <mjambor@suse.cz>
39240 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39241 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39242 and known_contexts.
39244 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39246 PR middle-end/79123
39247 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39248 casts from signed to unsigned really don't have a range.
39250 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39252 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39253 GMP_RNDx for compatiblity.
39255 2017-01-24 Martin Liska <mliska@suse.cz>
39258 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39259 that would prevent us to call alloca with -1 as argument.
39261 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39263 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39264 Avoid -Wformat-security warning.
39266 2017-01-23 Andrew Pinski <apinski@cavium.com>
39268 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39271 2017-01-23 Martin Sebor <msebor@redhat.com>
39273 PR middle-end/78703
39274 * gimple-ssa-sprintf.c (warn_level): New global.
39275 (format_integer): Use it here and throughout the rest of the file.
39276 Use the same switch to compute sign as base.
39277 (maybe_warn): New function.
39278 (format_directive): Factor out warnings into maybe_warn.
39279 Add debugging output. Use warn_level.
39280 (add_bytes): Use warn_level.
39281 (pass_sprintf_length::compute_format_length): Add debugging output.
39282 (try_substitute_return_value): Same.
39283 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39285 PR middle-end/78703
39286 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39287 (struct fmtresult, format_integer, format_floating): Adjust.
39288 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39289 (get_string_length, format_string,format_directive): Same.
39290 (pass_sprintf_length::compute_format_length): Same.
39291 (try_substitute_return_value): Simplify slightly.
39293 PR middle-end/78703
39294 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39295 (fmtresult::operator+=): Outlined.
39296 (struct fmtresult): Add ctors.
39297 (struct conversion_spec): Rename...
39298 (struct directive): ...to this. Add and remove data members.
39299 (directive::set_width, directive::set_precision): New functions.
39300 (format_percent): Use fmtresult ctor.
39301 (get_width_and_precision): Remove.
39302 (format_integer): Make naming changes. Avoid computing width and
39304 (format_floating): Same. Adjust indentation.
39305 (format_character, format_none): New functions.
39306 (format_string): Moved character handling to format_character.
39307 (format_directive): Remove arguments, change return type.
39308 (parse_directive): New function.
39309 (pass_sprintf_length::compute_format_length): Move directive
39310 parsing to parse_directive.
39312 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39314 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39315 (assign_assembler_name_if_needed): ... this.
39316 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39317 (assign_assembler_name_if_needed): ... this.
39318 (free_lang_data_in_cgraph): Adjust callers.
39319 * cgraphunit.c (cgraph_node::analyze): Likewise.
39320 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39323 2017-01-23 Richard Biener <rguenther@suse.de>
39325 PR tree-optimization/79088
39326 PR tree-optimization/79188
39327 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39328 resetting loop bounds after last path deletion. Reset loop
39329 bounds of the target loop, make code match the comments.
39330 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39331 Make sure loops need no fixups.
39333 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39335 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39336 exponent support with double type for first argument.
39337 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39338 type returned by __builtin_vec_extract_sig,
39339 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39340 functions from "vector int" to "vector unsigned int" or from
39341 "vector long long int" to "vector unsigned long long int".
39342 Changed type returned by __builtin_vec_extract_exp,
39343 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39344 functions from "vector int" to "vector unsigned int" or from
39345 "vector long long int" to "vector unsigned long long int".
39346 Changed return type of __builtin_vec_test_data_class,
39347 __builtin_vec_test_data_class_sp, and
39348 __builtin_vec_test_data_class_dp from "vector int" to
39349 "vector bool int" or from "vector long long int" to "vector bool
39350 long long int" and changed second argument type from "unsigned
39351 int" to "int". Added new overloaded function forms "vector float
39352 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39353 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39354 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39355 double, vector unsigned long long int)" and "vector double
39356 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39357 long int)". Changed return type of
39358 __builtin_scalar_test_data_class and
39359 __builtin_scalar_test_data_class_sp and
39360 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39361 int" and changed second argument from "unsigned int" to "int".
39362 Changed type returned by __builtin_scalar_test_neg,
39363 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39364 from "int" to "bool int". Added new overloaded function form
39365 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39366 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39367 exponent double-precision with floating point first argument.
39368 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39369 documentation of scalar_test_data_class, scalar_test_neg,
39370 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39371 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39372 vec_test_data_class built-in functions to reflect refinements in
39373 their type signatures.
39375 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39377 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39379 (aarch64_elf_asm_destructor): Likewise.
39381 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39383 PR rtl-optimization/78634
39384 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39385 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39386 * ifcvt.c (noce_try_cmove): Add missing cost check.
39388 PR rtl-optimization/71724
39389 * combine.c (if_then_else_cond): Look for situations where it is
39390 beneficial to undo the work of one of the recursive calls.
39392 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39394 PR tree-optimization/70754
39395 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39396 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39397 combined stmt before it if not NULL.
39398 (combine_chains): Process refs reversely and compute dominance point
39401 2017-01-23 Martin Liska <mliska@suse.cz>
39403 PR tree-optimization/79196
39404 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39405 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39407 (strlen_optimize_stmt): Call the renamed function.
39409 2017-01-23 Michael Matz <matz@suse.de>
39411 PR tree-optimization/78384
39412 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39414 2017-01-23 Richard Biener <rguenther@suse.de>
39416 PR tree-optimization/79186
39417 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39418 both incoming edges before moving an assert.
39420 2017-01-23 Martin Jambor <mjambor@suse.cz>
39422 * ipa-prop.c (load_from_param_1): Removed.
39423 (load_from_unmodified_param): Bits from load_from_param_1 put back
39425 (load_from_param): Removed.
39426 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39427 with stmt. Reverted back to use of load_from_unmodified_param.
39429 2017-01-23 Martin Jambor <mjambor@suse.cz>
39432 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39433 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39434 field a pointer to garbage collected vector, mark lattices and
39435 ipcp_orig_node with GTY((skip)).
39436 (ipa_get_param_count): Adjust to descriptors being a pointer.
39437 (ipa_get_param): Likewise.
39438 (ipa_get_type): Likewise.
39439 (ipa_get_param_move_cost): Likewise.
39440 (ipa_set_param_used): Likewise.
39441 (ipa_get_controlled_uses): Likewise.
39442 (ipa_set_controlled_uses): Likewise.
39443 (ipa_is_param_used): Likewise.
39444 (ipa_node_params_t): Move into garbage collector. New methods insert
39446 (ipa_node_params_sum): Annotate wth GTY(()).
39447 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39449 (ipa_load_from_parm_agg): Adjust declaration.
39450 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39451 * ipa-profile.c (ipa_profile): Likewise.
39452 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39453 (ipa_populate_param_decls): Make descriptors parameter garbage
39455 (ipa_dump_param): Adjust to descriptors being a pointer.
39456 (ipa_alloc_node_params): Likewise.
39457 (ipa_initialize_node_params): Likewise.
39458 (load_from_param_1): Make descriptors parameter garbage collected.
39459 (load_from_unmodified_param): Likewise.
39460 (load_from_param): Likewise.
39461 (ipa_load_from_parm_agg): Likewise.
39462 (ipa_node_params::~ipa_node_params): Removed.
39463 (ipa_free_all_node_params): Remove call to delete operator.
39464 (ipa_node_params_t::insert): New.
39465 (ipa_node_params_t::remove): Likewise.
39466 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39467 copy known_csts and known_contexts vectors.
39468 (ipa_read_node_info): Adjust to descriptors being a pointer.
39469 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39471 (ipcp_transform_function): Make descriptors variable garbage
39474 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39476 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39477 * config/i386/avx512dqintrin.h: Ditto.
39478 * config/i386/avx512fintrin.h: Ditto.
39479 * config/i386/i386.c: Handle new builtins.
39480 * config/i386/i386-builtin.def: Add new builtins.
39481 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39482 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39484 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39485 Martin Liska <mliska@suse.cz>
39487 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39488 * asan.c (asan_expand_poison_ifn): Support stores and use
39489 appropriate ASAN report function.
39490 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39491 * internal-fn.def (ASAN_POISON_USE): Declare.
39492 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39493 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39494 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39495 ASAN_POISON calls w/o LHS.
39496 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39497 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39498 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39499 * gimplify.c (asan_poison_variables): Add attribute
39500 use_after_scope_memory to variables that really needs to live
39502 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39503 having the attribute.
39505 2017-01-23 Martin Liska <mliska@suse.cz>
39507 * asan.c (create_asan_shadow_var): New function.
39508 (asan_expand_poison_ifn): Likewise.
39509 * asan.h (asan_expand_poison_ifn): New declaration.
39510 * internal-fn.c (expand_ASAN_POISON): Likewise.
39511 * internal-fn.def (ASAN_POISON): New builtin.
39512 * sanopt.c (pass_sanopt::execute): Expand
39513 asan_expand_poison_ifn.
39514 * tree-inline.c (copy_decl_for_dup_finish): Make function
39516 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39517 * tree-ssa.c (is_asan_mark_p): New function.
39518 (execute_update_addresses_taken): Rewrite local variables
39519 (identified just by use-after-scope as addressable) into SSA.
39521 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39523 * doc/install.texi (Specific): opensource.apple.com uses https
39524 now. Remove trailing slash.
39526 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39528 * README.Portability: Remove note on an Irix compatibility issue.
39530 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39532 * gcov.c (INCLUDE_ALGORITHM): Define.
39533 (INCLUDE_VECTOR): Define.
39534 No longer include <vector> and <algorithm> directly.
39536 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39538 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39540 * doc/invoke.texi (Code Gen Options): Ditto.
39542 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39545 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39547 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39549 rtl-optimization/79125
39550 * cprop.c (local_cprop_pass): Handle cases where we make an
39551 unconditional trap.
39553 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39557 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39558 read from, for big endian.
39560 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39562 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39563 register pauth builtins for LP64 only.
39565 2017-01-20 Marek Polacek <polacek@redhat.com>
39568 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39571 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39573 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39575 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39576 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39577 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39579 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39582 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39583 in big-endian mode when they are not a single duplicated value.
39585 2017-01-20 Richard Biener <rguenther@suse.de>
39587 * BASE-VER: Bump to 7.0.1.
39589 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39591 * omp-low.c (omplow_simd_context): New struct. Use it...
39592 (lower_rec_simd_input_clauses): ...here and...
39593 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39594 references to idx, lane, max_vf, is_simt.
39596 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39598 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39601 2017-01-20 Martin Jambor <mjambor@suse.cz>
39603 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39604 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39606 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39607 (GTFILES): Rename hsa.c to hsa-common.c.
39608 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39609 * hsa-dump.c: Likewise.
39610 * hsa-gen.c: Likewise.
39611 * hsa-regalloc.c: Likewise.
39612 * ipa-hsa.c: Likewise.
39613 * omp-expand.c: Likewise.
39614 * omp-low.c: Likewise.
39615 * toplev.c: Likewise.
39617 2017-01-20 Marek Polacek <polacek@redhat.com>
39620 * doc/invoke.texi: Document -Wduplicated-branches.
39621 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39622 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39623 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39624 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39625 return 0 only when not OEP_LEXICOGRAPHIC.
39626 (fold_build_cleanup_point_expr): Use the expression
39627 location when building CLEANUP_POINT_EXPR.
39628 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39629 * tree.c (add_expr): Handle error_mark_node.
39631 2017-01-20 Martin Liska <mliska@suse.cz>
39634 * tree-profile.c (init_ic_make_global_vars): Do not call
39636 (gimple_init_gcov_profiler): Likewise.
39638 2017-01-20 Martin Liska <mliska@suse.cz>
39641 * cgraph.h (maybe_create_reference): Remove argument and
39643 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39645 * ipa-cp.c (create_specialized_node): Likewise.
39646 * symtab.c (symtab_node::maybe_create_reference): Handle
39647 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39649 2017-01-20 Martin Liska <mliska@suse.cz>
39651 * read-rtl-function.c (function_reader::create_function): Use
39652 build_decl instread of build_decl_stat.
39654 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39656 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39657 * config/i386/avx512dqintrin.h: Ditto.
39658 * config/i386/avx512fintrin.h: Ditto.
39659 * config/i386/i386-builtin-types.def: Add new types.
39660 * config/i386/i386.c: Handle new types.
39661 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39662 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39663 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39664 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39665 (__builtin_ia32_kshiftridi): New.
39666 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39668 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39672 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39673 define to rs6000_init_stack_protect_guard.
39674 (rs6000_init_stack_protect_guard): New function.
39676 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39677 Yunqiang Su <yunqiang.su@imgtec.com>
39679 * config.gcc (supported_defaults): Add madd4.
39680 (with_madd4): Add validation.
39681 (all_defaults): Add madd4.
39682 * config/mips/mips.opt (mmadd4): New option.
39683 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39685 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39687 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39688 (ISA_HAS_FUSED_MADD4): Likewise.
39689 * doc/invoke.texi (-mmadd4): Document the new option.
39690 * doc/install.texi (--with-madd4): Document the new option.
39692 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39694 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39695 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39696 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39697 (aarch64_init_pauth_hint_builtins): New.
39698 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39699 (aarch64_expand_builtin): Expand new builtins.
39701 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39703 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39704 * combine-stack-adj.c (no_unhandled_cfa): Handle
39705 REG_CFA_TOGGLE_RA_MANGLE.
39706 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39707 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39708 info for return address signing.
39709 (aarch64_expand_epilogue): Likewise.
39711 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39713 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39714 * config/aarch64/aarch64-protos.h
39715 (aarch64_return_address_signing_enabled): New declaration.
39716 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39718 (aarch64_expand_prologue): Sign return address before it's pushed onto
39720 (aarch64_expand_epilogue): Authenticate return address fetched from
39722 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39723 (aarch64_attributes): New function attributes for "sign-return-address".
39724 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39725 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39726 ("*do_return"): Generate combined instructions according to key index.
39727 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39728 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39730 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39731 * config/aarch64/aarch64.opt (msign-return-address=): New.
39732 * doc/extend.texi (AArch64 Function Attributes): Documents
39733 "sign-return-address=".
39734 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39736 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39738 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39739 overall option summary.
39741 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39743 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39744 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39745 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39746 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39748 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39750 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39751 -mpower9-minmax by default for -mcpu=power9.
39752 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39753 128-bit floating point.
39755 2017-01-20 Alan Modra <amodra@gmail.com>
39757 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39758 optimizing for size.
39760 2017-01-20 Alan Modra <amodra@gmail.com>
39763 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39764 for strcmp and strncmp from corresponding builtin decl.
39766 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39768 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39769 instead of i386/rtems-64.h.
39770 * config/i386/rtems-64.h: Remove.
39772 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39776 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39778 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39780 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39782 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39783 Change int to HOST_WIDE_INT.
39784 * config/aarch64/aarch64-protos.h
39785 (aarch64_simd_gen_const_vector_dup): Likewise.
39786 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39788 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39790 * langhooks-def.h (lhd_type_for_size): New decl.
39791 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39792 * langhooks.c (lhd_type_for_size): New function, taken from
39795 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39797 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39798 define_bypass for CR latency.
39799 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39800 (power9-alu2): Add define_bypass for CR latency.
39802 (power9-mul): Update insn latency.
39803 (power9-mul-compare): Update insn latency, bypass latency and remove
39806 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39808 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39810 * config/aarch64/aarch64.md
39811 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39812 aarch64_nopcrelative_literal_loads.
39813 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39815 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39817 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39818 TARGET_LOONGSON_3A.
39819 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39821 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39824 * config.gcc (supported_defaults): Add lxc1-sxc1.
39825 (with_lxc1_sxc1): Add validation.
39826 (all_defaults): Add lxc1-sxc1.
39827 * config/mips/mips.opt (mlxc1-sxc1): New option.
39828 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39830 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39831 __mips_no_lxc1_sxc1.
39832 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39833 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39834 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39836 2017-01-19 Richard Biener <rguenther@suse.de>
39838 PR tree-optimization/72488
39839 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39840 sure to restore SSA info.
39841 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39843 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39845 PR rtl-optimization/79121
39846 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39847 of the inner type when shifting an extended value.
39849 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39852 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39853 interposable aliases.
39855 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39858 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39859 Use the evmergelohi instruction.
39860 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39861 (mov_sitf_e500_subreg8_2_be): Likewise.
39862 (mov_sitf_e500_subreg12_2_le): Likewise.
39863 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39864 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39865 (mov_sitf_e500_subreg8_2_le): Likewise.
39866 (mov_sitf_e500_subreg12_2_be): Likewise.
39868 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39870 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39871 attribute from vecsimple to vecperm.
39872 (altivec_vbpermq2): Likewise.
39874 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39877 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39879 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39880 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39881 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39882 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39883 case where N arg is SIZE_MAX.
39884 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39885 (cmpstrsi): Add pattern.
39887 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39889 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39890 __builtin_vec_revb builtins.
39891 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39892 built-in functions to support generation of the ISA 3.0 XXBR<x>
39893 vector byte reverse instructions.
39894 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39895 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39896 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39897 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39898 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39899 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39900 (P9V_BUILTIN_VEC_REVB): Likewise.
39901 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39902 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39903 (p9_xxbrq_v16qi): Likewise.
39904 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39905 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39906 (p9_xxbrh_v8hi): Likewise.
39907 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39908 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39909 vec_revb built-in functions.
39911 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39913 PR rtl-optimization/78952
39914 * config/i386/i386.md (any_extract): New code iterator.
39915 (*insvqi_2): Use any_extract for source operand.
39916 (*insvqi_3): Use any_shiftrt for source operand.
39918 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39920 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39922 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39924 2017-01-18 Matthias Klose <doko@ubuntu.com>
39926 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39928 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39930 * config/rs6000/altivec.h (vec_bperm): Change #define.
39931 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39932 (altivec_vbpermq2): New define_insn.
39933 (altivec_vbpermd): Likewise.
39934 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39935 function interface.
39936 (VBPERMD): Likewise.
39937 (VBPERM): New polymorphic function interface.
39938 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39939 Add entries for P9V_BUILTIN_VEC_VBPERM.
39940 * doc/extend.texi: Add interfaces for vec_bperm.
39942 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39944 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39945 first letter of error messages.
39946 (s390_resolve_overloaded_builtin): Likewise.
39947 * config/s390/s390.c (s390_expand_builtin): Likewise.
39948 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39949 (s390_valid_target_attribute_inner_p): Likewise.
39950 * config/s390/s390.md ("tabort"): Likewise.
39952 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39954 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39955 (ISA_AVOID_DIV_HILO): New macro.
39956 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39957 (ISA_HAS_DDIV): Likewise.
39959 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39961 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39963 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39965 * doc/invoke.texi (fabi-version): Spelling fix.
39967 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39970 * doc/invoke.texi (fabi-version): Mention mangling fix for
39973 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39976 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39978 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39981 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39982 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39984 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39986 (stack_protect_test): Ditto.
39987 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39988 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39990 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39991 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39992 -mstack-protector-guard-offset=.
39993 (RS/6000 and PowerPC Options): Ditto.
39995 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39997 * config/i386/i386.h (MASK_CLASS_P): New define.
39998 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39999 there are no registers from different register sets also when
40000 mask registers are used. Update function comment.
40001 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40002 to (*k/*r) and (*k/*km) alternatives.
40004 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40006 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40007 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40008 (EH_RETURN_HANDLER_RTX): New define.
40009 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40010 Force frame pointer in EH return functions.
40011 (aarch64_expand_epilogue): Add barrier for eh_return.
40012 (aarch64_final_eh_return_addr): Remove.
40013 (aarch64_eh_return_handler_rtx): New function.
40014 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40016 (aarch64_eh_return_handler_rtx): New prototype.
40018 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40020 * config/rs6000/altivec.h (vec_rlmi): New #define.
40021 (vec_vrlnm): Likewise.
40022 (vec_rlnm): Likewise.
40023 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40024 (UNSPEC_VRLNM): Likewise.
40025 (VIlong): New mode iterator.
40026 (altivec_vrl<VI_char>mi): New define_insn.
40027 (altivec_vrl<VI_char>nm): Likewise.
40028 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40030 (VRLDNM): Likewise.
40031 (RLNM): New polymorphic function entry.
40032 (VRLWMI): New monomorphic function entry.
40033 (VRLDMI): Likewise.
40034 (RLMI): New polymorphic function entry.
40035 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40036 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40037 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40040 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40043 * dwarf2out.c (field_byte_offset): Restore the
40044 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40045 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40046 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40049 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40052 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40054 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40057 * dwarf2out.c (add_data_member_location_attribute): For constant
40058 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40059 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40060 DW_AT_byte_size attributes.
40062 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40064 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40065 after forcing to constant memory when the code model is medium.
40067 2017-01-17 Julia Koval <julia.koval@intel.com>
40070 * config/i386/avx512fintrin.h
40071 (_mm512_i32gather_ps): Change __addr type to void const*.
40072 (_mm512_mask_i32gather_ps): Ditto.
40073 (_mm512_i32gather_pd): Ditto.
40074 (_mm512_mask_i32gather_pd): Ditto.
40075 (_mm512_i64gather_ps): Ditto.
40076 (_mm512_mask_i64gather_ps): Ditto.
40077 (_mm512_i64gather_pd): Ditto.
40078 (_mm512_mask_i64gather_pd): Ditto.
40079 (_mm512_i32gather_epi32): Ditto.
40080 (_mm512_mask_i32gather_epi32): Ditto.
40081 (_mm512_i32gather_epi64): Ditto.
40082 (_mm512_mask_i32gather_epi64): Ditto.
40083 (_mm512_i64gather_epi32): Ditto.
40084 (_mm512_mask_i64gather_epi32): Ditto.
40085 (_mm512_i64gather_epi64): Ditto.
40086 (_mm512_mask_i64gather_epi64): Ditto.
40087 (_mm512_i32scatter_ps): Change __addr type to void*.
40088 (_mm512_mask_i32scatter_ps): Ditto.
40089 (_mm512_i32scatter_pd): Ditto.
40090 (_mm512_mask_i32scatter_pd): Ditto.
40091 (_mm512_i64scatter_ps): Ditto.
40092 (_mm512_mask_i64scatter_ps): Ditto.
40093 (_mm512_i64scatter_pd): Ditto.
40094 (_mm512_mask_i64scatter_pd): Ditto.
40095 (_mm512_i32scatter_epi32): Ditto.
40096 (_mm512_mask_i32scatter_epi32): Ditto.
40097 (_mm512_i32scatter_epi64): Ditto.
40098 (_mm512_mask_i32scatter_epi64): Ditto.
40099 (_mm512_i64scatter_epi32): Ditto.
40100 (_mm512_mask_i64scatter_epi32): Ditto.
40101 (_mm512_i64scatter_epi64): Ditto.
40102 (_mm512_mask_i64scatter_epi64): Ditto.
40103 * config/i386/avx512pfintrin.h
40104 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40105 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40106 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40107 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40108 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40109 (_mm512_prefetch_i32scatter_ps): Ditto.
40110 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40111 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40112 (_mm512_prefetch_i64scatter_pd): Ditto.
40113 (_mm512_prefetch_i64scatter_ps): Ditto.
40114 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40115 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40116 * config/i386/avx512vlintrin.h
40117 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40118 (_mm_mmask_i32gather_ps): Ditto.
40119 (_mm256_mmask_i32gather_pd): Ditto.
40120 (_mm_mmask_i32gather_pd): Ditto.
40121 (_mm256_mmask_i64gather_ps): Ditto.
40122 (_mm_mmask_i64gather_ps): Ditto.
40123 (_mm256_mmask_i64gather_pd): Ditto.
40124 (_mm_mmask_i64gather_pd): Ditto.
40125 (_mm256_mmask_i32gather_epi32): Ditto.
40126 (_mm_mmask_i32gather_epi32): Ditto.
40127 (_mm256_mmask_i32gather_epi64): Ditto.
40128 (_mm_mmask_i32gather_epi64): Ditto.
40129 (_mm256_mmask_i64gather_epi32): Ditto.
40130 (_mm_mmask_i64gather_epi32): Ditto.
40131 (_mm256_mmask_i64gather_epi64): Ditto.
40132 (_mm_mmask_i64gather_epi64): Ditto.
40133 (_mm256_i32scatter_ps): Change __addr type to void*.
40134 (_mm256_mask_i32scatter_ps): Ditto.
40135 (_mm_i32scatter_ps): Ditto.
40136 (_mm_mask_i32scatter_ps): Ditto.
40137 (_mm256_i32scatter_pd): Ditto.
40138 (_mm256_mask_i32scatter_pd): Ditto.
40139 (_mm_i32scatter_pd): Ditto.
40140 (_mm_mask_i32scatter_pd): Ditto.
40141 (_mm256_i64scatter_ps): Ditto.
40142 (_mm256_mask_i64scatter_ps): Ditto.
40143 (_mm_i64scatter_ps): Ditto.
40144 (_mm_mask_i64scatter_ps): Ditto.
40145 (_mm256_i64scatter_pd): Ditto.
40146 (_mm256_mask_i64scatter_pd): Ditto.
40147 (_mm_i64scatter_pd): Ditto.
40148 (_mm_mask_i64scatter_pd): Ditto.
40149 (_mm256_i32scatter_epi32): Ditto.
40150 (_mm256_mask_i32scatter_epi32): Ditto.
40151 (_mm_i32scatter_epi32): Ditto.
40152 (_mm_mask_i32scatter_epi32): Ditto.
40153 (_mm256_i32scatter_epi64): Ditto.
40154 (_mm256_mask_i32scatter_epi64): Ditto.
40155 (_mm_i32scatter_epi64): Ditto.
40156 (_mm_mask_i32scatter_epi64): Ditto.
40157 (_mm256_i64scatter_epi32): Ditto.
40158 (_mm256_mask_i64scatter_epi32): Ditto.
40159 (_mm_i64scatter_epi32): Ditto.
40160 (_mm_mask_i64scatter_epi32): Ditto.
40161 (_mm256_i64scatter_epi64): Ditto.
40162 (_mm256_mask_i64scatter_epi64): Ditto.
40163 (_mm_i64scatter_epi64): Ditto.
40164 (_mm_mask_i64scatter_epi64): Ditto.
40165 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40166 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40167 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40168 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40169 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40170 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40171 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40172 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40173 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40174 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40175 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40176 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40177 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40178 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40179 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40180 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40181 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40182 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40183 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40184 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40185 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40186 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40187 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40188 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40189 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40190 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40191 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40192 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40193 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40194 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40195 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40196 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40197 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40198 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40199 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40200 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40201 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40202 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40203 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40204 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40205 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40206 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40207 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40208 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40209 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40210 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40211 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40212 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40213 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40214 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40215 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40216 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40217 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40218 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40219 definitions accordingly.
40221 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40222 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40225 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40228 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40231 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40232 hard regs for inner regno.
40234 2017-01-17 Martin Liska <mliska@suse.cz>
40237 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40238 assumption and add comment.
40240 2017-01-17 Nathan Sidwell <nathan@acm.org>
40242 * ipa-visibility.c (localize_node): New function, broken out of ...
40243 (function_and_variable_visibility): ... here. Call it.
40245 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40247 PR middle-end/77445
40248 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40249 correctly set frequency of oudgoing edge.
40250 (duplicate_thread_path): Fix profile updating.
40252 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40255 * configure.ac: Add GCC_BASE_VER.
40256 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40257 version from BASE-VER file.
40258 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40259 (gcc.o): Depend on $(BASEVER).
40260 * common.opt (dumpfullversion): New option.
40261 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40262 * doc/invoke.texi: Document -dumpfullversion.
40263 * doc/install.texi: Document --with-gcc-major-version-only.
40264 * configure: Regenerated.
40266 2017-01-17 Richard Biener <rguenther@suse.de>
40268 PR tree-optimization/71433
40269 * tree-vrp.c (register_new_assert_for): Merge same asserts
40270 on all incoming edges.
40271 (process_assert_insertions_for): Handle insertions at the
40274 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40276 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40277 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40279 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40282 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40285 2017-01-17 Alan Modra <amodra@gmail.com>
40288 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40289 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40290 symbolic stack limit when pic.
40292 2017-01-16 Martin Sebor <msebor@redhat.com>
40294 PR tree-optimization/78608
40295 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40297 2017-01-16 Jeff Law <law@redhat.com>
40300 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40301 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40302 for several include directories that may be relative to sysroot.
40303 * config/i386/x-mingw32 (gplus_includedir): Define.
40304 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40305 (native_system_includedir): Likewise.
40306 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40307 override if TARGET_SYSTEM_ROOT is defined.
40308 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40310 PR tree-optimization/79090
40311 PR tree-optimization/33562
40312 PR tree-optimization/61912
40313 PR tree-optimization/77485
40314 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40315 and computed trims into the dump file.
40317 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40319 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40321 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40324 * gimplify.c (gimplify_init_constructor): If want_value and
40325 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40329 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40330 sequence. Formatting fixes.
40331 (doloop_optimize): Formatting fixes.
40334 * gcc.c (debug_level_greater_than_spec_func): New function.
40335 (static_spec_functions): Add debug-level-gt spec function.
40336 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40338 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40339 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40340 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40341 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40342 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40343 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40345 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40347 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40348 QImode fixups to general and mask registers only.
40350 2017-01-16 Carl Love <cel@us.ibm.com>
40352 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40353 for built-in functions
40354 vector signed char vec_nabs (vector signed char)
40355 vector signed short vec_nabs (vector signed short)
40356 vector signed int vec_nabs (vector signed int)
40357 vector signed long long vec_nabs (vector signed long long)
40358 vector float vec_nabs (vector float)
40359 vector double vec_nabs (vector double)
40360 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40362 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40363 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40364 * doc/extend.texi: Update the documentation file for the new built-in
40367 2017-01-16 Martin Sebor <msebor@redhat.com>
40369 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40372 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40374 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40375 UNSPEC_VSX__XXSPLTD to require special splat handling.
40377 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40380 * system.h: Poison strndup.
40382 2017-01-16 Alan Modra <amodra@gmail.com>
40385 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40388 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40390 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40392 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40394 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40395 call recog here. Assert that INSN_CODE (insn) is non-negative.
40397 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40400 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40402 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40403 in the currently scheduled RTL fragment.
40405 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40407 PR rtl-optimization/78751
40408 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40411 2017-01-14 Jeff Law <law@redhat.com>
40413 PR tree-optimization/79090
40414 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40415 variable length stores.
40416 (compute_trims): Delete dead assignment to *trim_tail.
40417 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40420 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40422 PR rtl-optimization/78626
40423 PR rtl-optimization/78727
40424 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40425 of a block, and split such blocks after everything else is finished.
40427 2017-01-14 Alan Modra <amodra@gmail.com>
40430 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40431 target legitimate_combined_insn.
40432 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40433 (rs6000_legitimate_combined_insn): New function.
40434 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40436 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40437 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40438 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40440 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40442 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40444 2017-01-13 Jeff Law <law@redhat.com>
40446 PR tree-optimization/33562
40447 PR tree-optimization/61912
40448 PR tree-optimization/77485
40449 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40451 (delete_dead_assignment): Likewise.
40452 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40453 statement to delete_dead_call and delete_dead_assignment.
40455 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40458 * substring-locations.c (format_warning_va): Strengthen case 1 so
40459 that both endpoints of the substring must be within the format
40460 range for just the substring to be printed.
40462 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40464 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40465 * config/i386/i386.c (ix86_target_string): Add missing options
40466 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40467 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40468 flags_other and ix86_target_other to flags2_other. Display unknown
40470 (ix86_valid_target_attribute_inner_p): Add missing options and
40471 reorder options by implied ISAs, as in ix86_target_string.
40473 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40475 * hash-table.h (hash_table::too_empty_p): New function.
40476 (hash_table::expand): Use it.
40477 (hash_table::traverse): Likewise.
40478 (hash_table::empty_slot): Use sizeof (value_type) instead of
40479 sizeof (PTR) to convert bytes to elements. Shrink the table
40480 if the current size is excessive for the current number of
40483 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40485 * ira-costs.c (record_reg_classes): Break from the inner loop
40486 early once alt_fail is known to be true. Update outer loop
40487 handling accordingly.
40489 2017-01-13 Jeff Law <law@redhat.com>
40491 * tree-ssa-dse.c (decrement_count): New function.
40492 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40493 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40494 when we know the partially dead statement is a mem* function.
40496 PR tree-optimization/61912
40497 PR tree-optimization/77485
40498 * tree-ssa-dse.c: Include expr.h.
40499 (maybe_trim_constructor_store): New function.
40500 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40502 PR tree-optimization/33562
40503 PR tree-optimization/61912
40504 PR tree-optimization/77485
40505 * doc/invoke.texi: Document new dse-max-object-size param.
40506 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40507 * tree-ssa-dse.c: Include params.h.
40508 (dse_store_status): New enum.
40509 (initialize_ao_ref_for_dse): New, partially extracted from
40511 (valid_ao_ref_for_dse, normalize_ref): New.
40512 (setup_live_bytes_from_ref, compute_trims): Likewise.
40513 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40514 (maybe_trim_partially_dead_store): Likewise.
40515 (maybe_trim_complex_store): Likewise.
40516 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40517 Track what bytes live from the original store. Return tri-state
40518 for dead, partially dead or live.
40519 (dse_dom_walker): Add constructor, destructor and new private members.
40520 (delete_dead_call, delete_dead_assignment): New extracted from
40522 (dse_optimize_stmt): Make a member of dse_dom_walker.
40523 Use initialize_ao_ref_for_dse.
40525 PR tree-optimization/33562
40526 PR tree-optimization/61912
40527 PR tree-optimization/77485
40528 * sbitmap.h (bitmap_count_bits): Prototype.
40529 (bitmap_clear_range, bitmap_set_range): Likewise.
40530 * sbitmap.c (bitmap_clear_range): New function.
40531 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40533 2017-01-13 Martin Liska <mliska@suse.cz>
40536 * function.c (set_cfun): Add new argument force.
40537 * function.h (set_cfun): Likewise.
40538 * ipa-inline-transform.c (inline_call): Use the function when
40539 strict alising from is dropped for function we inline to.
40541 2017-01-13 Richard Biener <rguenther@suse.de>
40543 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40544 for dumping GIMPLE INTEGER_CSTs.
40546 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40548 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40549 to 201112L since C++17.
40551 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40554 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40555 if -fsanitize=kernel-address is present.
40557 2017-01-13 Richard Biener <rguenther@suse.de>
40559 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40560 as _Literal ( type ) number in case usual suffixes do not
40561 preserve all information.
40563 2017-01-13 Richard Biener <rguenther@suse.de>
40565 PR tree-optimization/77283
40566 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40567 and ssa-iterators.h.
40568 (is_feasible_trace): Implement a cost model based on joiner
40571 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40574 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40575 char or short to __float128/_Float128 directly.
40577 2017-01-12 Martin Sebor <msebor@redhat.com>
40579 to -Wformat-overflow.
40580 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40581 (min_bytes_remaining): Same.
40582 (get_string_length): Same.
40583 (format_string): Same.
40584 (format_directive): Same.
40586 (pass_sprintf_length::handle_gimple_call): Same.
40588 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40590 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40591 info.nowrite calls with no lhs that can't throw. Return bool
40592 whether gsi_remove has been called or not.
40593 (pass_sprintf_length::handle_gimple_call): Return bool whether
40594 try_substitute_return_value called gsi_remove. Formatting fix.
40595 (pass_sprintf_length::execute): Don't use gsi_remove if
40596 handle_gimple_call returned true.
40599 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40600 be removed due to side-effects, don't remove following barrier nor
40601 turn the successor edge into fallthru edge.
40603 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40606 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40607 element-reversing loads and stores as not swappable.
40609 2017-01-12 Nathan Sidwell <nathan@acm.org>
40610 Nicolai Stange <nicstange@gmail.com>
40612 * combine.c (try_combine): Don't ignore result of overlap checking
40613 loop. Combine overlap & asm check into single loop.
40615 2017-01-12 Richard Biener <rguenther@suse.de>
40617 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40618 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40620 2017-01-12 Richard Biener <rguenther@suse.de>
40622 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40623 and TS_TARGET_OPTION directly derive from TS_BASE.
40624 * tree-core.h (tree_optimization_option): Derive from tree_base.
40625 (tree_target_option): Likewise.
40627 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40629 * config/i386/i386.c (memory_address_length): Increase len
40630 only when rip_relative_addr_p returns false.
40632 2017-01-11 Julia Koval <julia.koval@intel.com>
40634 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40635 (OPTION_MASK_ISA_SGX_SET): New.
40636 (ix86_handle_option): Handle OPT_msgx.
40637 * config.gcc: Added sgxintrin.h.
40638 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40639 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40640 * config/i386/i386.c (ix86_target_string): Add -msgx.
40642 (ix86_option_override_internal): Handle new options.
40643 (ix86_valid_target_attribute_inner_p): Add sgx.
40644 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40645 * config/i386/i386.opt: Add msgx.
40646 * config/i386/sgxintrin.h: New file.
40647 * config/i386/x86intrin.h: Add sgxintrin.h.
40649 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40652 * fold-const.c (maybe_nonzero_address): Return 1 for function
40654 (tree_single_nonzero_warnv_p): Don't handle function local objects
40658 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40661 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40664 * opts.c: Include "spellcheck.h"
40665 (struct string_fragment): New struct.
40666 (struct edit_distance_traits<const string_fragment &>): New
40668 (get_closest_sanitizer_option): New function.
40669 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40671 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40673 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40675 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40676 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40677 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40678 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40679 for initial die_offset if dwarf_split_debug_info.
40680 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40681 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40683 (output_skeleton_debug_sections): Formatting fix. Use
40684 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40685 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40687 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40689 * config/arm/cortex-a53.md: Add bypasses for
40690 cortex_a53_r2f_cvt.
40691 (cortex_a53_r2f): Only use for transfers.
40692 (cortex_a53_f2r): Likewise.
40693 (cortex_a53_r2f_cvt): Add reservation for conversions.
40694 (cortex_a53_f2r_cvt): Likewise.
40696 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40698 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40699 to all inlined functions, change static to extern.
40701 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40704 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40706 (arm_assemble_integer): Likewise.
40708 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40710 * config.gcc: Use new awk script to check CPU, FPU and architecture
40711 parameters for --with-... options.
40712 * config/arm/parsecpu.awk: New file
40713 * config/arm/arm-cpus.in: New file.
40714 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40716 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40718 * config/arm/t-arm: Update dependency rules.
40719 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40720 of processing .def files.
40721 * config/arm/genopt.sh: Deleted.
40722 * config/arm/gentune.sh: Deleted.
40723 * config/arm/arm-cores.def: Deleted.
40724 * config/arm/arm-arches.def: Deleted.
40725 * config/arm/arm-fpus.def: Deleted.
40726 * config/arm/arm-tune.md: Regenerated.
40727 * config/arm/arm-tables.opt: Regenerated.
40728 * config/arm/arm-cpu.h: New generated file.
40729 * config/arm/arm-cpu-data.h: New generated file.
40730 * config/arm/arm-cpu-cdata.h: New generated file.
40732 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40735 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40737 (input_varpool_node): Unpack dynamically_initialized bit.
40739 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40741 PR rtl-optimization/79032
40742 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40743 the alignment of the adjusted memory reference against that of MODE,
40744 instead of the alignment of the original memory reference.
40746 2017-01-11 Martin Jambor <mjambor@suse.cz>
40748 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40750 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40751 decorated functions.
40753 2017-01-11 Richard Biener <rguenther@suse.de>
40755 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40756 set range/nonnull info for PHI results. Do not set it on
40757 stmts marked for removal.
40759 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40761 * expr.c (store_field): In the bitfield case, fetch the return value
40762 from the registers before applying a single big-endian adjustment.
40763 Always do a final load for a BLKmode value not larger than a word.
40765 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40768 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40769 that we correctly handle column numbers greater than
40770 LINE_MAP_MAX_COLUMN_NUMBER.
40772 2017-01-10 Martin Sebor <msebor@redhat.com>
40774 PR middle-end/78245
40775 * gimple-ssa-sprintf.c (get_destination_size): Call
40776 {init,fini}object_sizes.
40777 * tree-object-size.c (addr_object_size): Adjust.
40778 (pass_through_call): Adjust.
40779 (pass_object_sizes::execute): Adjust.
40780 * tree-object-size.h (fini_object_sizes): Declare.
40782 2017-01-10 Martin Sebor <msebor@redhat.com>
40784 PR tree-optimization/78775
40785 * builtins.c (get_size_range): Move...
40786 * calls.c: ...to here.
40787 (alloc_max_size): Accept zero argument.
40788 (operand_signed_p): Remove.
40789 (maybe_warn_alloc_args_overflow): Call get_size_range.
40790 * calls.h (get_size_range): Declare.
40792 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40794 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40795 from TI's devices.csv file as of September 2016.
40796 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40798 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40800 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40801 * doc/invoke.texi: Likewise.
40802 * doc/md.texi: Likewise.
40803 * doc/objc.texi: Likewise.
40805 2017-01-10 Joshua Conner <joshconner@google.com>
40807 * config/arm/fuchsia-elf.h: New file.
40808 * config/fuchsia.h: New file.
40809 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40810 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40812 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40814 2016-01-10 Richard Biener <rguenther@suse.de>
40816 PR tree-optimization/79034
40817 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40818 Propagate out degenerate PHIs in the joiner.
40820 2017-01-10 Martin Liska <mliska@suse.cz>
40822 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40823 (sort_congruence_classes_by_decl_uid): Likewise.
40824 (sort_congruence_class_groups_by_decl_uid): Likewise.
40825 (sem_item_optimizer::merge_classes): Sort class, groups in these
40826 classes and members in the groups by DECL_UID of declarations.
40827 This would make merge operations stable.
40829 2017-01-10 Martin Liska <mliska@suse.cz>
40831 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40832 usage of m_classes_vec.
40833 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40834 (sem_item_optimizer::get_group_by_hash): Likewise.
40835 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40836 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40837 (sem_item_optimizer::verify_classes): Likewise.
40838 (sem_item_optimizer::process_cong_reduction): Likewise.
40839 (sem_item_optimizer::dump_cong_classes): Likewise.
40840 (sem_item_optimizer::merge_classes): Likewise.
40841 * ipa-icf.h (congruence_class_hash): Rename from
40842 congruence_class_group_hash. Remove declaration of m_classes_vec.
40844 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40846 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40847 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40848 * config.gcc: Add avx512vpopcntdqintrin.h.
40849 * config/i386/avx512vpopcntdqintrin.h: New.
40850 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40851 * config/i386/i386-builtin-types.def: Add new types.
40852 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40853 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40854 __builtin_ia32_vpopcountq_v8di_mask): New.
40855 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40856 __AVX512VPOPCNTDQ__.
40857 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40858 (PTA_AVX512VPOPCNTDQ): Define.
40859 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40860 TARGET_AVX512VPOPCNTDQ_P): Define.
40861 * config/i386/i386.opt: Add mavx512vpopcntdq.
40862 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40863 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40865 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40867 PR middle-end/77484
40868 * predict.def (PRED_CALL): Set to 67.
40870 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40872 * expr.c (store_field): In the bitfield case, if the value comes from
40873 a function call and is of an aggregate type returned in registers, do
40874 not modify the field mode; extract the value in all cases if the mode
40875 is BLKmode and the size is not larger than a word.
40877 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40880 * config/i386/cpuid.h: Fix undefined behavior.
40882 2017-01-04 Jeff Law <law@redhat.com>
40884 PR tree-optimization/79007
40885 PR tree-optimization/67955
40886 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40887 conservative for pt.null when flag_non_call_exceptions is on.
40889 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40891 PR translation/79019
40892 PR translation/79020
40893 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40894 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40895 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40897 * config/avr/avr.opt (maccumulate-args): Likewise.
40898 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40899 * common.opt (freport-bug): Likewise.
40900 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40901 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40902 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40903 translatable string.
40904 * config/i386/i386.c (function_value_32): Likewise.
40905 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40906 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40908 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40909 * common/config/msp430/msp430-common.c (msp430_handle_option):
40911 * symtab.c (symtab_node::verify_base): Likewise.
40912 * opts.c (set_debug_level): Likewise.
40913 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40914 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40915 missing whitespace to translatable strings.
40916 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40917 * config/sh/superh.h: Likewise.
40918 * config/i386/xopintrin.h: Likewise.
40919 * config/i386/znver1.md: Likewise.
40920 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40921 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40922 * double-int.h (struct double_int): Likewise.
40923 * double-int.c (div_and_round_double): Likewise.
40924 * wide-int.cc: Likewise.
40925 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40926 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40927 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40928 (crossjumps_occurred): ... this.
40929 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40932 PR tree-optimization/78899
40933 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40934 returning bool return struct loop *, NULL for failure and the new
40936 (versionable_outer_loop_p): Don't version outer loop if it has
40937 dont_vectorized bit set.
40938 (tree_if_conversion): When versioning outer loop, ensure
40939 tree_if_conversion is performed also on the inner loop of the
40940 non-vectorizable outer loop copy.
40941 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40942 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40943 prevent vectorization of it.
40944 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40945 the outer loop vectorization of the non-scalar version is attempted
40946 before vectorization of the inner loop in scalar version. If
40947 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40948 vectorization of its inner loop.
40949 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40950 has 2 inner loops, rename also on edges from bb whose single pred
40951 is outer_loop->header. Fix typo in function comment.
40953 2017-01-09 Martin Sebor <msebor@redhat.com>
40956 * asan.c (asan_emit_stack_protection): Increase local buffer size
40957 to avoid snprintf truncation warning.
40959 2017-01-09 Andrew Pinski <apinski@cavium.com>
40961 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40962 to reference thunderx2t99 for the tuning structure
40963 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40965 (thunderx2t99_extra_costs): This.
40966 * config/aarch64/aarch64-tune.md: Regenerate.
40967 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40968 (vulcan_addrcost_table): This.
40969 (vulcan_regmove_cost): Rename to ...
40970 (thunderx2t99_regmove_cost): This.
40971 (vulcan_vector_cost): Rename to ...
40972 (thunderx2t99_vector_cost): this.
40973 (vulcan_branch_cost): Rename to ...
40974 (thunderx2t99_branch_cost): This.
40975 (vulcan_tunings): Rename to ...
40976 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40977 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40979 2017-01-09 Martin Jambor <mjambor@suse.cz>
40983 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40984 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40985 (propagate_vr_accross_jump_function): Use the above function for all
40986 value range computations for pass-through jump functions and type
40987 converasion from explicit value range values.
40988 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40989 parameters from TYPE_ARG_TYPES.
40990 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40991 (ipa_write_node_info): Stream type of the actual argument.
40992 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40994 2017-01-09 Martin Liska <mliska@suse.cz>
40997 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40998 (lookup_compiler): Do not show error message with have_E.
41000 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41002 PR tree-optimization/78938
41003 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41004 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41005 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41006 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41009 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41011 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41014 2017-01-09 Richard Biener <rguenther@suse.de>
41016 PR tree-optimization/78997
41017 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41018 name condition properly.
41020 2017-01-09 Richard Biener <rguenther@suse.de>
41023 * dwarf2out.c (is_cxx): New overload with context.
41024 (is_naming_typedef_decl): Use it.
41026 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41028 * invoke.texi (Option Summary): Correct spacing in option lists
41029 and add line breaks to fix over-long lines.
41031 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41033 PR middle-end/17660
41035 * extend.texi (Common Variable Attributes): Add xref to GCC
41036 Internals manual to explain mode attribute keywords.
41038 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41041 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41042 and Preprocessor Options.
41043 (Options for Linking): Document -pthread here....
41044 (RS/6000 and PowerPC Options): ...not here.
41045 (Solaris 2 Options): ...or here.
41046 * doc/cppopts.texi: Document -pthread.
41048 2017-01-08 Martin Sebor <msebor@redhat.com>
41050 PR middle-end/77708
41051 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41052 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41053 New member functions.
41054 (format_directive): Used them.
41056 (pass_sprintf_length::handle_gimple_call): Same.
41057 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41058 to avoid truncation for any argument.
41059 (extract_affine_mul): Same.
41060 * tree.c (get_file_function_name): Same.
41062 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41064 PR middle-end/77484
41065 * predict.def (PRED_INDIR_CALL): Set to 86.
41067 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41069 PR preprocessor/54124
41070 * doc/cppopts.texi: Reformat -d subtable to list the full name
41071 of the options. Add cross-reference to the docs for the general
41072 compiler -d options.
41073 * doc/invoke.texi (Developer Options): Add cross-reference to the
41074 preprocessor-specific -d option documentation.
41076 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41078 PR preprocessor/13498
41079 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41080 redudant material, and reflect new command-line options.
41081 (System Headers): Likewise.
41083 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41085 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41086 -isystem, and -idirafter. Copy-edit.
41087 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41088 default for -ftrack-macro-expansion. Delete obsolete and
41089 badly-formatted implementation details about -fdebug-cpp output.
41090 * doc/cppwarnopts.texi: Copy-edit.
41092 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41095 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41096 that the transition from a max line width >= 1<<10 to narrower
41097 lines works correctly.
41099 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41101 * doc/options.texi (PerFunction): New.
41102 * opt-functions.awk (switch_flags): Map both Optimization and
41103 PerFunction to CL_OPTIMIZATION.
41104 * opth-gen.awk: Test for PerFunction flag along with
41106 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41107 it only when the latter is present. Skip those that don't in
41108 the hash function generator.
41109 * common.opt (fvar-tracking): Mark as PerFunction instead of
41111 (fvar-tracking-assignments): Likewise.
41112 (fvar-tracking-assignments-toggle): Likewise.
41113 (fvar-tracking-uninit): Likewise.
41115 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41117 PR translation/79018
41118 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41121 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41124 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41125 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41126 TARGET_LONG_JUMP_TABLE_OFFSETS.
41127 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41128 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41129 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41130 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41131 * config/m68k/m68k.md (tablejump expander): Likewise.
41132 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41133 TARGET_LONG_JUMP_TABLE_OFFSETS.
41134 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41135 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41137 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41138 David Holsgrove <david.holsgrove@xilinx.com>
41140 * common/config/microblaze/microblaze-common.c
41141 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41142 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41144 * config/microblaze/microblaze.c (microblaze_must_save_register)
41145 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41147 (microblaze_eh_return): New function.
41148 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41149 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41150 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41151 * config/microblaze/microblaze.md (eh_return): New pattern.
41153 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41155 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41156 GCC_DIAGNOSTIC_STRINGIFY): Define.
41158 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41160 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41162 * config/arm/arm.md (<mcrr>): New.
41164 * config/arm/arm.c (arm_arch5te): New.
41165 (arm_option_override): Set arm_arch5te.
41166 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41168 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41169 (arm_mcrr_qualifiers): ... this. New.
41170 (MRRC_QUALIFIERS): Define to...
41171 (arm_mrrc_qualifiers): ... this. New.
41172 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41174 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41175 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41176 (MRRCI, mrrc, MRRC): New.
41177 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41178 VUNSPEC_MRRC2): New.
41180 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41182 * config/arm/arm.md (<mcr>): New.
41184 * config/arm/arm.c (arm_coproc_builtin_available): Add
41185 support for mcr, mrc, mcr2 and mrc2.
41186 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41187 (arm_mcr_qualifiers): ... this. New.
41188 (MRC_QUALIFIERS): Define to ...
41189 (arm_mrc_qualifiers): ... this. New.
41190 (MCR_QUALIFIERS): Define to ...
41191 (arm_mcr_qualifiers): ... this. New.
41192 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41194 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41195 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41196 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41197 VUNSPEC_MRC2): New.
41199 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41201 * config/arm/arm.md (*ldc): New.
41205 * config/arm/arm.c (arm_coproc_builtin_available): Add
41206 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41207 (arm_coproc_ldc_stc_legitimate_address): New.
41208 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41209 'qualifier_const_pointer'.
41210 (LDC_QUALIFIERS): Define to...
41211 (arm_ldc_qualifiers): ... this. New.
41212 (STC_QUALIFIERS): Define to...
41213 (arm_stc_qualifiers): ... this. New.
41214 * config/arm/arm-protos.h
41215 (arm_coproc_ldc_stc_legitimate_address): New.
41216 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41217 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41218 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41219 stc2, stcl, stc2l): New.
41220 * config/arm/constraints.md (Uz): New.
41221 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41222 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41223 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41224 VUNSPEC_STC2L): New.
41226 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41228 * config/arm/arm.md (<cdp>): New.
41229 * config/arm/arm.c (neon_const_bounds): Rename this ...
41230 (arm_const_bounds): ... this.
41231 (arm_coproc_builtin_available): New.
41232 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41233 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41234 (CDP_QUALIFIERS): Define to...
41235 (arm_cdp_qualifiers): ... this. New.
41237 (arm_expand_builtin_args): Add case for 6 arguments.
41238 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41239 (arm_const_bounds): ... this.
41240 (arm_coproc_builtin_available): New.
41241 * config/arm/arm_acle.h (__arm_cdp): New.
41243 * config/arm/arm_acle_builtins.def (cdp): New.
41245 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41246 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41247 'arm_const_bounds'.
41248 * config/arm/types.md (coproc): New.
41249 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41250 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41251 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41252 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41254 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41256 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41257 (UBINOP_QUALIFIERS): New.
41259 (acle_builtin_data): New. Change comment.
41260 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41261 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41262 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41263 arm_acle_builtins.def.
41264 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41265 (arm_init_acle_builtins): New.
41266 (CRC32_BUILTIN): Remove.
41267 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41268 crc32cb, crc32ch and crc32cw.
41269 (arm_init_crc32_builtins): Remove.
41270 (arm_init_builtins): Use arm_init_acle_builtins rather
41271 than arm_init_crc32_builtins.
41272 (arm_expand_acle_builtin): New.
41273 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41274 * config/arm/arm_acle_builtins.def: New.
41276 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41278 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41279 (arm_builtin_datum): ... this.
41280 (arm_init_neon_builtin): Rename to ...
41281 (arm_init_builtin): ... this. Add a new parameters PREFIX
41282 and USE_SIG_IN_NAME.
41283 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41284 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41285 'arm_builtin_datum'.
41286 (arm_init_vfp_builtins): Likewise.
41287 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41288 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41289 (arm_expand_neon_args): Rename to ...
41290 (arm_expand_builtin_args): ... this. Rename builtin_arg
41291 enum values and differentiate between ARG_BUILTIN_MEMORY
41292 and ARG_BUILTIN_NEON_MEMORY.
41293 (arm_expand_neon_builtin_1): Rename to ...
41294 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41295 values, arm_expand_builtin_args and add bool parameter NEON.
41296 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41297 (arm_expand_vfp_builtin): Likewise.
41298 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41300 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41302 PR middle-end/77484
41303 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41304 * predict.c (tree_estimate_probability_bb): Reverse direction of
41305 polymorphic call predictor.
41307 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41309 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41310 (determine_pass_name_match): ...this new function and...
41311 (should_skip_pass_p): ...this new function.
41313 2017-01-06 Nathan Sidwell <nathan@acm.org>
41315 * ipa-visibility.c (function_and_variable_visibility): Reformat
41316 comments and long lines. Remove extrneous if.
41317 * symtab.c (symtab_node::make_decl_local): Fix code format.
41318 (symtab_node::set_section_for_node): Fix comment typo.
41320 2017-01-06 Martin Liska <mliska@suse.cz>
41323 * lra-constraints.c: Rename invariant to lra_invariant.
41324 * predict.c (set_even_probabilities): Initialize e to NULL.
41326 2017-01-05 Martin Sebor <msebor@redhat.com>
41328 PR tree-optimization/78910
41329 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41330 (format_integer): Correct off-by-one error in the handling
41331 of precision with negative numbers in signed conversions..
41333 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41335 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41337 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41339 PR tree-optimization/71016
41340 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41341 factor_out_conditional_conversion. Formatting fix.
41342 (factor_out_conditional_conversion): Add cond_stmt argument.
41343 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41344 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41347 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41349 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41350 read-rtl-function.o, and selftest-rtl.o.
41351 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41352 (selftest::aarch64_test_loading_full_dump): New function.
41353 (selftest::aarch64_run_selftests): New function.
41354 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41355 selftest::aarch64_run_selftests.
41356 * config/i386/i386.c
41357 (selftest::ix86_test_loading_dump_fragment_1): New function.
41358 (selftest::ix86_test_loading_call_insn): New function.
41359 (selftest::ix86_test_loading_full_dump): New function.
41360 (selftest::ix86_test_loading_unspec): New function.
41361 (selftest::ix86_run_selftests): Call the new functions.
41362 * emit-rtl.c (maybe_set_max_label_num): New function.
41363 * emit-rtl.h (maybe_set_max_label_num): New decl.
41364 * function.c (instantiate_decls): Guard call to
41365 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41366 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41368 * gensupport.c (gen_reader::gen_reader): Pass "false"
41369 for new "compact" param of rtx_reader.
41370 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41371 rather than an empty string for NULL strings.
41372 * read-md.c: Potentially include config.h rather than bconfig.h.
41373 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41374 (have_error): New global, copied from errors.c.
41375 (md_reader::read_name): Rename to...
41376 (md_reader::read_name_1): ...this, adding "out_loc" param,
41377 and converting "missing name or number" to returning false, rather
41379 (md_reader::read_name): Reimplement in terms of read_name_1.
41380 (md_reader::read_name_or_nil): New function.
41381 (md_reader::read_string): Handle "(nil)" by returning NULL.
41382 (md_reader::md_reader): Add new param "compact".
41383 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41384 (md_reader::read_file): New method.
41385 * read-md.h (md_reader::md_reader): Add new param "compact".
41386 (md_reader::read_file): New method.
41387 (md_reader::is_compact): New accessor.
41388 (md_reader::read_name): Convert return type from void to file_location.
41389 (md_reader::read_name_or_nil): New decl.
41390 (md_reader::read_name_1): New decl.
41391 (md_reader::m_compact): New field.
41392 (noop_reader::noop_reader): Pass "false" for new "compact" param
41394 (rtx_reader::rtx_reader): Add new "compact" param.
41395 (rtx_reader::read_rtx_operand): Make virtual and convert return
41396 type from void to rtx.
41397 (rtx_reader::read_until): New decl.
41398 (rtx_reader::handle_any_trailing_information): New virtual function.
41399 (rtx_reader::postprocess): New virtual function.
41400 (rtx_reader::finalize_string): New virtual function.
41401 (rtx_reader::m_in_call_function_usage): New field.
41402 (rtx_reader::m_reuse_rtx_by_id): New field.
41403 * read-rtl-function.c: New file.
41404 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41405 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41406 (selftest::verify_three_block_rtl_cfg): New decl.
41407 * read-rtl-function.h: New file.
41408 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41409 For host, include function.h, memmodel.h, and emit-rtl.h.
41410 (one_time_initialization): New function.
41411 (struct compact_insn_name): New struct.
41412 (compact_insn_names): New array.
41413 (find_code): Handle insn codes in compact dumps.
41414 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41415 (bind_subst_iter_and_attr): Likewise.
41416 (add_condition_to_string): Likewise.
41417 (add_condition_to_rtx): Likewise.
41418 (apply_attribute_uses): Likewise.
41419 (add_current_iterators): Likewise.
41420 (apply_iterators): Likewise.
41421 (initialize_iterators): Guard usage of apply_subst_iterator with
41422 #ifdef GENERATOR_FILE.
41423 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41424 (md_reader::read_mapping): Likewise.
41425 (add_define_attr_for_define_subst): Likewise.
41426 (add_define_subst_attr): Likewise.
41427 (read_subst_mapping): Likewise.
41428 (check_code_iterator): Likewise.
41429 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41431 (one_time_initialization): New function.
41432 (rtx_reader::read_until): New method.
41433 (read_flags): New function.
41434 (parse_reg_note_name): New function.
41435 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41436 Handle reuse_rtx ids.
41437 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41438 Add parsing support for RTL dumps, mirroring the special-cases in
41439 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41440 values, and calling handle_any_trailing_information.
41441 (rtx_reader::read_rtx_operand): Convert return type from void
41442 to rtx, returning return_rtx. Handle case 'e'. Call
41443 finalize_string on XSTR and XTMPL fields.
41444 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41445 "(nil)" values were omitted. Call the postprocess vfunc on the
41447 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41448 class ctor. Initialize m_in_call_function_usage. Call
41449 one_time_initialization.
41450 * rtl-tests.c (selftest::test_uncond_jump): Call
41451 set_new_first_and_last_insn.
41452 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41453 * selftest-rtl.c: New file.
41454 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41455 (selftest::get_insn_by_uid): New decl.
41456 * selftest-run-tests.c (selftest::run_tests): Call
41457 read_rtl_function_c_tests.
41458 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41459 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41462 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41464 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41465 operands in a special way. Assert that pos+len <= mode precision.
41467 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41469 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41470 3 argument Alias with unlimited for the negative form.
41471 (fno-vect-cost-model): Removed.
41473 2017-01-05 Martin Liska <mliska@suse.cz>
41475 * hsa-gen.c (gen_hsa_divmod): New function.
41476 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41478 2017-01-05 Martin Liska <mliska@suse.cz>
41481 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41484 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41486 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41487 small constant length operands.
41489 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41491 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41492 between loop iterations.
41494 2017-01-05 Martin Liska <mliska@suse.cz>
41497 * gimplify.c (gimplify_decl_expr): Compare to
41498 asan_poisoned_variables instread of checking flags.
41499 (gimplify_target_expr): Likewise.
41500 (gimplify_expr): Likewise.
41501 (gimplify_function_tree): Conditionally initialize
41502 asan_poisoned_variables.
41504 2017-01-04 Jeff Law <law@redhat.com>
41506 PR tree-optimizatin/78812
41507 * rtl.h (contains_mem_rtx_p): Prototype.
41508 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41509 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41510 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41511 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41512 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41514 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41516 * input.c (assert_char_at_range): Default-initialize actual_range.
41518 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41520 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41521 to match the caller.
41523 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41525 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41526 insns after final jump in test to emit dummy move.
41528 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41530 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41531 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41533 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41535 * multiple_target.c (create_dispatcher_calls): Init e_next.
41536 * tree-ssa-loop-split.c (split_loop): Init border.
41537 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41540 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41545 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41546 (altivec_register_operand): Do not return true if the operand
41547 contains a SUBREG mixing SImode and SFmode.
41548 (vsx_register_operand): Likewise.
41549 (vsx_reg_sfsubreg_ok): New predicate.
41550 (vfloat_operand): Do not return true if the operand contains a
41551 SUBREG mixing SImode and SFmode.
41552 (vint_operand): Likewise.
41553 (vlogical_operand): Likewise.
41554 (gpc_reg_operand): Likewise.
41555 (int_reg_operand): Likewise.
41556 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41557 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41558 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41560 (rs6000_emit_move_si_sf_subreg): New helper function.
41561 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41562 fixup SUBREGs involving SImode and SFmode.
41563 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41564 numbers for the new peephole2 optimization.
41565 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41566 the GLIBC math library that do AND/IOR/XOR operations on single
41567 precision floating point.
41568 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41569 target macros to say whether we need to avoid SUBREGs mixing
41571 (TARGET_ALLOW_SF_SUBREG): Likewise.
41572 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41573 (UNSPEC_SI_FROM_SF): Likewise.
41574 (iorxor): Change spacing.
41575 (and_ior_xor): New iterator for AND, IOR, and XOR.
41576 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41577 (movdi_from_sf_zero_ext): Likewise.
41578 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41579 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41580 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41581 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41582 (fms<mode>4): Likewise.
41583 (fnma<mode>4): Likewise.
41584 (fnms<mode>4): Likewise.
41585 (nfma<mode>4): Likewise.
41586 (nfms<mode>4): Likewise.
41588 2017-01-04 Marek Polacek <polacek@redhat.com>
41591 * doc/invoke.texi: Document -Wpointer-compare.
41593 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41595 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41598 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41599 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41602 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41605 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41606 documentation of the powerpc_popcntb_ok attribute.
41607 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41608 code to issue warning messages if a requested CPU configuration is
41609 not supported by the binary (assembler and loader) toolchain.
41610 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41611 made to define a built-in function that has been disabled.
41612 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41613 made to define a built-in function that has been disabled.
41614 (altivec_init_builtins): Add comment explaining why definition
41615 of the DST built-in functions is not preceded by an assertion
41616 check. Add assertions to prevent ICE if attempts are made to
41617 define an altivec predicate or an abs* built-in function that has
41619 (htm_init_builtins): Add comment explaining why definition of the
41620 htm built-in functions is not preceded by an assertion check.
41622 2017-01-04 Jeff Law <law@redhat.com>
41624 PR tree-optimizatin/67955
41625 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41626 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41627 the points-to solution does not include pt_null. Use DECL_PT_UID
41630 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41632 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41633 Use gen_int_mode instead of gen_lopwart for const_int operands.
41635 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41637 PR tree-optimization/71563
41638 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41639 out of range value - has low bits known to be zero.
41641 2017-01-04 Alan Modra <amodra@gmail.com>
41643 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41644 * configure: Regenerate.
41645 * config.in: Regenerate.
41647 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41650 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41651 a substring of the message, but strcmp with the whole message. Ifdef
41652 ENABLE_NLS, translate the message first using dgettext.
41654 2017-01-03 Jeff Law <law@redhat.com>
41656 PR tree-optimizatin/78856
41657 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41658 (mark_threaded_blocks): Remove code to truncate thread paths that
41659 cross multiple loop headers. Instead invalidate the cached loop
41660 iteration information and handle case of a thread path walking
41661 into an irreducible region.
41663 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41666 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41667 assertions. Add support for doing the signbit if the IEEE 128-bit
41668 floating point value is in a GPR.
41669 * config/rs6000/rs6000.md (Fsignbit): Delete.
41670 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41671 Update the length attribute if the value is in a GPR.
41672 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41673 the sign or zero extension instruction, since the value is always 0/1.
41674 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41677 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41678 extracting SImode to a GPR register so that we can generate a
41679 store, limit the vector to be in a traditional Altivec register
41680 for the vextuwrx instruction.
41682 2017-01-03 Ian Lance Taylor <iant@google.com>
41684 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41686 2017-01-03 Martin Sebor <msebor@redhat.com>
41688 PR tree-optimization/78696
41689 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41690 precision. Use MPFR for %f for greater fidelity. Correct handling
41692 (pass_sprintf_length::compute_format_length): Set width and precision
41693 specified by asrerisk to void_node for vararg functions.
41694 (try_substitute_return_value): Adjust dump output.
41696 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41698 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41700 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41702 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41703 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41704 -mlra/-mno-lra was passed to the compiler.
41706 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41708 PR rtl-optimization/65618
41709 * emit-rtl.c (try_split): Move initialization of "before" and
41710 "after" to just before the call to emit_insn_after_setloc.
41712 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41714 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41716 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41718 * dwarf2out.c (gen_enumeration_type_die): When
41719 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41721 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41723 PR tree-optimization/78965
41724 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41725 Change first argument from const call_info & to call_info &. For %n
41726 set info.nowrite to false.
41728 PR middle-end/78901
41729 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41730 possibly throwing calls.
41732 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41733 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41734 and fns handling, rather than in a separate case SSA_NAME.
41736 2017-01-02 Jeff Law <law@redhat.com>
41738 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41739 fixes for first_period and second_period variables.
41741 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41744 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41745 (*insvqi_1): New insn pattern.
41746 (*insvqi_1_mem_rex64): Ditto.
41747 (*insvqi_2): Ditto.
41748 (*insvqi_3): Rename from *insvqi.
41750 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41752 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41754 * doc/cfg.texi (Edges): Remove reference to Java.
41755 (Maintaining the CFG): Ditto.
41757 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41759 PR middle-end/77674
41760 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41761 transparent aliases.
41763 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41765 PR middle-end/77484
41766 * predict.def (PRED_CALL): Update hitrate.
41767 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41768 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41769 into direct/indirect/polymorphic variants.
41771 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41773 Update copyright years.
41775 * gcc.c (process_command): Update copyright notice dates.
41776 * gcov-dump.c (print_version): Ditto.
41777 * gcov.c (print_version): Ditto.
41778 * gcov-tool.c (print_version): Ditto.
41779 * gengtype.c (create_file): Ditto.
41780 * doc/cpp.texi: Bump @copying's copyright year.
41781 * doc/cppinternals.texi: Ditto.
41782 * doc/gcc.texi: Ditto.
41783 * doc/gccint.texi: Ditto.
41784 * doc/gcov.texi: Ditto.
41785 * doc/install.texi: Ditto.
41786 * doc/invoke.texi: Ditto.
41788 Copyright (C) 2017 Free Software Foundation, Inc.
41790 Copying and distribution of this file, with or without modification,
41791 are permitted in any medium without royalty provided the copyright
41792 notice and this notice are preserved.